Keep real radar leads under ACC authority

.gitmodules

@@ -4,6 +4,7 @@
 [submodule "opendbc"]
   path = opendbc_repo
   url = https://github.com/sunnypilot/opendbc.git
+  branch = tn
 [submodule "msgq"]
   path = msgq_repo
   url = https://github.com/commaai/msgq.git

cereal/custom.capnp

@@ -194,6 +194,7 @@ struct LongitudinalPlanSP @0xf35cc4560bbf6ec2 {
   aTarget @5 :Float32;
   events @6 :List(OnroadEventSP.Event);
   e2eAlerts @7 :E2eAlerts;
+  accelController @8 :AccelController;
 
   struct DynamicExperimentalControl {
     state @0 :DynamicExperimentalControlState;
@@ -296,6 +297,47 @@ struct LongitudinalPlanSP @0xf35cc4560bbf6ec2 {
     greenLightAlert @0 :Bool;
     leadDepartAlert @1 :Bool;
   }
+
+  struct AccelController {
+    enabled @0 :Bool;
+    active @1 :Bool;
+    shadowOnly @2 :Bool;
+    profile @3 :Profile;
+    state @4 :State;
+    vTargetBase @5 :Float32;
+    vTargetRaw @6 :Float32;
+    vTargetFiltered @7 :Float32;
+    vTargetShadow @8 :Float32;
+    leadIndex @9 :Int8 = -1;
+    usableGap @10 :Float32;
+    closingSpeed @11 :Float32;
+    requiredDecel @12 :Float32;
+    aMaxProfile @13 :Float32;
+    aMaxEffective @14 :Float32;
+
+    enum Profile {
+      eco @0;
+      normal @1;
+      sport @2;
+    }
+
+    enum State {
+      inactive @0;
+      free @1;
+      restrict @2;
+      hold @3;
+      release @4;
+      stopHold @5;
+    }
+  }
+
+  # Compatibility type for vehicle integrations that map physical drive modes
+  # onto AccelPersonality. New controller telemetry uses AccelController.Profile.
+  enum AccelerationPersonality {
+    eco @0;
+    normal @1;
+    sport @2;
+  }
 }
 
 struct OnroadEventSP @0xda96579883444c35 {
@@ -342,6 +384,7 @@ struct OnroadEventSP @0xda96579883444c35 {
     speedLimitChanged @21;
     speedLimitPending @22;
     e2eChime @23;
+    laneChangeRoadEdge @24;
   }
 }
 
@@ -448,6 +491,8 @@ struct LiveMapDataSP @0xf416ec09499d9d19 {
 
 struct ModelDataV2SP @0xa1680744031fdb2d {
   laneTurnDirection @0 :TurnDirection;
+  leftLaneChangeEdgeBlock @1 :Bool;
+  rightLaneChangeEdgeBlock @2 :Bool;
 
   enum TurnDirection {
     none @0;

common/params_keys.h

@@ -179,12 +179,19 @@ inline static std::unordered_map<std::string, ParamKeyAttributes> keys = {
     {"QuickBootToggle", {PERSISTENT | BACKUP, BOOL, "0"}},
     {"QuietMode", {PERSISTENT | BACKUP, BOOL, "0"}},
     {"RainbowMode", {PERSISTENT | BACKUP, BOOL, "0"}},
+    {"RoadEdgeLaneChangeEnabled", {PERSISTENT | BACKUP, BOOL, "0"}},
     {"RocketFuel", {PERSISTENT | BACKUP, BOOL, "0"}},
     {"ShowAdvancedControls", {PERSISTENT | BACKUP, BOOL, "0"}},
     {"ShowTurnSignals", {PERSISTENT | BACKUP, BOOL, "0"}},
     {"StandstillTimer", {PERSISTENT | BACKUP, BOOL, "0"}},
     {"TrueVEgoUI", {PERSISTENT | BACKUP, BOOL, "0"}},
 
+    // toyota specific params
+    {"ToyotaAutoHold", {PERSISTENT | BACKUP, BOOL, "0"}},
+    {"ToyotaEnhancedBsm", {PERSISTENT | BACKUP, BOOL, "0"}},
+    {"ToyotaTSS2Long", {PERSISTENT | BACKUP, BOOL, "0"}},
+    {"ToyotaDriveMode", {PERSISTENT | BACKUP, BOOL, "0"}},
+
     // MADS params
     {"Mads", {PERSISTENT | BACKUP, BOOL, "1"}},
     {"MadsMainCruiseAllowed", {PERSISTENT | BACKUP, BOOL, "1"}},
@@ -228,6 +235,10 @@ inline static std::unordered_map<std::string, ParamKeyAttributes> keys = {
     {"DynamicExperimentalControl", {PERSISTENT | BACKUP, BOOL, "0"}},
     {"BlindSpot", {PERSISTENT | BACKUP, BOOL, "0"}},
 
+    // Accel Controller profiles (Eco / Normal / Sport)
+    {"AccelPersonalityEnabled", {PERSISTENT | BACKUP, BOOL, "0"}},
+    {"AccelPersonality", {PERSISTENT | BACKUP, INT, "1"}},
+
     // sunnypilot model params
     {"CameraOffset", {PERSISTENT | BACKUP, FLOAT, "0.0"}},
     {"LagdToggle", {PERSISTENT | BACKUP, BOOL, "1"}},

common/tests/test_params.py

@@ -112,12 +112,16 @@ class TestParams:
   def test_params_default_value(self):
     self.params.remove("LanguageSetting")
     self.params.remove("LongitudinalPersonality")
+    self.params.remove("AccelPersonalityEnabled")
+    self.params.remove("AccelPersonality")
     self.params.remove("LiveParameters")
 
     assert self.params.get("LanguageSetting") is None
     assert self.params.get("LanguageSetting", return_default=False) is None
     assert isinstance(self.params.get("LanguageSetting", return_default=True), str)
     assert isinstance(self.params.get("LongitudinalPersonality", return_default=True), int)
+    assert self.params.get("AccelPersonalityEnabled", return_default=True) is False
+    assert self.params.get("AccelPersonality", return_default=True) == 1
     assert self.params.get("LiveParameters") is None
     assert self.params.get("LiveParameters", return_default=True) is None
 

selfdrive/car/card.py

@@ -10,7 +10,7 @@ from cereal import car, log, custom
 from openpilot.common.params import Params
 from openpilot.common.realtime import config_realtime_process, Priority, Ratekeeper
 from openpilot.common.swaglog import cloudlog, ForwardingHandler
-
+from opendbc.safety import ALTERNATIVE_EXPERIENCE
 from opendbc.car import DT_CTRL, structs
 from opendbc.car.can_definitions import CanData, CanRecvCallable, CanSendCallable
 from opendbc.car.carlog import carlog
@@ -121,7 +121,13 @@ class Car:
       self.CI, self.CP, self.CP_SP = CI, CI.CP, CI.CP_SP
       self.RI = RI
 
+    # set alternative experiences from parameters
+    sp_toyota_auto_brake_hold = self.params.get_bool("ToyotaAutoHold")
     self.CP.alternativeExperience = 0
+    if sp_toyota_auto_brake_hold:
+      self.CP.alternativeExperience |= ALTERNATIVE_EXPERIENCE.ALLOW_AEB
+
+
     # mads
     set_alternative_experience(self.CP, self.CP_SP, self.params)
     set_car_specific_params(self.CP, self.CP_SP, self.params)

selfdrive/controls/lib/desire_helper.py

@@ -56,7 +56,7 @@ class DesireHelper:
   def get_lane_change_direction(CS):
     return LaneChangeDirection.left if CS.leftBlinker else LaneChangeDirection.right
 
-  def update(self, carstate, lateral_active, lane_change_prob):
+  def update(self, carstate, lateral_active, lane_change_prob, left_edge_detected=False, right_edge_detected=False):
     self.alc.update_params()
     self.lane_turn_controller.update_params()
     v_ego = carstate.vEgo
@@ -88,8 +88,8 @@ class DesireHelper:
                          ((carstate.steeringTorque > 0 and self.lane_change_direction == LaneChangeDirection.left) or
                           (carstate.steeringTorque < 0 and self.lane_change_direction == LaneChangeDirection.right))
 
-        blindspot_detected = ((carstate.leftBlindspot and self.lane_change_direction == LaneChangeDirection.left) or
-                              (carstate.rightBlindspot and self.lane_change_direction == LaneChangeDirection.right))
+        blindspot_detected = (((carstate.leftBlindspot or left_edge_detected) and self.lane_change_direction == LaneChangeDirection.left) or
+                              ((carstate.rightBlindspot or right_edge_detected) and self.lane_change_direction == LaneChangeDirection.right))
 
         self.alc.update_lane_change(blindspot_detected, carstate.brakePressed)
 

selfdrive/controls/lib/longitudinal_mpc_lib/long_mpc.py

@@ -217,6 +217,7 @@ class LongitudinalMpc:
   def __init__(self, dt=DT_MDL):
     self.dt = dt
     self.solver = AcadosOcpSolverCython(MODEL_NAME, ACADOS_SOLVER_TYPE, N)
+    self.last_solution_status = 0
     self.reset()
     self.source = LongitudinalPlanSource.cruise
 
@@ -313,7 +314,8 @@ class LongitudinalMpc:
     lead_xv = self.extrapolate_lead(x_lead, v_lead, a_lead, a_lead_tau)
     return lead_xv
 
-  def update(self, radarstate, v_cruise, personality=log.LongitudinalPersonality.standard):
+  def update(self, radarstate, v_cruise, personality=log.LongitudinalPersonality.standard,
+             accel_max: float | tuple[float, ...] | np.ndarray | None = None):
     t_follow = get_T_FOLLOW(personality)
     v_ego = self.x0[1]
     self.status = radarstate.leadOne.status or radarstate.leadTwo.status
@@ -345,6 +347,17 @@ class LongitudinalMpc:
 
     self.params[:,0] = ACCEL_MIN
     self.params[:,1] = ACCEL_MAX
+    if accel_max is not None:
+      try:
+        accel_max_trajectory = np.asarray(accel_max, dtype=float)
+      except (TypeError, ValueError):
+        accel_max_trajectory = np.empty(0)
+      if accel_max_trajectory.ndim == 0 and np.isfinite(accel_max_trajectory) and accel_max_trajectory >= 0.0:
+        accel_max_trajectory = np.full(N + 1, float(accel_max_trajectory))
+      valid_accel_max = accel_max_trajectory.shape == (N + 1,) and np.all(np.isfinite(accel_max_trajectory))
+      if valid_accel_max:
+        self.params[:,1] = np.clip(accel_max_trajectory, ACCEL_MIN, ACCEL_MAX)
+        self.params[0,1] = max(self.params[0,1], float(np.clip(self.x0[2], ACCEL_MIN, ACCEL_MAX)))
     self.params[:,2] = np.min(x_obstacles, axis=1)
     self.params[:,3] = np.copy(self.a_prev)
     self.params[:,4] = t_follow
@@ -364,6 +377,7 @@ class LongitudinalMpc:
     self.solver.constraints_set(0, "ubx", self.x0)
 
     self.solution_status = self.solver.solve()
+    self.last_solution_status = self.solution_status
     self.solve_time = float(self.solver.get_stats('time_tot')[0])
     self.time_qp_solution = float(self.solver.get_stats('time_qp')[0])
     self.time_linearization = float(self.solver.get_stats('time_lin')[0])

selfdrive/controls/lib/longitudinal_planner.py

@@ -51,7 +51,7 @@ class LongitudinalPlanner(LongitudinalPlannerSP):
   def __init__(self, CP, CP_SP, init_v=0.0, init_a=0.0, dt=DT_MDL):
     self.CP = CP
     self.mpc = LongitudinalMpc(dt=dt)
-    LongitudinalPlannerSP.__init__(self, self.CP, CP_SP, self.mpc)
+    LongitudinalPlannerSP.__init__(self, self.CP, CP_SP, self.mpc, dt=dt)
     self.fcw = False
     self.dt = dt
     self.allow_throttle = True
@@ -113,6 +113,7 @@ class LongitudinalPlanner(LongitudinalPlannerSP):
     accel_clip = [ACCEL_MIN, get_max_accel(v_ego)]
     steer_angle_without_offset = sm['carState'].steeringAngleDeg - sm['liveParameters'].angleOffsetDeg
     accel_clip = limit_accel_in_turns(v_ego, steer_angle_without_offset, accel_clip, self.CP)
+    profile_accel_clip = limit_accel_in_turns(v_ego, steer_angle_without_offset, [ACCEL_MIN, ACCEL_MAX], self.CP)
 
     if reset_state:
       self.v_desired_filter.x = v_ego
@@ -129,16 +130,21 @@ class LongitudinalPlanner(LongitudinalPlannerSP):
       clipped_accel_coast = max(accel_coast, accel_clip[0])
       clipped_accel_coast_interp = np.interp(v_ego, [MIN_ALLOW_THROTTLE_SPEED, MIN_ALLOW_THROTTLE_SPEED*2], [accel_clip[1], clipped_accel_coast])
       accel_clip[1] = min(accel_clip[1], clipped_accel_coast_interp)
+    controller_accel_max = profile_accel_clip[1] if self.allow_throttle else 0.0
 
     # Get new v_cruise and a_desired from Smart Cruise Control and Speed Limit Assist
+    previous_output_a_target = self.output_a_target
     v_cruise, self.a_desired = LongitudinalPlannerSP.update_targets(self, sm, self.v_desired_filter.x, self.a_desired, v_cruise)
+    base_v_cruise = v_cruise
 
     if force_slow_decel:
       v_cruise = 0.0
 
-    self.mpc.set_weights(prev_accel_constraint, personality=sm['selfdriveState'].personality)
-    self.mpc.set_cur_state(self.v_desired_filter.x, self.a_desired)
-    self.mpc.update(sm['radarState'], v_cruise, personality=sm['selfdriveState'].personality)
+    is_e2e = LongitudinalPlannerSP.update_accel_controller_mpc(
+      self, sm, base_v_cruise, v_cruise, prev_accel_constraint, reset_state=reset_state,
+      cruise_initialized=v_cruise_initialized, planner_accel=self.a_desired, previous_output_accel=previous_output_a_target,
+      available_accel_max=controller_accel_max, previous_should_stop=self.output_should_stop, force_decel=force_slow_decel,
+    )
 
     self.v_desired_trajectory = np.interp(CONTROL_N_T_IDX, T_IDXS_MPC, self.mpc.v_solution)
     self.a_desired_trajectory = np.interp(CONTROL_N_T_IDX, T_IDXS_MPC, self.mpc.a_solution)
@@ -154,13 +160,14 @@ class LongitudinalPlanner(LongitudinalPlannerSP):
     self.a_desired = float(np.interp(self.dt, CONTROL_N_T_IDX, self.a_desired_trajectory))
     self.v_desired_filter.x = self.v_desired_filter.x + self.dt * (self.a_desired + a_prev) / 2.0
 
-    action_t =  self.CP.longitudinalActuatorDelay + DT_MDL
-    output_a_target_mpc, output_should_stop_mpc = get_accel_from_plan(self.v_desired_trajectory, self.a_desired_trajectory, CONTROL_N_T_IDX,
-                                                                        action_t=action_t, vEgoStopping=self.CP.vEgoStopping)
+    action_t = self.CP.longitudinalActuatorDelay + DT_MDL
+    output_a_target_mpc, output_should_stop_mpc = get_accel_from_plan(
+      self.v_desired_trajectory, self.a_desired_trajectory, CONTROL_N_T_IDX, action_t=action_t, vEgoStopping=self.CP.vEgoStopping,
+    )
     output_a_target_e2e = sm['modelV2'].action.desiredAcceleration
     output_should_stop_e2e = sm['modelV2'].action.shouldStop
 
-    if self.is_e2e(sm):
+    if is_e2e:
       output_a_target = min(output_a_target_e2e, output_a_target_mpc)
       self.output_should_stop = output_should_stop_e2e or output_should_stop_mpc
       if output_a_target < output_a_target_mpc:

selfdrive/selfdrived/selfdrived.py

@@ -321,9 +321,16 @@ class SelfdriveD(CruiseHelper):
     # Handle lane change
     if self.sm['modelV2'].meta.laneChangeState == LaneChangeState.preLaneChange:
       direction = self.sm['modelV2'].meta.laneChangeDirection
+      mdv2sp = self.sm['modelDataV2SP']
+
       if (CS.leftBlindspot and direction == LaneChangeDirection.left) or \
-         (CS.rightBlindspot and direction == LaneChangeDirection.right):
+            (CS.rightBlindspot and direction == LaneChangeDirection.right):
         self.events.add(EventName.laneChangeBlocked)
+
+      elif (mdv2sp.leftLaneChangeEdgeBlock and direction == LaneChangeDirection.left) or \
+           (mdv2sp.rightLaneChangeEdgeBlock and direction == LaneChangeDirection.right):
+        self.events_sp.add(custom.OnroadEventSP.EventName.laneChangeRoadEdge)
+
       else:
         if direction == LaneChangeDirection.left:
           self.events.add(EventName.preLaneChangeLeft)

selfdrive/test/longitudinal_maneuvers/plant.py

@@ -1,5 +1,11 @@
 #!/usr/bin/env python3
+from collections import deque
+from collections.abc import Callable
+from dataclasses import dataclass
+import math
 import time
+from typing import Any
+
 import numpy as np
 
 from cereal import log
@@ -11,12 +17,105 @@ from openpilot.selfdrive.controls.lib.longitudinal_planner import LongitudinalPl
 from openpilot.selfdrive.controls.radard import _LEAD_ACCEL_TAU
 
 
+LeadObservation = dict[str, Any]
+LeadObservationFn = Callable[[float, str, LeadObservation], LeadObservation | None]
+ModelActionFn = Callable[[float, float, float], tuple[float, bool]]
+EgoObservationFn = Callable[[float, float, float], tuple[float, float]]
+
+
+@dataclass(frozen=True)
+class ActuatorModel:
+  planner_delay: float
+  transport_delay: float
+  actuator_lag: float
+  command_rate_limit: float
+  stopping_acceleration: float
+  standstill_breakaway_acceleration: float
+  standstill_breakaway_time: float
+
+  def __post_init__(self):
+    nonnegative_fields = {
+      "planner_delay": self.planner_delay,
+      "transport_delay": self.transport_delay,
+      "actuator_lag": self.actuator_lag,
+      "standstill_breakaway_acceleration": self.standstill_breakaway_acceleration,
+      "standstill_breakaway_time": self.standstill_breakaway_time,
+    }
+    if any(not math.isfinite(value) or value < 0.0 for value in nonnegative_fields.values()):
+      raise ValueError(f"ActuatorModel fields must be finite and non-negative: {nonnegative_fields}")
+    if not math.isfinite(self.command_rate_limit) or self.command_rate_limit <= 0.0:
+      raise ValueError("command_rate_limit must be finite and positive")
+    if not math.isfinite(self.stopping_acceleration) or self.stopping_acceleration > 0.0:
+      raise ValueError("stopping_acceleration must be finite and non-positive")
+
+
+# Route-derived conservative Prius TSS2 stress model for the acceleration-controller
+# regression suite. The 1.0 m/s² gate represents prompt takeoffs, not a universal
+# physical threshold: the supplied routes also contain low-command creep departures.
+# This models vehicle response only and does not emulate Toyota's CAN controller.
+PRIUS_TSS2_ROUTE_MODEL = ActuatorModel(
+  planner_delay=0.05,
+  transport_delay=0.0,
+  actuator_lag=0.20,
+  command_rate_limit=4.0,
+  stopping_acceleration=-2.0,
+  standstill_breakaway_acceleration=1.0,
+  standstill_breakaway_time=0.05,
+)
+
+
 class Plant:
   messaging_initialized = False
 
-  def __init__(self, lead_relevancy=False, speed=0.0, distance_lead=2.0,
-               enabled=True, only_lead2=False, only_radar=False, e2e=False, personality=0, force_decel=False):
-    self.rate = 1. / DT_MDL
+  def __init__(
+    self,
+    lead_relevancy=False,
+    speed=0.0,
+    distance_lead=2.0,
+    enabled=True,
+    only_lead2=False,
+    only_radar=False,
+    e2e=False,
+    personality=0,
+    force_decel=False,
+    lead_observation_fn: LeadObservationFn | None = None,
+    model_action_fn: ModelActionFn | None = None,
+    ego_observation_fn: EgoObservationFn | None = None,
+    actuator_delay: float | None = None,
+    actuator_lag: float = 0.0,
+    actuator_model: ActuatorModel | None = None,
+  ):
+    """Closed-loop longitudinal planner plant.
+
+    ``lead_observation_fn(time, lead_name, truth)`` may return a complete or partial
+    observed LeadData mapping, or ``None`` for an absent lead. It is called separately
+    for ``leadOne`` and ``leadTwo``. The supplied truth mapping is a copy, and observed
+    values never affect the physical lead trajectory.
+
+    ``model_action_fn(time, v_ego, a_ego)`` returns
+    ``(desired_acceleration, should_stop)``.
+
+    ``ego_observation_fn(time, true_v_ego, true_a_ego)`` returns the observed
+    ``(v_ego, a_ego)`` published in ``carState``. It can inject measurement noise
+    without changing the physical plant state.
+
+    Passing ``actuator_delay`` both overrides ``CP.longitudinalActuatorDelay`` and
+    adds the corresponding command transport delay to the plant. ``None`` keeps the
+    historical Honda planner delay with instantaneous plant response. ``actuator_lag``
+    is an optional first-order acceleration-response time constant. Both defaults keep
+    historical plant dynamics unchanged.
+
+    ``actuator_model`` opts into a staged vehicle-response model. Its planner delay
+    is used by MPC, while its independent transport delay is used by the command
+    queue before rate limiting, standstill breakaway confirmation, and first-order
+    lag. Leaving it unset preserves the historical actuator path.
+    """
+    if actuator_delay is not None and (not math.isfinite(actuator_delay) or actuator_delay < 0.0):
+      raise ValueError("actuator_delay must be finite and non-negative")
+    if not math.isfinite(actuator_lag) or actuator_lag < 0.0:
+      raise ValueError("actuator_lag must be finite and non-negative")
+
+    self.rate = 1.0 / DT_MDL
 
     if not Plant.messaging_initialized:
       Plant.radar = messaging.pub_sock('radarState')
@@ -28,10 +127,15 @@ class Plant:
 
     self.v_lead_prev = 0.0
 
-    self.distance = 0.
+    self.distance = 0.0
     self.speed = speed
     self.should_stop = False
     self.acceleration = 0.0
+    self.a_target = 0.0
+    self.actuator_command = 0.0
+    self.applied_actuator_command = 0.0
+    self.breakaway_confirmed = False
+    self._breakaway_timer = 0.0
 
     # lead car
     self.lead_relevancy = lead_relevancy
@@ -42,9 +146,18 @@ class Plant:
     self.e2e = e2e
     self.personality = personality
     self.force_decel = force_decel
+    self.lead_observation_fn = lead_observation_fn
+    self.model_action_fn = model_action_fn
+    self.ego_observation_fn = ego_observation_fn
+    self.actuator_model = actuator_model
+    self.actuator_delay = actuator_model.planner_delay if actuator_model is not None else actuator_delay
+    self.transport_delay = actuator_model.transport_delay if actuator_model is not None else actuator_delay
+    self.actuator_lag = actuator_model.actuator_lag if actuator_model is not None else actuator_lag
+    self.publish_realized_a_ego = any((lead_observation_fn is not None, model_action_fn is not None, ego_observation_fn is not None,
+                                      actuator_delay is not None, actuator_lag > 0.0, actuator_model is not None))
 
     self.rk = Ratekeeper(self.rate, print_delay_threshold=100.0)
-    self.ts = 1. / self.rate
+    self.ts = 1.0 / self.rate
     time.sleep(0.1)
     self.sm = messaging.SubMaster(['longitudinalPlan'])
 
@@ -52,14 +165,86 @@ class Plant:
     from opendbc.car.honda.interface import CarInterface
 
     CP = CarInterface.get_non_essential_params(CAR.HONDA_CIVIC)
+    if self.actuator_delay is not None:
+      CP.longitudinalActuatorDelay = self.actuator_delay
     CP_SP = CarInterface.get_non_essential_params_sp(CP, CAR.HONDA_CIVIC)
     self.planner = LongitudinalPlanner(CP, CP_SP, init_v=self.speed)
 
+    if self.actuator_model is not None and self.speed >= 0.01:
+      self.breakaway_confirmed = True
+    delay_steps = 0 if self.transport_delay is None else round(self.transport_delay / self.ts)
+    self._actuator_delay_queue = deque([self.acceleration] * delay_steps)
+
   @property
   def current_time(self):
     return float(self.rk.frame) / self.rate
 
-  def step(self, v_lead=0.0, prob_lead=1.0, v_cruise=50., pitch=0.0, prob_throttle=1.0):
+  @staticmethod
+  def _lead_message(observation: LeadObservation):
+    lead = log.RadarState.LeadData.new_message()
+    for field, value in observation.items():
+      setattr(lead, field, value)
+    return lead
+
+  def _observe_lead(self, lead_name: str, truth: LeadObservation, present_by_default: bool) -> LeadObservation | None:
+    if self.lead_observation_fn is None:
+      return dict(truth) if present_by_default else None
+
+    observed = self.lead_observation_fn(self.current_time, lead_name, dict(truth))
+    if observed is None:
+      return None
+
+    # Partial overrides are convenient for individual sensor glitches, while copying
+    # from truth ensures every field written to cereal is deterministic.
+    complete_observation = dict(truth)
+    complete_observation.update(observed)
+    return complete_observation
+
+  def _update_actuator(self, command: float) -> tuple[float, float]:
+    if self._actuator_delay_queue:
+      self._actuator_delay_queue.append(command)
+      delayed_command = self._actuator_delay_queue.popleft()
+    else:
+      delayed_command = command
+
+    if self.actuator_model is not None:
+      max_command_delta = self.actuator_model.command_rate_limit * self.ts
+      self.applied_actuator_command = float(np.clip(delayed_command,
+                                                   self.applied_actuator_command - max_command_delta,
+                                                   self.applied_actuator_command + max_command_delta))
+
+      if self.speed < 0.01:
+        if self.applied_actuator_command <= 0.0:
+          self.breakaway_confirmed = False
+          self._breakaway_timer = 0.0
+        elif not self.breakaway_confirmed:
+          breakaway_ready = self.applied_actuator_command + 1e-9 >= self.actuator_model.standstill_breakaway_acceleration
+          if breakaway_ready:
+            self._breakaway_timer += self.ts
+          else:
+            self._breakaway_timer = 0.0
+
+          self.breakaway_confirmed = breakaway_ready and self._breakaway_timer + 1e-9 >= self.actuator_model.standstill_breakaway_time
+        if not self.breakaway_confirmed:
+          self.acceleration = 0.0
+          return delayed_command, self.acceleration
+      else:
+        self.breakaway_confirmed = True
+
+      response_command = self.applied_actuator_command
+    else:
+      # Preserve the historical response path exactly when no staged model is used.
+      self.applied_actuator_command = delayed_command
+      response_command = delayed_command
+
+    if self.actuator_lag > 0.0:
+      alpha = 1.0 - math.exp(-self.ts / self.actuator_lag)
+      self.acceleration += alpha * (response_command - self.acceleration)
+    else:
+      self.acceleration = response_command
+    return delayed_command, self.acceleration
+
+  def step(self, v_lead=0.0, prob_lead=1.0, v_cruise=50.0, pitch=0.0, prob_throttle=1.0):
     # ******** publish a fake model going straight and fake calibration ********
     # note that this is worst case for MPC, since model will delay long mpc by one time step
     radar = messaging.new_message('radarState')
@@ -72,39 +257,48 @@ class Plant:
     car_state_sp = messaging.new_message('carStateSP')
     live_map_data_sp = messaging.new_message('liveMapDataSP')
     gps_data = messaging.new_message('gpsLocation')
-    a_lead = (v_lead - self.v_lead_prev)/self.ts
+    a_lead = (v_lead - self.v_lead_prev) / self.ts
     self.v_lead_prev = v_lead
 
     if self.lead_relevancy:
-      d_rel = np.maximum(0., self.distance_lead - self.distance)
+      d_rel = np.maximum(0.0, self.distance_lead - self.distance)
       v_rel = v_lead - self.speed
       if self.only_radar:
         status = True
-      elif prob_lead > .5:
+      elif prob_lead > 0.5:
         status = True
       else:
         status = False
     else:
-      d_rel = 200.
-      v_rel = 0.
+      d_rel = 200.0
+      v_rel = 0.0
       prob_lead = 0.0
       status = False
 
-    lead = log.RadarState.LeadData.new_message()
-    lead.dRel = float(d_rel)
-    lead.yRel = 0.0
-    lead.vRel = float(v_rel)
-    lead.aRel = float(a_lead - self.acceleration)
-    lead.vLead = float(v_lead)
-    lead.vLeadK = float(v_lead)
-    lead.aLeadK = float(a_lead)
-    # TODO use real radard logic for this
-    lead.aLeadTau = float(_LEAD_ACCEL_TAU)
-    lead.status = status
-    lead.modelProb = float(prob_lead)
-    if not self.only_lead2:
-      radar.radarState.leadOne = lead
-    radar.radarState.leadTwo = lead
+    truth_lead: LeadObservation = {
+      "dRel": float(d_rel),
+      "yRel": 0.0,
+      "vRel": float(v_rel),
+      "aRel": float(a_lead - self.acceleration),
+      "vLead": float(v_lead),
+      "dPath": 0.0,
+      "vLat": 0.0,
+      "vLeadK": float(v_lead),
+      "aLeadK": float(a_lead),
+      "fcw": False,
+      "status": bool(status),
+      # TODO use real radard logic for this
+      "aLeadTau": float(_LEAD_ACCEL_TAU),
+      "modelProb": float(prob_lead),
+      "radar": bool(self.only_radar),
+      "radarTrackId": -1,
+    }
+    lead_one_observation = self._observe_lead("leadOne", truth_lead, not self.only_lead2)
+    lead_two_observation = self._observe_lead("leadTwo", truth_lead, True)
+    if lead_one_observation is not None:
+      radar.radarState.leadOne = self._lead_message(lead_one_observation)
+    if lead_two_observation is not None:
+      radar.radarState.leadTwo = self._lead_message(lead_two_observation)
 
     # Simulate model predicting slightly faster speed
     # this is to ensure lead policy is effective when model
@@ -112,10 +306,15 @@ class Plant:
     position = log.XYZTData.new_message()
     position.x = [float(x) for x in (self.speed + 0.5) * np.array(ModelConstants.T_IDXS)]
     model.modelV2.position = position
-    model.modelV2.action.desiredAcceleration = float(self.acceleration + 0.1)
+    if self.model_action_fn is None:
+      model_acceleration, model_should_stop = self.acceleration + 0.1, False
+    else:
+      model_acceleration, model_should_stop = self.model_action_fn(self.current_time, self.speed, self.acceleration)
+    model.modelV2.action.desiredAcceleration = float(model_acceleration)
+    model.modelV2.action.shouldStop = bool(model_should_stop)
     velocity = log.XYZTData.new_message()
     velocity.x = [float(x) for x in (self.speed + 0.5) * np.ones_like(ModelConstants.T_IDXS)]
-    velocity.x[0] = float(self.speed) # always start at current speed
+    velocity.x[0] = float(self.speed)  # always start at current speed
     model.modelV2.velocity = velocity
     acceleration = log.XYZTData.new_message()
     acceleration.x = [float(x) for x in np.zeros_like(ModelConstants.T_IDXS)]
@@ -126,33 +325,45 @@ class Plant:
     ss.selfdriveState.experimentalMode = self.e2e
     ss.selfdriveState.personality = self.personality
     control.controlsState.forceDecel = self.force_decel
-    car_state.carState.vEgo = float(self.speed)
+    true_v_ego = self.speed
+    true_a_ego = self.acceleration
+    published_v_ego = true_v_ego
+    published_a_ego = true_a_ego if self.publish_realized_a_ego else 0.0
+    if self.ego_observation_fn is not None:
+      published_v_ego, published_a_ego = self.ego_observation_fn(self.current_time, true_v_ego, true_a_ego)
+    car_state.carState.vEgo = float(published_v_ego)
+    car_state.carState.aEgo = float(published_a_ego)
     car_state.carState.standstill = bool(self.speed < 0.01)
     car_state.carState.vCruise = float(v_cruise * 3.6)
-    car_control.carControl.orientationNED = [0., float(pitch), 0.]
+    car_control.carControl.orientationNED = [0.0, float(pitch), 0.0]
 
     # ******** get controlsState messages for plotting ***
-    sm = {'radarState': radar.radarState,
-          'carState': car_state.carState,
-          'carControl': car_control.carControl,
-          'controlsState': control.controlsState,
-          'selfdriveState': ss.selfdriveState,
-          'liveParameters': lp.liveParameters,
-          'modelV2': model.modelV2,
-          'carStateSP': car_state_sp.carStateSP,
-          'liveMapDataSP': live_map_data_sp.liveMapDataSP,
-          'gpsLocation': gps_data.gpsLocation}
+    sm = {
+      'radarState': radar.radarState,
+      'carState': car_state.carState,
+      'carControl': car_control.carControl,
+      'controlsState': control.controlsState,
+      'selfdriveState': ss.selfdriveState,
+      'liveParameters': lp.liveParameters,
+      'modelV2': model.modelV2,
+      'carStateSP': car_state_sp.carStateSP,
+      'liveMapDataSP': live_map_data_sp.liveMapDataSP,
+      'gpsLocation': gps_data.gpsLocation,
+    }
     self.planner.update(sm)
-    self.acceleration = self.planner.output_a_target
+    self.a_target = self.planner.output_a_target
+    self.actuator_command = self.a_target
     if self.planner.output_should_stop:
-      self.acceleration = min(-0.5, self.acceleration)
+      stopping_acceleration = -0.5 if self.actuator_model is None else self.actuator_model.stopping_acceleration
+      self.actuator_command = min(stopping_acceleration, self.actuator_command)
+    delayed_actuator_command, _ = self._update_actuator(self.actuator_command)
     self.speed = self.speed + self.acceleration * self.ts
     self.should_stop = self.planner.output_should_stop
     fcw = self.planner.fcw
     self.distance_lead = self.distance_lead + v_lead * self.ts
 
     # ******** run the car ********
-    #print(self.distance, speed)
+    # print(self.distance, speed)
     if self.speed <= 0:
       self.speed = 0
       self.acceleration = 0
@@ -160,30 +371,64 @@ class Plant:
 
     # *** radar model ***
     if self.lead_relevancy:
-      d_rel = np.maximum(0., self.distance_lead - self.distance)
+      d_rel = np.maximum(0.0, self.distance_lead - self.distance)
       v_rel = v_lead - self.speed
     else:
-      d_rel = 200.
-      v_rel = 0.
+      d_rel = 200.0
+      v_rel = 0.0
 
     # print at 5hz
     # if (self.rk.frame % (self.rate // 5)) == 0:
     #   print("%2.2f sec   %6.2f m  %6.2f m/s  %6.2f m/s2   lead_rel: %6.2f m  %6.2f m/s"
     #         % (self.current_time, self.distance, self.speed, self.acceleration, d_rel, v_rel))
 
-
     # ******** update prevs ********
     self.rk.monitor_time()
 
+    accel_controller_result = getattr(self.planner, "accel_controller_result", None)
     return {
       "distance": self.distance,
       "speed": self.speed,
       "acceleration": self.acceleration,
+      "realized_acceleration": self.acceleration,
+      "a_target": self.a_target,
+      "planner_acceleration": self.a_target,
+      "actuator_command": self.actuator_command,
+      "stop_clamped_actuator_command": self.actuator_command,
+      "delayed_actuator_command": delayed_actuator_command,
+      "applied_actuator_command": self.applied_actuator_command,
+      "vehicle_actuator_command": self.applied_actuator_command,
+      "true_v_ego": true_v_ego,
+      "true_a_ego": true_a_ego,
+      "published_a_ego": published_a_ego,
+      "published_v_ego": published_v_ego,
+      "observed_a_ego": published_a_ego,
+      "observed_v_ego": published_v_ego,
+      "planner_delay": self.actuator_delay,
+      "transport_delay": self.transport_delay,
+      "breakaway_confirmed": self.breakaway_confirmed,
+      "breakaway_time": self._breakaway_timer,
       "should_stop": self.should_stop,
       "distance_lead": self.distance_lead,
       "fcw": fcw,
+      "mpc_source": self.planner.mpc.source,
+      "dec_mode": self.planner.dec.mode(),
+      "pace_cap": getattr(accel_controller_result, "target_speed", None),
+      "base_target": getattr(accel_controller_result, "base_speed", None),
+      "raw_energy_cap": getattr(accel_controller_result, "raw_energy_cap", None),
+      "live_filtered_cap": getattr(accel_controller_result, "live_filtered_cap", None),
+      "shadow_filtered_cap": getattr(accel_controller_result, "shadow_filtered_cap", None),
+      "accel_controller_selected_lead": getattr(accel_controller_result, "selected_lead", None),
+      "model_action": {
+        "desiredAcceleration": float(model_acceleration),
+        "shouldStop": bool(model_should_stop),
+      },
+      "truth_lead": dict(truth_lead),
+      "lead_one_observation": None if lead_one_observation is None else dict(lead_one_observation),
+      "lead_two_observation": None if lead_two_observation is None else dict(lead_two_observation),
     }
 
+
 # simple engage in standalone mode
 def plant_thread():
   plant = Plant()

selfdrive/test/longitudinal_maneuvers/test_plant.py

@@ -0,0 +1,80 @@
+import math
+
+import pytest
+
+from openpilot.common.realtime import DT_MDL
+from openpilot.selfdrive.test.longitudinal_maneuvers.plant import Plant
+
+
+def test_full_lead_observation_is_independent_from_truth():
+  callback_inputs = []
+
+  def observe_lead(current_time, lead_name, truth):
+    callback_inputs.append((current_time, lead_name, truth))
+    if lead_name == "leadOne":
+      return {
+        "dRel": 12.5,
+        "vRel": -4.0,
+        "vLead": 6.0,
+        "vLeadK": 5.5,
+        "aLeadK": -1.25,
+        "aLeadTau": 0.7,
+        "status": True,
+        "modelProb": 0.9,
+        "radarTrackId": 42,
+      }
+    return None
+
+  plant = Plant(lead_relevancy=True, speed=10.0, distance_lead=50.0, lead_observation_fn=observe_lead)
+  result = plant.step(v_lead=8.0)
+
+  assert [entry[1] for entry in callback_inputs] == ["leadOne", "leadTwo"]
+  assert callback_inputs[0][2]["dRel"] == pytest.approx(50.0)
+  assert result["truth_lead"]["dRel"] == pytest.approx(50.0)
+  assert result["lead_one_observation"]["dRel"] == pytest.approx(12.5)
+  assert result["lead_one_observation"]["radarTrackId"] == 42
+  assert result["lead_two_observation"] is None
+  assert result["distance_lead"] == pytest.approx(50.0 + 8.0 * DT_MDL)
+
+
+def test_model_action_realized_acceleration_and_source_logging():
+  def model_action(current_time, v_ego, a_ego):
+    return -1.25, True
+
+  plant = Plant(speed=10.0, e2e=True, force_decel=True, model_action_fn=model_action, actuator_lag=0.5)
+  first = plant.step()
+  second = plant.step()
+
+  assert first["model_action"] == {"desiredAcceleration": -1.25, "shouldStop": True}
+  assert first["published_a_ego"] == pytest.approx(0.0)
+  assert second["published_a_ego"] == pytest.approx(first["realized_acceleration"])
+  assert first["acceleration"] == first["realized_acceleration"]
+  assert abs(first["realized_acceleration"]) < abs(first["actuator_command"])
+  assert first["mpc_source"] is not None
+  assert first["dec_mode"] in ("acc", "blended")
+  assert "pace_cap" in first
+  assert "raw_energy_cap" in first
+  assert "live_filtered_cap" in first
+  assert first["lead_one_observation"] is not None
+  assert first["truth_lead"] == first["lead_one_observation"]
+
+
+def test_configurable_transport_delay_and_first_order_lag():
+  plant = Plant(speed=10.0, actuator_delay=2 * DT_MDL, actuator_lag=0.2)
+
+  assert plant.planner.CP.longitudinalActuatorDelay == pytest.approx(2 * DT_MDL)
+  delayed_commands = [plant._update_actuator(-1.0) for _ in range(3)]
+  assert [command for command, _ in delayed_commands[:2]] == [0.0, 0.0]
+
+  expected_acceleration = -(1.0 - math.exp(-DT_MDL / 0.2))
+  assert delayed_commands[2][0] == -1.0
+  assert delayed_commands[2][1] == pytest.approx(expected_acceleration)
+
+
+@pytest.mark.parametrize(
+  ("delay", "lag"),
+  [(-0.1, 0.0), (float("nan"), 0.0), (float("inf"), 0.0), (None, -0.1), (None, float("nan")), (None, float("inf"))],
+)
+def test_invalid_actuator_dynamics(delay, lag):
+  with pytest.raises(ValueError):
+    Plant(actuator_delay=delay, actuator_lag=lag)

selfdrive/ui/layouts/settings/toggles.py

@@ -27,6 +27,12 @@ DESCRIPTIONS = {
     "In relaxed mode sunnypilot will stay further away from lead cars. On supported cars, you can cycle through these personalities with " +
     "your steering wheel distance button."
   ),
+  "AccelPersonalityEnabled": tr_noop(
+    "Begin slowing early and smoothly behind lead vehicles. Stock longitudinal control retains braking and stopping authority."
+  ),
+  "AccelPersonality": tr_noop(
+    "Eco slows earliest and recovers gently, Normal balances comfort and response, and Sport reacts and recovers more quickly."
+  ),
   "IsLdwEnabled": tr_noop(
     "Receive alerts to steer back into the lane when your vehicle drifts over a detected lane line " +
     "without a turn signal activated while driving over 31 mph (50 km/h)."
@@ -106,6 +112,24 @@ class TogglesLayout(Widget):
       icon="speed_limit.png"
     )
 
+    self._accel_personality_enabled = toggle_item(
+      lambda: tr("Enable Accel Controller"),
+      lambda: tr(DESCRIPTIONS["AccelPersonalityEnabled"]),
+      self._params.get_bool("AccelPersonalityEnabled"),
+      callback=self._set_accel_personality_enabled,
+      icon="speed_limit.png",
+    )
+
+    self._accel_personality_setting = multiple_button_item(
+      lambda: tr("Acceleration Profile"),
+      lambda: tr(DESCRIPTIONS["AccelPersonality"]),
+      buttons=[lambda: tr("Eco"), lambda: tr("Normal"), lambda: tr("Sport")],
+      button_width=300,
+      callback=self._set_accel_personality,
+      selected_index=self._params.get("AccelPersonality", return_default=True),
+      icon="speed_limit.png"
+    )
+
     self._toggles = {}
     self._locked_toggles = set()
     for param, (title, desc, icon, needs_restart) in self._toggle_defs.items():
@@ -135,9 +159,11 @@ class TogglesLayout(Widget):
 
       self._toggles[param] = toggle
 
-      # insert longitudinal personality after NDOG toggle
+      # insert longitudinal personality and Accel Controller settings after NDOG toggle
       if param == "DisengageOnAccelerator":
         self._toggles["LongitudinalPersonality"] = self._long_personality_setting
+        self._toggles["AccelPersonalityEnabled"] = self._accel_personality_enabled
+        self._toggles["AccelPersonality"] = self._accel_personality_setting
 
     self._update_experimental_mode_icon()
     self._scroller = Scroller(list(self._toggles.values()), line_separator=True, spacing=0)
@@ -158,6 +184,7 @@ class TogglesLayout(Widget):
 
   def _update_toggles(self):
     ui_state.update_params()
+    accel_personality_enabled = self._params.get_bool("AccelPersonalityEnabled")
 
     e2e_description = tr(
       "sunnypilot defaults to driving in chill mode. Experimental mode enables alpha-level features that aren't ready for chill mode. " +
@@ -176,11 +203,15 @@ class TogglesLayout(Widget):
         self._toggles["ExperimentalMode"].action_item.set_enabled(True)
         self._toggles["ExperimentalMode"].set_description(e2e_description)
         self._long_personality_setting.action_item.set_enabled(True)
+        self._accel_personality_enabled.action_item.set_enabled(True)
+        self._accel_personality_setting.action_item.set_enabled(accel_personality_enabled)
       else:
         # no long for now
         self._toggles["ExperimentalMode"].action_item.set_enabled(False)
         self._toggles["ExperimentalMode"].action_item.set_state(False)
         self._long_personality_setting.action_item.set_enabled(False)
+        self._accel_personality_enabled.action_item.set_enabled(False)
+        self._accel_personality_setting.action_item.set_enabled(False)
         self._params.remove("ExperimentalMode")
 
         unavailable = tr("Experimental mode is currently unavailable on this car since the car's stock ACC is used for longitudinal control.")
@@ -203,6 +234,10 @@ class TogglesLayout(Widget):
     # refresh toggles from params to mirror external changes
     for param in self._toggle_defs:
       self._toggles[param].action_item.set_state(self._params.get_bool(param))
+    self._accel_personality_enabled.action_item.set_state(accel_personality_enabled)
+    self._accel_personality_setting.action_item.set_selected_button(
+      self._params.get("AccelPersonality", return_default=True)
+    )
 
     # these toggles need restart, block while engaged
     for toggle_def in self._toggle_defs:
@@ -247,3 +282,10 @@ class TogglesLayout(Widget):
 
   def _set_longitudinal_personality(self, button_index: int):
     self._params.put("LongitudinalPersonality", button_index, block=True)
+
+  def _set_accel_personality(self, button_index: int):
+    self._params.put("AccelPersonality", button_index, block=True)
+
+  def _set_accel_personality_enabled(self, state: bool):
+    self._params.put_bool("AccelPersonalityEnabled", state, block=True)
+    self._accel_personality_setting.action_item.set_enabled(state and ui_state.has_longitudinal_control)

selfdrive/ui/mici/layouts/main.py

@@ -13,6 +13,7 @@ from openpilot.system.ui.lib.application import gui_app
 
 if gui_app.sunnypilot_ui():
   from openpilot.selfdrive.ui.sunnypilot.mici.layouts.settings import SettingsLayoutSP as SettingsLayout
+  from openpilot.selfdrive.ui.sunnypilot.mici.layouts.onroad import OnroadViewContainerSP as AugmentedRoadView
 
 ONROAD_DELAY = 2.5  # seconds
 
@@ -118,13 +119,15 @@ class MiciMainLayout(Scroller):
 
     # FIXME: these two pops can interrupt user interacting in the settings
     if self._onroad_time_delay is not None and rl.get_time() - self._onroad_time_delay >= ONROAD_DELAY:
-      gui_app.pop_widgets_to(self, lambda: self._scroll_to(self._onroad_layout))
+      if not gui_app.sunnypilot_ui() or self._should_auto_scroll_to_onroad():
+        gui_app.pop_widgets_to(self, lambda: self._scroll_to(self._onroad_layout))
       self._onroad_time_delay = None
 
     # When car leaves standstill, pop nav stack and scroll to onroad
     CS = ui_state.sm["carState"]
     if not CS.standstill and self._prev_standstill:
-      gui_app.pop_widgets_to(self, lambda: self._scroll_to(self._onroad_layout))
+      if not gui_app.sunnypilot_ui() or self._should_auto_scroll_to_onroad():
+        gui_app.pop_widgets_to(self, lambda: self._scroll_to(self._onroad_layout))
     self._prev_standstill = CS.standstill
 
   def _on_interactive_timeout(self):

selfdrive/ui/mici/layouts/settings/toggles.py

@@ -14,6 +14,8 @@ class TogglesLayoutMici(NavScroller):
     super().__init__()
 
     self._personality_toggle = BigMultiParamToggle("driving personality", "LongitudinalPersonality", ["aggressive", "standard", "relaxed"])
+    self._accel_personality_enabled = BigParamControl("enable accel controller", "AccelPersonalityEnabled")
+    self._accel_personality_toggle = BigMultiParamToggle("acceleration profile", "AccelPersonality", ["eco", "normal", "sport"])
     self._experimental_btn = BigParamControl("experimental mode", "ExperimentalMode")
     is_metric_toggle = BigParamControl("use metric units", "IsMetric")
     ldw_toggle = BigParamControl("lane departure warnings", "IsLdwEnabled")
@@ -24,6 +26,8 @@ class TogglesLayoutMici(NavScroller):
 
     self._scroller.add_widgets([
       self._personality_toggle,
+      self._accel_personality_enabled,
+      self._accel_personality_toggle,
       self._experimental_btn,
       is_metric_toggle,
       ldw_toggle,
@@ -36,6 +40,7 @@ class TogglesLayoutMici(NavScroller):
     # Toggle lists
     self._refresh_toggles = (
       ("ExperimentalMode", self._experimental_btn),
+      ("AccelPersonalityEnabled", self._accel_personality_enabled),
       ("IsMetric", is_metric_toggle),
       ("IsLdwEnabled", ldw_toggle),
       ("AlwaysOnDM", always_on_dm_toggle),
@@ -45,6 +50,9 @@ class TogglesLayoutMici(NavScroller):
     )
 
     enable_openpilot.set_enabled(lambda: not ui_state.engaged)
+    self._accel_personality_toggle.set_enabled(
+      lambda: ui_state.has_longitudinal_control and ui_state.params.get_bool("AccelPersonalityEnabled")
+    )
     record_front.set_enabled(False if ui_state.params.get_bool("RecordFrontLock") else (lambda: not ui_state.engaged))
     record_mic.set_enabled(lambda: not ui_state.engaged)
 
@@ -75,13 +83,18 @@ class TogglesLayoutMici(NavScroller):
       if ui_state.has_longitudinal_control:
         self._experimental_btn.set_visible(True)
         self._personality_toggle.set_visible(True)
+        self._accel_personality_enabled.set_visible(True)
+        self._accel_personality_toggle.set_visible(True)
       else:
         # no long for now
         self._experimental_btn.set_visible(False)
         self._experimental_btn.set_checked(False)
         self._personality_toggle.set_visible(False)
+        self._accel_personality_enabled.set_visible(False)
+        self._accel_personality_toggle.set_visible(False)
         ui_state.params.remove("ExperimentalMode")
 
     # Refresh toggles from params to mirror external changes
     for key, item in self._refresh_toggles:
       item.set_checked(ui_state.params.get_bool(key))
+    self._accel_personality_toggle.refresh()

selfdrive/ui/mici/widgets/button.py

@@ -382,13 +382,18 @@ class BigMultiParamToggle(BigMultiToggle):
     self._load_value()
 
   def _load_value(self):
-    self.set_value(self._options[self._params.get(self._param) or 0])
+    value = self._params.get(self._param, return_default=True)
+    index = value if isinstance(value, int) else 0
+    self.set_value(self._options[max(0, min(index, len(self._options) - 1))])
 
   def _handle_mouse_release(self, mouse_pos: MousePos):
     super()._handle_mouse_release(mouse_pos)
     new_idx = self._options.index(self.value)
     self._params.put(self._param, new_idx)
 
+  def refresh(self):
+    self._load_value()
+
 
 class BigParamControl(BigToggle):
   def __init__(self, text: str, param: str, toggle_callback: Callable | None = None):

selfdrive/ui/sunnypilot/layouts/settings/steering_sub_layouts/lane_change_settings.py

@@ -51,11 +51,17 @@ class LaneChangeSettingsLayout(Widget):
       description=lambda: tr("Toggle to enable a delay timer for seamless lane changes when blind spot monitoring " +
                              "(BSM) detects a obstructing vehicle, ensuring safe maneuvering."),
     )
+    self._road_edge_block = toggle_item_sp(
+      param="RoadEdgeLaneChangeEnabled",
+      title=lambda: tr("Block Lane Change: Road Edge Detection"),
+      description=lambda: tr("Blocks the lane change if the model sees a road edge on your signaled side."),
+    )
 
     items = [
       self._lane_change_timer,
       LineSeparatorSP(40),
       self._bsm_delay,
+      self._road_edge_block,
     ]
 
     return items

selfdrive/ui/sunnypilot/mici/layouts/main.py

@@ -0,0 +1,13 @@
+"""
+Copyright (c) 2021-, Haibin Wen, sunnypilot, and a number of other contributors.
+
+This file is part of sunnypilot and is licensed under the MIT License.
+See the LICENSE.md file in the root directory for more details.
+"""
+
+from openpilot.selfdrive.ui.mici.layouts.main import MiciMainLayout
+
+
+class MiciMainLayoutSP(MiciMainLayout):
+  def _should_auto_scroll_to_onroad(self) -> bool:
+    return not self._onroad_layout.is_on_info_panel()

selfdrive/ui/sunnypilot/mici/layouts/onroad.py

@@ -0,0 +1,63 @@
+"""
+Copyright (c) 2021-, Haibin Wen, sunnypilot, and a number of other contributors.
+
+This file is part of sunnypilot and is licensed under the MIT License.
+See the LICENSE.md file in the root directory for more details.
+"""
+import pyray as rl
+from openpilot.system.ui.lib.application import gui_app
+from openpilot.selfdrive.ui.sunnypilot.mici.widgets.scroller_sp import ScrollerSP
+from openpilot.selfdrive.ui.sunnypilot.mici.onroad.augmented_road_view import AugmentedRoadViewSP
+from openpilot.selfdrive.ui.sunnypilot.mici.layouts.onroad_info_panel import OnroadInfoPanel
+
+CONFIDENCE_BALL_VISIBLE_RATIO = 0.4
+HORIZONTAL_SETTLE_PX = 5
+HORIZONTAL_RESET_RATIO = 0.5
+
+
+class OnroadViewContainerSP(ScrollerSP):
+  def __init__(self, bookmark_callback=None):
+    super().__init__(horizontal=False, snap_items=True, spacing=0, pad=0, scroll_indicator=False, edge_shadows=False)
+    self.road_view = AugmentedRoadViewSP(bookmark_callback=bookmark_callback)
+    self.onroad_info_panel = OnroadInfoPanel(bookmark_callback=bookmark_callback)
+
+    self._scroller.add_widgets([
+      self.road_view,
+      self.onroad_info_panel,
+    ])
+    self._scroller.set_reset_scroll_at_show(False)
+    self._scroller.set_scrolling_enabled(lambda: abs(self.rect.x) < HORIZONTAL_SETTLE_PX)
+
+    for child in (self.road_view, self.onroad_info_panel):
+      inner_touch_valid = child._touch_valid_callback
+      child.set_touch_valid_callback(
+        lambda inner=inner_touch_valid: self._touch_valid() and (inner() if inner else True)
+      )
+
+  def set_rect(self, rect: rl.Rectangle):
+    super().set_rect(rect)
+    self.road_view.set_rect(rect)
+    self.onroad_info_panel.set_rect(rect)
+    return self
+
+  def is_swiping_left(self) -> bool:
+    return self.road_view.is_swiping_left() or self.onroad_info_panel.is_swiping_left()
+
+  def set_click_callback(self, callback) -> None:
+    self.road_view.set_click_callback(callback)
+    self.onroad_info_panel.set_click_callback(callback)
+
+  def is_on_info_panel(self) -> bool:
+    """True when scrolled past halfway toward onroad_info_panel (used by main layout
+    to skip auto-pop-back-to-camera while user is reading the info panel)."""
+    return abs(self._scroller.scroll_panel.get_offset()) > self._rect.height / 2
+
+  def _render(self, rect: rl.Rectangle):
+    if abs(self.rect.x) > gui_app.width * HORIZONTAL_RESET_RATIO:
+      self._scroller.scroll_panel.set_offset(0)
+
+    vertical_offset = self._scroller.scroll_panel.get_offset()
+    show_ball = abs(vertical_offset) < rect.height * CONFIDENCE_BALL_VISIBLE_RATIO
+    self.road_view.set_show_confidence_ball(show_ball)
+
+    super()._render(rect)

selfdrive/ui/sunnypilot/mici/layouts/onroad_info_panel.py

@@ -0,0 +1,324 @@
+"""
+Copyright (c) 2021-, Haibin Wen, sunnypilot, and a number of other contributors.
+
+This file is part of sunnypilot and is licensed under the MIT License.
+See the LICENSE.md file in the root directory for more details.
+"""
+
+import pyray as rl
+from dataclasses import dataclass
+from openpilot.common.constants import CV
+from openpilot.common.filter_simple import FirstOrderFilter
+from openpilot.selfdrive.ui.ui_state import ui_state
+from openpilot.system.ui.lib.application import gui_app, FontWeight
+from openpilot.system.ui.lib.multilang import tr
+from openpilot.system.ui.lib.text_measure import measure_text_cached
+from openpilot.system.ui.lib.application import MousePos
+from openpilot.system.ui.widgets import Widget
+from openpilot.selfdrive.ui.mici.onroad.alert_renderer import AlertRenderer
+from openpilot.selfdrive.ui.mici.onroad.augmented_road_view import BookmarkIcon
+
+METER_TO_KM = 0.001
+METER_TO_MILE = 0.000621371
+
+
+@dataclass(frozen=True)
+class OnroadInfoPanelColors:
+  white: rl.Color = rl.WHITE
+  black: rl.Color = rl.BLACK
+  red: rl.Color = rl.Color(255, 0, 0, 255)
+  green: rl.Color = rl.Color(0, 255, 0, 255)
+  grey: rl.Color = rl.Color(190, 195, 190, 255)
+  light_grey: rl.Color = rl.Color(200, 200, 200, 255)
+  dark_grey: rl.Color = rl.Color(100, 100, 100, 255)
+  bg_dark: rl.Color = rl.Color(0, 0, 0, 255)
+  card_bg: rl.Color = rl.Color(50, 50, 50, 200)
+  badge_bg: rl.Color = rl.Color(60, 60, 60, 255)
+
+
+COLORS = OnroadInfoPanelColors()
+
+
+class OnroadInfoPanel(Widget):
+  def __init__(self, bookmark_callback=None):
+    super().__init__()
+    self.speed_limit: float = 0.0
+    self.speed_limit_valid: bool = False
+    self.speed_limit_offset: float = 0.0
+    self.next_speed_limit: float = 0.0
+    self.next_speed_limit_distance: float = 0.0
+    self.road_name: str = ""
+    self.current_speed: float = 0.0
+    self.set_speed: float = 0.0
+    self.cruise_enabled: bool = False
+
+    self._sign_slide: float = 0.0
+
+    self._font_bold: rl.Font = gui_app.font(FontWeight.BOLD)
+    self._font_semi_bold: rl.Font = gui_app.font(FontWeight.SEMI_BOLD)
+    self._font_medium: rl.Font = gui_app.font(FontWeight.MEDIUM)
+
+    self._marquee_offset: float = 0.0
+    self._marquee_direction: int = 1
+    self._marquee_pause_timer: float = 0.0
+    self._marquee_speed: float = 40.0
+    self._marquee_pause_duration: float = 1.5
+
+    self._alert_renderer = AlertRenderer()
+    self._alert_alpha_filter = FirstOrderFilter(0, 0.05, 1 / gui_app.target_fps)
+
+    self._bookmark_icon = BookmarkIcon(bookmark_callback)
+
+  def is_swiping_left(self) -> bool:
+    return self._bookmark_icon.is_swiping_left()
+
+  def _handle_mouse_release(self, mouse_pos: MousePos) -> None:
+    # Mirror stock AugmentedRoadView: suppress click while bookmark gesture active
+    if not self._bookmark_icon.interacting():
+      super()._handle_mouse_release(mouse_pos)
+
+  def _update_state(self) -> None:
+    sm = ui_state.sm
+    speed_conv = CV.MS_TO_KPH if ui_state.is_metric else CV.MS_TO_MPH
+
+    if sm.valid["longitudinalPlanSP"]:
+      lp_sp = sm["longitudinalPlanSP"]
+      resolver = lp_sp.speedLimit.resolver
+      self.speed_limit = resolver.speedLimit * speed_conv
+      self.speed_limit_valid = resolver.speedLimitValid
+      self.speed_limit_offset = resolver.speedLimitOffset * speed_conv
+
+    if sm.valid["liveMapDataSP"]:
+      lmd = sm["liveMapDataSP"]
+      self.next_speed_limit = lmd.speedLimitAhead * speed_conv
+      self.next_speed_limit_distance = lmd.speedLimitAheadDistance
+      self.road_name = lmd.roadName
+
+    if sm.updated["carState"]:
+      self.current_speed = sm["carState"].vEgo * speed_conv
+
+    if sm.valid["carState"] and sm.valid["controlsState"]:
+      self.cruise_enabled = sm["carState"].cruiseState.enabled
+      v_cruise_cluster = sm["carState"].vCruiseCluster
+      set_speed_kph = sm["controlsState"].vCruiseDEPRECATED if v_cruise_cluster == 0.0 else v_cruise_cluster
+      self.set_speed = set_speed_kph * (METER_TO_MILE / METER_TO_KM) if not ui_state.is_metric else set_speed_kph
+
+  def _render(self, rect: rl.Rectangle) -> None:
+    self._update_state()
+
+    rl.draw_rectangle(int(rect.x), int(rect.y), int(rect.width), int(rect.height), COLORS.bg_dark)
+    margin = 20
+    mid_y = rect.y + rect.height / 2
+
+    left_x = rect.x + margin
+
+    if self.cruise_enabled:
+      unit = tr("MAX")
+      display_speed = self.set_speed
+    else:
+      unit = tr("km/h") if ui_state.is_metric else tr("MPH")
+      display_speed = self.current_speed
+
+    speed_val = str(round(display_speed))
+    if self.speed_limit_valid and display_speed > self.speed_limit:
+      speed_color = COLORS.red
+    else:
+      speed_color = COLORS.white
+
+    rl.draw_text_ex(self._font_semi_bold, unit, rl.Vector2(left_x, mid_y - 95), 38, 0, COLORS.grey)
+    rl.draw_text_ex(self._font_bold, speed_val, rl.Vector2(left_x, mid_y - 60), 110, 0, speed_color)
+
+    sign_width = 135
+    sign_height = 135 if ui_state.is_metric else 175
+
+    has_next = self.next_speed_limit > 0 and self.next_speed_limit != self.speed_limit
+    target_slide = 1.0 if has_next else 0.0
+    slide_speed = 3.0 * rl.get_frame_time()
+    if self._sign_slide < target_slide:
+      self._sign_slide = min(self._sign_slide + slide_speed, target_slide)
+    elif self._sign_slide > target_slide:
+      self._sign_slide = max(self._sign_slide - slide_speed, target_slide)
+
+    next_w = int(sign_width * 0.7)
+    next_h = int(sign_height * 0.7)
+    next_peek = int(next_w * 0.85) + 5
+    centered_x = rect.x + rect.width - sign_width - margin
+    shifted_x = rect.x + rect.width - sign_width - margin - next_peek
+    sign_x = centered_x + (shifted_x - centered_x) * self._sign_slide
+    sign_y = rect.y + (rect.height - sign_height) / 2
+
+    road_y = mid_y + 55
+    road_width = sign_x - left_x - margin
+    self._draw_road_name(left_x, road_y, road_width)
+
+    if has_next and self._sign_slide > 0.01:
+      next_val = str(round(self.next_speed_limit))
+      dist_str = self._format_distance(self.next_speed_limit_distance)
+      next_x = sign_x + sign_width - int(next_w * 0.15)
+      next_y = sign_y + (sign_height - next_h) / 2
+
+      next_speed_color = COLORS.black
+      if ui_state.is_metric:
+        self._draw_vienna_sign(next_x, next_y, next_w, next_h, next_val, next_speed_color, is_upcoming=True)
+      else:
+        self._draw_mutcd_sign(next_x, next_y, next_w, next_h, next_val, next_speed_color, is_upcoming=True)
+
+      dist_size = measure_text_cached(self._font_medium, dist_str, 24)
+      rl.draw_text_ex(self._font_medium, dist_str, rl.Vector2(next_x + next_w / 2 - dist_size.x / 2, next_y + next_h + 4), 24, 0, COLORS.grey)
+
+    self._draw_speed_limit_sign(sign_x, sign_y, sign_width, sign_height)
+
+    if self.speed_limit_offset != 0 and self.speed_limit_valid:
+      offset_val = str(abs(round(self.speed_limit_offset)))
+      badge_sz = 42
+      badge_x = sign_x + sign_width - badge_sz * 0.85
+      badge_y = sign_y - badge_sz * 0.25
+
+      if ui_state.is_metric:
+        badge_r = badge_sz / 2
+        badge_cx = badge_x + badge_r
+        badge_cy = badge_y + badge_r
+        rl.draw_circle(int(badge_cx), int(badge_cy), badge_r + 2, COLORS.dark_grey)
+        rl.draw_circle(int(badge_cx), int(badge_cy), badge_r, COLORS.badge_bg)
+        self._draw_text_centered(self._font_bold, offset_val, 24, rl.Vector2(badge_cx, badge_cy), COLORS.white)
+      else:
+        mutcd_badge_x = sign_x + sign_width - badge_sz * 0.65
+        mutcd_badge_y = sign_y - badge_sz * 0.50
+        badge_rect = rl.Rectangle(mutcd_badge_x, mutcd_badge_y, badge_sz, badge_sz)
+        rl.draw_rectangle_rounded(badge_rect, 0.25, 10, COLORS.badge_bg)
+        rl.draw_rectangle_rounded_lines_ex(badge_rect, 0.25, 10, 2, COLORS.dark_grey)
+        self._draw_text_centered(self._font_bold, offset_val, 24, rl.Vector2(mutcd_badge_x + badge_sz / 2, mutcd_badge_y + badge_sz / 2), COLORS.white)
+
+    # SCC
+    speed_size = measure_text_cached(self._font_bold, speed_val, 110)
+    scc_x = left_x + speed_size.x + 30
+    scc_y = mid_y - 50
+    self._draw_scc_icons(scc_x, scc_y)
+
+    self._bookmark_icon.render(rect)
+
+    if ui_state.started:
+      alert_obj, no_alert = self._alert_renderer.will_render()
+      self._alert_alpha_filter.update(0 if no_alert else 1)
+      alpha = self._alert_alpha_filter.x
+      if alpha > 0.01:
+        rl.draw_rectangle(int(rect.x), int(rect.y), int(rect.width), int(rect.height), rl.Color(0, 0, 0, int(150 * alpha)))
+      self._alert_renderer.render(rect)
+
+  def _draw_scc_icons(self, x: float, y: float) -> None:
+    sm = ui_state.sm
+    if not sm.valid["longitudinalPlanSP"]:
+      return
+    scc = sm["longitudinalPlanSP"].smartCruiseControl
+
+    box_w, box_h = 100, 36
+    gap = 6
+    drawn = 0
+
+    for label, active in [("SCC-V", scc.vision.active), ("SCC-M", scc.map.active)]:
+      if not active:
+        continue
+      bx = x
+      by = y + drawn * (box_h + gap)
+      rl.draw_rectangle_rounded(rl.Rectangle(bx, by, box_w, box_h), 0.3, 10, COLORS.green)
+      self._draw_text_centered(self._font_bold, label, 20, rl.Vector2(bx + box_w / 2, by + box_h / 2), COLORS.black)
+      drawn += 1
+
+  def _draw_speed_limit_sign(self, x: float, y: float, sign_width: float, sign_height: float) -> None:
+    speed_str = str(round(self.speed_limit)) if self.speed_limit_valid and self.speed_limit > 0 else "--"
+    speed_color = COLORS.black if not self.speed_limit_valid or self.current_speed <= self.speed_limit else COLORS.red
+
+    if ui_state.is_metric:
+      self._draw_vienna_sign(x, y, sign_width, sign_height, speed_str, speed_color, is_upcoming=False)
+    else:
+      self._draw_mutcd_sign(x, y, sign_width, sign_height, speed_str, speed_color, is_upcoming=False)
+
+  def _draw_road_name(self, x: float, y: float, width: float) -> None:
+    road_display = self.road_name if self.road_name else "--"
+    font_size = 30
+    road_size = measure_text_cached(self._font_semi_bold, road_display, font_size)
+    text_width = road_size.x
+
+    if text_width <= width:
+      self._marquee_offset = 0.0
+      self._marquee_direction = 1
+      self._marquee_pause_timer = 0.0
+      rl.draw_text_ex(self._font_semi_bold, road_display, rl.Vector2(x, y), font_size, 0, COLORS.white)
+    else:
+      overflow = text_width - width
+      dt = rl.get_frame_time()
+
+      if self._marquee_pause_timer > 0:
+        self._marquee_pause_timer -= dt
+      else:
+        self._marquee_offset += self._marquee_direction * self._marquee_speed * dt
+
+        if self._marquee_offset >= overflow:
+          self._marquee_offset = overflow
+          self._marquee_direction = -1
+          self._marquee_pause_timer = self._marquee_pause_duration
+        elif self._marquee_offset <= 0:
+          self._marquee_offset = 0
+          self._marquee_direction = 1
+          self._marquee_pause_timer = self._marquee_pause_duration
+
+      rl.begin_scissor_mode(int(x), int(y), int(width), int(road_size.y + 4))
+      text_pos = rl.Vector2(x - self._marquee_offset, y)
+      rl.draw_text_ex(self._font_semi_bold, road_display, text_pos, font_size, 0, COLORS.white)
+      rl.end_scissor_mode()
+
+  def _draw_vienna_sign(self, x: float, y: float, width: float, height: float, speed_str: str, speed_color: rl.Color, is_upcoming: bool = False) -> None:
+    center = rl.Vector2(x + width / 2, y + height / 2)
+    outer_radius = min(width, height) / 2
+
+    rl.draw_circle_v(center, outer_radius, COLORS.white)
+    ring_width = outer_radius * 0.18
+    rl.draw_ring(center, outer_radius - ring_width, outer_radius, 0, 360, 36, COLORS.red)
+
+    font_size = outer_radius * (0.7 if len(speed_str) >= 3 else 0.9)
+    text_size = measure_text_cached(self._font_bold, speed_str, int(font_size))
+    text_pos = rl.Vector2(center.x - text_size.x / 2, center.y - text_size.y / 2)
+    rl.draw_text_ex(self._font_bold, speed_str, text_pos, font_size, 0, speed_color)
+
+  def _draw_mutcd_sign(self, x: float, y: float, width: float, height: float, speed_str: str, speed_color: rl.Color, is_upcoming: bool = False) -> None:
+    sign_rect = rl.Rectangle(x, y, width, height)
+    rl.draw_rectangle_rounded(sign_rect, 0.35, 10, COLORS.white)
+
+    inset = max(4, width * 0.05)
+    inner_rect = rl.Rectangle(x + inset, y + inset, width - inset * 2, height - inset * 2)
+    outer_radius = 0.35 * width / 2.0
+    inner_radius = outer_radius - inset
+    inner_roundness = inner_radius / (inner_rect.width / 2.0)
+    rl.draw_rectangle_rounded_lines_ex(inner_rect, inner_roundness, 10, 3, COLORS.black)
+
+    mid_x = x + width / 2
+    label_size = max(18, int(width * 0.26))
+    if is_upcoming:
+      self._draw_text_centered(self._font_bold, tr("AHEAD"), label_size, rl.Vector2(mid_x, y + height * 0.27), COLORS.black)
+    else:
+      self._draw_text_centered(self._font_bold, tr("SPEED"), label_size, rl.Vector2(mid_x, y + height * 0.20), COLORS.black)
+      self._draw_text_centered(self._font_bold, tr("LIMIT"), label_size, rl.Vector2(mid_x, y + height * 0.40), COLORS.black)
+
+    speed_font_size = int(width * 0.52) if len(speed_str) >= 3 else int(width * 0.62)
+    self._draw_text_centered(self._font_bold, speed_str, speed_font_size, rl.Vector2(mid_x, y + height * 0.72), speed_color)
+
+  def _draw_text_centered(self, font, text, size, pos_center, color):
+    sz = measure_text_cached(font, text, size)
+    rl.draw_text_ex(font, text, rl.Vector2(pos_center.x - sz.x / 2, pos_center.y - sz.y / 2), size, 0, color)
+
+  def _format_distance(self, distance: float) -> str:
+    if ui_state.is_metric:
+      if distance < 50:
+        return tr("Near")
+      if distance >= 1000:
+        return f"{distance * METER_TO_KM:.1f}" + tr("km")
+      if distance < 200:
+        rounded = max(10, int(distance / 10) * 10)
+      else:
+        rounded = int(distance / 100) * 100
+      return str(rounded) + tr("m")
+    else:
+      distance_mi = distance * METER_TO_MILE
+      if distance_mi < 0.1:
+        return tr("Near")
+      return f"{distance_mi:.1f}" + tr("mi")

selfdrive/ui/sunnypilot/mici/onroad/augmented_road_view.py

@@ -0,0 +1,30 @@
+"""
+Copyright (c) 2021-, Haibin Wen, sunnypilot, and a number of other contributors.
+
+This file is part of sunnypilot and is licensed under the MIT License.
+See the LICENSE.md file in the root directory for more details.
+"""
+
+import pyray as rl
+from openpilot.selfdrive.ui.mici.onroad.augmented_road_view import AugmentedRoadView
+
+
+class _SuppressedConfidenceBall:
+  def render(self, *_):
+    pass
+
+
+class AugmentedRoadViewSP(AugmentedRoadView):
+  def __init__(self, **kwargs):
+    super().__init__(**kwargs)
+    self._show_confidence_ball: bool = True
+    self._real_confidence_ball = self._confidence_ball
+    self._confidence_ball = _SuppressedConfidenceBall()
+
+  def set_show_confidence_ball(self, show: bool) -> None:
+    self._show_confidence_ball = show
+
+  def _render(self, rect: rl.Rectangle) -> None:
+    super()._render(rect)
+    if self._show_confidence_ball:
+      self._real_confidence_ball.render(self.rect)

selfdrive/ui/sunnypilot/mici/widgets/scroll_panel_sp.py

@@ -0,0 +1,34 @@
+"""
+Copyright (c) 2021-, Haibin Wen, sunnypilot, and a number of other contributors.
+
+This file is part of sunnypilot and is licensed under the MIT License.
+See the LICENSE.md file in the root directory for more details.
+"""
+
+import pyray as rl
+from openpilot.system.ui.lib.application import MouseEvent
+from openpilot.system.ui.lib.scroll_panel2 import GuiScrollPanel2, ScrollState
+
+
+class GuiScrollPanel2SP(GuiScrollPanel2):
+  """Reject orthogonal-dominant drags so nested scrollers (outer horizontal +
+  inner vertical) don't both engage on a slightly diagonal swipe.
+
+  Implemented as a post-super state rollback rather than reimplementing the
+  PRESSED state machine — keeps stock behaviour authoritative."""
+
+  def _handle_mouse_event(self, mouse_event: MouseEvent, bounds: rl.Rectangle, bounds_size: float,
+                          content_size: float) -> None:
+    pre_state = self._state
+    super()._handle_mouse_event(mouse_event, bounds, bounds_size, content_size)
+
+    if self._state == ScrollState.MANUAL_SCROLL and pre_state == ScrollState.PRESSED and \
+       self._initial_click_event is not None:
+      diff_x = abs(mouse_event.pos.x - self._initial_click_event.pos.x)
+      diff_y = abs(mouse_event.pos.y - self._initial_click_event.pos.y)
+      along = diff_x if self._horizontal else diff_y
+      anti = diff_y if self._horizontal else diff_x
+      if anti > along:
+        self._state = ScrollState.STEADY
+        self._velocity = 0.0
+        self._velocity_buffer.clear()

selfdrive/ui/sunnypilot/mici/widgets/scroller_sp.py

@@ -0,0 +1,16 @@
+"""
+Copyright (c) 2021-, Haibin Wen, sunnypilot, and a number of other contributors.
+
+This file is part of sunnypilot and is licensed under the MIT License.
+See the LICENSE.md file in the root directory for more details.
+"""
+
+from openpilot.system.ui.widgets.scroller import Scroller
+from openpilot.selfdrive.ui.sunnypilot.mici.widgets.scroll_panel_sp import GuiScrollPanel2SP
+
+
+class ScrollerSP(Scroller):
+  def __init__(self, **kwargs):
+    super().__init__(**kwargs)
+    inner = self._scroller
+    inner.scroll_panel = GuiScrollPanel2SP(inner._horizontal, handle_out_of_bounds=not inner._snap_items)

selfdrive/ui/ui.py

@@ -10,6 +10,9 @@ from openpilot.selfdrive.ui.layouts.main import MainLayout
 from openpilot.selfdrive.ui.mici.layouts.main import MiciMainLayout
 from openpilot.selfdrive.ui.ui_state import ui_state
 
+if gui_app.sunnypilot_ui():
+  from openpilot.selfdrive.ui.sunnypilot.mici.layouts.main import MiciMainLayoutSP as MiciMainLayout
+
 BIG_UI = gui_app.big_ui()
 
 

sunnypilot/modeld_v2/modeld.py

@@ -40,6 +40,7 @@ from openpilot.sunnypilot.modeld_v2.camera_offset_helper import CameraOffsetHelp
 from openpilot.sunnypilot.livedelay.helpers import get_lat_delay
 from openpilot.sunnypilot.modeld_v2.modeld_base import ModelStateBase
 from openpilot.sunnypilot.models.helpers import get_active_bundle
+from openpilot.sunnypilot.selfdrive.controls.lib.relc import RoadEdgeLaneChangeController
 
 PROCESS_NAME = "selfdrive.modeld.modeld_tinygrad"
 
@@ -329,6 +330,7 @@ def main(demo=False):
   prev_action = log.ModelDataV2.Action()
 
   DH = DesireHelper()
+  RELC = RoadEdgeLaneChangeController(DH)
   meta_constants = load_meta_constants()
 
   while True:
@@ -433,7 +435,10 @@ def main(demo=False):
       l_lane_change_prob = desire_state[log.Desire.laneChangeLeft]
       r_lane_change_prob = desire_state[log.Desire.laneChangeRight]
       lane_change_prob = l_lane_change_prob + r_lane_change_prob
-      DH.update(sm['carState'], sm['carControl'].latActive, lane_change_prob)
+      RELC.update(modelv2_send.modelV2.roadEdgeStds, modelv2_send.modelV2.laneLineProbs, v_ego)
+      mdv2sp_send.modelDataV2SP.leftLaneChangeEdgeBlock = RELC.left_edge_detected
+      mdv2sp_send.modelDataV2SP.rightLaneChangeEdgeBlock = RELC.right_edge_detected
+      DH.update(sm['carState'], sm['carControl'].latActive, lane_change_prob, RELC.left_edge_detected, RELC.right_edge_detected)
       modelv2_send.modelV2.meta.laneChangeState = DH.lane_change_state
       modelv2_send.modelV2.meta.laneChangeDirection = DH.lane_change_direction
       mdv2sp_send.modelDataV2SP.laneTurnDirection = DH.lane_turn_direction

sunnypilot/selfdrive/controls/lib/accel_personality/__init__.py

@@ -0,0 +1,8 @@
+from openpilot.sunnypilot.selfdrive.controls.lib.accel_personality.accel_controller import (
+  AccelController,
+  AccelControllerResult,
+  AccelControllerState,
+)
+from openpilot.sunnypilot.selfdrive.controls.lib.accel_personality.constants import AccelProfile
+
+__all__ = ["AccelController", "AccelControllerResult", "AccelControllerState", "AccelProfile"]

sunnypilot/selfdrive/controls/lib/accel_personality/accel_controller.py

@@ -0,0 +1,619 @@
+#!/usr/bin/env python3
+from collections import deque
+from dataclasses import dataclass, field
+from enum import IntEnum
+import math
+
+import numpy as np
+
+from cereal import log
+from opendbc.car.interfaces import ACCEL_MIN, ACCEL_MAX
+from openpilot.common.realtime import DT_MDL
+from openpilot.selfdrive.controls.lib.longitudinal_mpc_lib.long_mpc import LongitudinalMpc, STOP_DISTANCE, T_IDXS, get_T_FOLLOW, get_stopped_equivalence_factor
+from openpilot.selfdrive.controls.radard import _LEAD_ACCEL_TAU
+from openpilot.sunnypilot.selfdrive.controls.lib.accel_personality.constants import (
+  ACCEL_PROFILE_MAX_BP, ACCEL_PROFILE_MAX_V, APPROACH_CLOSING_SPEED, APPROACH_LEAD_DECEL, APPROACH_LEAD_SPEED_MARGIN, APPROACH_MIN_SPEED,
+  BRAKE_CAP_MARGIN, CAP_FILTER_FRAMES, CAP_RELAX_JERK, CAP_TIGHTEN_JERK, COAST_MATCH_CLOSING_SPEED, COAST_MATCH_USABLE_GAP,
+  DROPOUT_ACTION_ACCEL_MARGIN, HORIZON_DOWN_JERK, HORIZON_HOLD_TIME, HORIZON_SPEED_BUDGET, HORIZON_UP_JERK, MAX_LEAD_ACCEL_TAU,
+  MIN_LEAD_SPEED, POSITIVE_MPC_HEADROOM, PROFILE_CONFIGS, PROFILE_TRANSITION_JERK, RADAR_STALE_TIMEOUT, RELIEF_CAP_MARGIN,
+  RELIEF_CONFIRM_FRAMES, RELIEF_LEAD_SPEED_STEP, RELIEF_MPC_JERK, REQUIRED_DECEL_MARGIN, ROUTINE_DECEL_MAX, STOP_HOLD_EGO_SPEED,
+  SHALLOW_BRAKE_BOUND, SHALLOW_BRAKE_RELIEF_TIME, STOP_GAP_RESERVE, STOP_GAP_RESERVE_DECEL_BP, STOP_GAP_RESERVE_LEAD_SPEED,
+  STOP_HOLD_CREEP_ABORT_FRAMES, STOP_HOLD_CREEP_DISTANCE, STOP_HOLD_CREEP_SPEED,
+  STOP_HOLD_EXIT_FRAMES, STOP_HOLD_EXIT_SPEED,
+  STOPPED_LEAD_SPEED, URGENT_CLOSING_SPEED, URGENT_RELEASE_ACCEL, URGENT_REQUIRED_DECEL, URGENT_TTC, URGENT_TTC_MIN_CLOSING,
+  VEGO_NOISE_TOLERANCE, AccelProfile,
+)
+
+
+class AccelControllerState(IntEnum):
+  inactive = 0
+  free = 1
+  restrict = 2
+  hold = 3
+  release = 4
+  stopHold = 5
+
+
+@dataclass(frozen=True)
+class EnergyEnvelope:
+  cap: float = math.inf
+  selected_lead: int = -1
+  selected_lead_speed: float = math.inf
+  selected_lead_decel: float = 0.0
+  departure_lead_index: int = -1
+  departure_lead_speed: float = math.inf
+  departure_cap: float = math.inf
+  departure_lead_speeds: tuple[float, float] = (math.inf, math.inf)
+  departure_lead_separations: tuple[float, float] = (-math.inf, -math.inf)
+  usable_gap: float = math.inf
+  safety_usable_gap: float = math.inf
+  closing_speed: float = 0.0
+  required_decel: float = 0.0
+  has_nearly_stopped_lead: bool = False
+  lead_status: bool = False
+
+
+@dataclass(frozen=True)
+class AccelControllerResult:
+  target_speed: float
+  enabled: bool
+  active: bool
+  shadow_active: bool
+  launching: bool
+  stock_mode: bool
+  profile: AccelProfile
+  profile_accel_max: float
+  positive_accel_max: float
+  effective_accel_max: float
+  mpc_accel_max: tuple[float, ...] | None
+  state: AccelControllerState
+  shadow_state: AccelControllerState
+  base_speed: float
+  raw_energy_cap: float
+  live_filtered_cap: float
+  shadow_filtered_cap: float
+  selected_lead: int
+  selected_lead_speed: float
+  usable_gap: float
+  closing_speed: float
+  required_decel: float
+
+
+@dataclass
+class _ControllerPath:
+  cap_samples: deque[float] = field(default_factory=lambda: deque([math.inf] * CAP_FILTER_FRAMES, maxlen=CAP_FILTER_FRAMES))
+  required_samples: deque[float] = field(default_factory=lambda: deque(maxlen=CAP_FILTER_FRAMES))
+  lead_decel_samples: deque[float] = field(default_factory=lambda: deque(maxlen=CAP_FILTER_FRAMES))
+  departure_samples: tuple[deque[float], deque[float]] = field(
+    default_factory=lambda: (deque(maxlen=CAP_FILTER_FRAMES), deque(maxlen=CAP_FILTER_FRAMES)),
+  )
+  departure_references: list[float | None] = field(default_factory=lambda: [None, None])
+  bound: float | None = None
+  state: AccelControllerState = AccelControllerState.inactive
+  relief_frames: int = 0
+  bound_relief_frames: int = 0
+  bound_relief_required_frames: int = 0
+  departure_frames: int = 0
+  creep_abort_frames: int = 0
+  stale_frames: int = 0
+  urgent: bool = False
+  urgent_severe: bool = False
+  urgent_safe_frames: int = 0
+  departing_from_stop: bool = False
+  previous_lead_speed: float | None = None
+  lead_speed_relief: bool = False
+
+  @property
+  def filtered_cap(self) -> float:
+    return sorted(self.cap_samples)[CAP_FILTER_FRAMES // 2]
+
+  @property
+  def robust_required_decel(self) -> float:
+    return float(np.median(self.required_samples)) if self.required_samples else 0.0
+
+  @property
+  def robust_lead_decel(self) -> float:
+    return float(np.median(self.lead_decel_samples)) if self.lead_decel_samples else 0.0
+
+  def robust_departure_separation(self, lead_index: int) -> float:
+    samples = self.departure_samples[lead_index]
+    return float(np.median(samples)) if samples else -math.inf
+
+  def reset(self) -> None:
+    self.cap_samples = deque([math.inf] * CAP_FILTER_FRAMES, maxlen=CAP_FILTER_FRAMES)
+    self.required_samples.clear()
+    self.lead_decel_samples.clear()
+    self.departure_samples = (deque(maxlen=CAP_FILTER_FRAMES), deque(maxlen=CAP_FILTER_FRAMES))
+    self.departure_references = [None, None]
+    self.bound = None
+    self.state = AccelControllerState.inactive
+    self.relief_frames = 0
+    self.bound_relief_frames = 0
+    self.bound_relief_required_frames = 0
+    self.departure_frames = 0
+    self.creep_abort_frames = 0
+    self.stale_frames = 0
+    self.urgent = False
+    self.urgent_severe = False
+    self.urgent_safe_frames = 0
+    self.departing_from_stop = False
+    self.previous_lead_speed = None
+    self.lead_speed_relief = False
+
+
+class AccelController:
+  def __init__(self, CP, dt: float = DT_MDL):
+    if not math.isfinite(dt) or dt <= 0.0:
+      raise ValueError("dt must be finite and positive")
+
+    self.CP = CP
+    self.dt = dt
+    self.radar_stale_frames = max(1, math.ceil(RADAR_STALE_TIMEOUT / dt))
+    self.shallow_brake_relief_frames = max(RELIEF_CONFIRM_FRAMES, math.ceil(SHALLOW_BRAKE_RELIEF_TIME / dt))
+    self.live = _ControllerPath()
+    self.shadow = _ControllerPath()
+
+  @staticmethod
+  def _profile(profile: int | AccelProfile) -> AccelProfile:
+    try:
+      return AccelProfile(profile)
+    except (TypeError, ValueError):
+      return AccelProfile.normal
+
+  @classmethod
+  def get_profile_accel_max(cls, profile: int | AccelProfile, v_ego: float) -> float:
+    if not math.isfinite(v_ego):
+      return math.nan
+    selected_profile = cls._profile(profile)
+    return float(np.interp(max(v_ego, 0.0), ACCEL_PROFILE_MAX_BP, ACCEL_PROFILE_MAX_V[selected_profile]))
+
+  def _delay(self) -> float:
+    try:
+      return float(self.CP.longitudinalActuatorDelay) + DT_MDL
+    except (AttributeError, OverflowError, TypeError, ValueError):
+      return math.nan
+
+  @staticmethod
+  def _project_ego(v_ego: float, a_ego: float, delay: float) -> tuple[float, float]:
+    if a_ego < 0.0:
+      stop_time = -v_ego / a_ego if v_ego > 0.0 else 0.0
+      if stop_time <= delay:
+        distance = -v_ego**2 / (2.0 * a_ego) if v_ego > 0.0 else 0.0
+        return distance, 0.0
+    return max(v_ego * delay + 0.5 * a_ego * delay**2, 0.0), max(v_ego + a_ego * delay, 0.0)
+
+  @staticmethod
+  def _lead_values(lead) -> tuple[float, float, float, float] | None:
+    try:
+      if not lead.status:
+        return None
+      d_rel = float(lead.dRel)
+      v_lead = float(lead.vLeadK)
+    except (AttributeError, OverflowError, TypeError, ValueError):
+      return None
+    if not math.isfinite(d_rel) or d_rel < 0.0 or not math.isfinite(v_lead) or v_lead < MIN_LEAD_SPEED:
+      return None
+
+    try:
+      a_lead = float(lead.aLeadK)
+    except (AttributeError, OverflowError, TypeError, ValueError):
+      a_lead = 0.0
+    if not math.isfinite(a_lead):
+      a_lead = 0.0
+
+    try:
+      a_lead_tau = float(lead.aLeadTau)
+    except (AttributeError, OverflowError, TypeError, ValueError):
+      a_lead_tau = _LEAD_ACCEL_TAU
+    if not math.isfinite(a_lead_tau) or not 0.0 < a_lead_tau <= MAX_LEAD_ACCEL_TAU:
+      a_lead_tau = _LEAD_ACCEL_TAU
+    return d_rel, max(v_lead, 0.0), float(np.clip(a_lead, -10.0, 5.0)), a_lead_tau
+
+  def calculate_energy_envelope(self, radar_state, v_ego: float, a_ego: float, profile: int | AccelProfile,
+                                follow_personality=log.LongitudinalPersonality.standard) -> EnergyEnvelope:
+    delay = self._delay()
+    if not all(math.isfinite(value) for value in (v_ego, a_ego, delay)) or v_ego < 0.0 or delay < 0.0:
+      return EnergyEnvelope()
+
+    try:
+      leads = (radar_state.leadOne, radar_state.leadTwo)
+      lead_status = any(bool(lead.status) for lead in leads)
+    except (AttributeError, TypeError, ValueError):
+      return EnergyEnvelope()
+
+    try:
+      t_follow = get_T_FOLLOW(follow_personality)
+    except (NotImplementedError, TypeError, ValueError):
+      t_follow = get_T_FOLLOW(log.LongitudinalPersonality.standard)
+    if not math.isfinite(t_follow) or t_follow < 0.0:
+      return EnergyEnvelope(lead_status=lead_status)
+
+    x_ego, v_ego_delay = self._project_ego(v_ego, a_ego, delay)
+    comfort_decel = PROFILE_CONFIGS[self._profile(profile)].comfort_decel
+    candidates: list[EnergyEnvelope] = []
+    departure_candidates: list[tuple[float, int]] = []
+    departure_speeds = [math.inf, math.inf]
+    departure_separations = [-math.inf, -math.inf]
+    departure_caps = [math.inf, math.inf]
+    for lead_index, lead in enumerate(leads):
+      values = self._lead_values(lead)
+      if values is None:
+        continue
+      try:
+        d_rel, v_lead, a_lead, a_lead_tau = values
+        lead_xv = LongitudinalMpc.extrapolate_lead(d_rel, v_lead, a_lead, a_lead_tau)
+        x_lead = float(np.interp(delay, T_IDXS, lead_xv[:, 0]))
+        v_lead_delay = float(np.interp(delay, T_IDXS, lead_xv[:, 1]))
+        safety_usable_gap = max(x_lead - x_ego - STOP_DISTANCE - t_follow * v_lead_delay, 0.0)
+        closing_speed = max(v_ego_delay - v_lead_delay, 0.0)
+        required_decel = (0.0 if closing_speed == 0.0 else math.inf if safety_usable_gap == 0.0
+                          else closing_speed**2 / (2.0 * safety_usable_gap))
+        reserve_speed = float(np.interp(v_lead_delay, (0.0, STOP_GAP_RESERVE_LEAD_SPEED), (STOP_GAP_RESERVE, 0.0)))
+        reserve_scale = float(np.interp(required_decel, STOP_GAP_RESERVE_DECEL_BP, (1.0, 0.0)))
+        usable_gap = max(safety_usable_gap - reserve_speed * reserve_scale, 0.0)
+        cap = v_lead_delay + math.sqrt(2.0 * comfort_decel * usable_gap)
+        departure_cap = v_lead_delay + math.sqrt(2.0 * comfort_decel * safety_usable_gap)
+        projected_separation = x_lead - x_ego
+        departure_distance = x_lead + float(get_stopped_equivalence_factor(v_lead_delay))
+      except (FloatingPointError, OverflowError, TypeError, ValueError):
+        continue
+      finite_values = (x_lead, v_lead_delay, usable_gap, safety_usable_gap, closing_speed, cap, departure_cap, departure_distance)
+      if not all(math.isfinite(value) and value >= 0.0 for value in finite_values) or math.isnan(required_decel) or required_decel < 0.0:
+        continue
+      if not math.isfinite(projected_separation):
+        continue
+      candidates.append(EnergyEnvelope(cap=cap, selected_lead=lead_index, selected_lead_speed=v_lead_delay,
+                                       selected_lead_decel=max(-a_lead, 0.0), usable_gap=usable_gap,
+                                       safety_usable_gap=safety_usable_gap, closing_speed=closing_speed,
+                                       required_decel=required_decel, lead_status=lead_status))
+      departure_candidates.append((departure_distance, lead_index))
+      departure_speeds[lead_index] = v_lead_delay
+      departure_separations[lead_index] = projected_separation
+      departure_caps[lead_index] = departure_cap
+
+    if not candidates:
+      return EnergyEnvelope(lead_status=lead_status)
+    selected = min(candidates, key=lambda candidate: candidate.cap)
+    departure_lead_index = min(departure_candidates, key=lambda candidate: candidate[0])[1]
+    departure_lead_speed = departure_speeds[departure_lead_index]
+    return EnergyEnvelope(
+      cap=selected.cap, selected_lead=selected.selected_lead, selected_lead_speed=selected.selected_lead_speed,
+      selected_lead_decel=selected.selected_lead_decel, departure_lead_index=departure_lead_index,
+      departure_lead_speed=departure_lead_speed, departure_cap=departure_caps[departure_lead_index],
+      departure_lead_speeds=tuple(departure_speeds), departure_lead_separations=tuple(departure_separations),
+      usable_gap=selected.usable_gap, safety_usable_gap=selected.safety_usable_gap, closing_speed=selected.closing_speed,
+      required_decel=selected.required_decel, has_nearly_stopped_lead=departure_lead_speed < STOPPED_LEAD_SPEED,
+      lead_status=lead_status,
+    )
+
+  @staticmethod
+  def _move(value: float, target: float, rate: float, dt: float) -> float:
+    return float(np.clip(target, value - rate * dt, value + rate * dt))
+
+  @staticmethod
+  def _ttc(envelope: EnergyEnvelope) -> float:
+    return envelope.safety_usable_gap / envelope.closing_speed if envelope.closing_speed > 0.0 else math.inf
+
+  def _update_samples(self, path: _ControllerPath, envelope: EnergyEnvelope) -> None:
+    has_lead = envelope.selected_lead >= 0
+    path.lead_speed_relief = (has_lead and path.previous_lead_speed is not None
+                              and envelope.selected_lead_speed > path.previous_lead_speed + RELIEF_LEAD_SPEED_STEP)
+    path.previous_lead_speed = envelope.selected_lead_speed if has_lead else None
+    path.cap_samples.append(envelope.cap if has_lead else math.inf)
+    for lead_index, separation in enumerate(envelope.departure_lead_separations):
+      if math.isfinite(separation):
+        path.departure_samples[lead_index].append(separation)
+    if has_lead:
+      if math.isfinite(envelope.required_decel):
+        path.required_samples.append(envelope.required_decel)
+      if math.isfinite(envelope.selected_lead_decel):
+        path.lead_decel_samples.append(envelope.selected_lead_decel)
+    else:
+      path.required_samples.append(0.0)
+      path.lead_decel_samples.append(0.0)
+
+  @staticmethod
+  def _reset_departure_tracking(path: _ControllerPath, envelope: EnergyEnvelope) -> None:
+    path.departure_samples = (deque(maxlen=CAP_FILTER_FRAMES), deque(maxlen=CAP_FILTER_FRAMES))
+    path.departure_references = [None, None]
+    for lead_index, separation in enumerate(envelope.departure_lead_separations):
+      if math.isfinite(separation):
+        path.departure_samples[lead_index].append(separation)
+        path.departure_references[lead_index] = separation
+    path.departure_frames = 0
+    path.creep_abort_frames = 0
+
+  @staticmethod
+  def _clear_bound_relief(path: _ControllerPath) -> None:
+    path.bound_relief_frames = 0
+    path.bound_relief_required_frames = 0
+
+  @staticmethod
+  def _creep_departure(path: _ControllerPath, envelope: EnergyEnvelope) -> bool:
+    lead_index = envelope.departure_lead_index
+    if lead_index < 0 or envelope.departure_lead_speed <= STOP_HOLD_CREEP_SPEED:
+      return False
+
+    separation = path.robust_departure_separation(lead_index)
+    reference = path.departure_references[lead_index]
+    return reference is not None and separation - reference >= STOP_HOLD_CREEP_DISTANCE
+
+  def _update_path(self, path: _ControllerPath, envelope: EnergyEnvelope, base_speed: float, v_ego: float, action_accel: float,
+                   positive_accel_max: float, profile: AccelProfile, previous_should_stop: bool) -> bool:
+    self._update_samples(path, envelope)
+    has_lead = envelope.selected_lead >= 0
+    filtered_cap = path.filtered_cap
+    robust_required = path.robust_required_decel
+    robust_lead_decel = path.robust_lead_decel
+    ttc = self._ttc(envelope)
+    moving_away = (has_lead and not envelope.has_nearly_stopped_lead
+                   and envelope.selected_lead_speed > v_ego + APPROACH_CLOSING_SPEED
+                   and envelope.cap > v_ego + RELIEF_CAP_MARGIN)
+
+    if path.departing_from_stop:
+      if v_ego >= STOP_HOLD_EGO_SPEED:
+        path.departing_from_stop = False
+        path.creep_abort_frames = 0
+      elif envelope.lead_status and (not has_lead or envelope.departure_lead_speed <= STOP_HOLD_CREEP_SPEED):
+        path.creep_abort_frames += 1
+        if path.creep_abort_frames >= STOP_HOLD_CREEP_ABORT_FRAMES:
+          path.departing_from_stop = False
+          path.creep_abort_frames = 0
+      else:
+        path.creep_abort_frames = 0
+
+    stop_hold = (v_ego < STOP_HOLD_EGO_SPEED and not path.departing_from_stop
+                 and (previous_should_stop or (envelope.lead_status and not has_lead)
+                      or (has_lead and (envelope.has_nearly_stopped_lead or envelope.cap < 0.50))))
+
+    if path.state == AccelControllerState.stopHold:
+      self._clear_bound_relief(path)
+      for lead_index in range(len(path.departure_references)):
+        separation = path.robust_departure_separation(lead_index)
+        if math.isfinite(separation) and path.departure_references[lead_index] is None:
+          path.departure_references[lead_index] = separation
+      departed = (not envelope.lead_status
+                  or (has_lead and envelope.departure_lead_speed > STOP_HOLD_EXIT_SPEED
+                      and envelope.departure_cap > STOP_HOLD_EXIT_SPEED)
+                  or self._creep_departure(path, envelope))
+      path.departure_frames = path.departure_frames + 1 if departed else 0
+      path.bound = 0.0
+      if path.departure_frames < STOP_HOLD_EXIT_FRAMES:
+        return False
+      path.state = AccelControllerState.free
+      path.bound = positive_accel_max
+      path.departure_frames = 0
+      path.departing_from_stop = True
+      return False
+
+    if stop_hold:
+      path.state = AccelControllerState.stopHold
+      path.bound = 0.0
+      path.relief_frames = 0
+      self._clear_bound_relief(path)
+      path.departure_frames = 0
+      path.urgent = False
+      path.urgent_severe = False
+      path.urgent_safe_frames = 0
+      path.departing_from_stop = False
+      self._reset_departure_tracking(path, envelope)
+      return False
+
+    urgent_closing = envelope.closing_speed > URGENT_TTC_MIN_CLOSING
+    raw_urgent = (has_lead and v_ego >= STOP_HOLD_EGO_SPEED
+                  and (envelope.closing_speed >= URGENT_CLOSING_SPEED
+                       or (urgent_closing and envelope.required_decel >= URGENT_REQUIRED_DECEL)
+                       or (urgent_closing and ttc <= URGENT_TTC)))
+    if raw_urgent:
+      path.urgent = True
+      path.urgent_severe |= envelope.closing_speed >= URGENT_CLOSING_SPEED or envelope.required_decel >= URGENT_REQUIRED_DECEL
+      path.urgent_safe_frames = 0
+      path.bound = None
+      path.state = AccelControllerState.hold
+      path.relief_frames = 0
+      self._clear_bound_relief(path)
+      return True
+
+    if path.urgent:
+      matched = has_lead and envelope.closing_speed <= APPROACH_CLOSING_SPEED and robust_lead_decel <= 0.05
+      urgent_safe = (not has_lead or moving_away or matched) and (not path.urgent_severe or action_accel >= URGENT_RELEASE_ACCEL)
+      path.urgent_safe_frames = path.urgent_safe_frames + 1 if urgent_safe else 0
+      if path.urgent_safe_frames < RELIEF_CONFIRM_FRAMES:
+        path.bound = None
+        path.state = AccelControllerState.hold
+        self._clear_bound_relief(path)
+        return True
+      path.urgent = False
+      path.urgent_severe = False
+      path.urgent_safe_frames = 0
+      if not has_lead or moving_away:
+        path.state = AccelControllerState.free
+        path.bound = min(action_accel, 0.0)
+      else:
+        path.state = AccelControllerState.hold
+        path.bound = 0.0
+
+    if path.state == AccelControllerState.inactive and has_lead and not math.isfinite(filtered_cap):
+      path.bound = min(action_accel, 0.0)
+      self._clear_bound_relief(path)
+      return False
+
+    dropout_guard = (not has_lead and math.isfinite(filtered_cap)
+                     and path.state in (AccelControllerState.restrict, AccelControllerState.hold) and path.bound is not None)
+    if dropout_guard:
+      path.bound = min(path.bound, action_accel + DROPOUT_ACTION_ACCEL_MARGIN)
+
+    profile_config = PROFILE_CONFIGS[profile]
+    lead_demand = (envelope.closing_speed > APPROACH_CLOSING_SPEED
+                   or (robust_lead_decel > APPROACH_LEAD_DECEL
+                       and envelope.selected_lead_speed < v_ego + APPROACH_LEAD_SPEED_MARGIN))
+    braking_zone = filtered_cap < v_ego + BRAKE_CAP_MARGIN
+    anticipation = filtered_cap < base_speed - profile_config.anticipation_margin
+    approach = (has_lead and (v_ego > APPROACH_MIN_SPEED or path.state == AccelControllerState.restrict)
+                and lead_demand and (braking_zone or anticipation))
+    retaining_lead = path.state in (AccelControllerState.restrict, AccelControllerState.hold) and has_lead and not moving_away
+    if approach or retaining_lead:
+      entering = path.state not in (AccelControllerState.restrict, AccelControllerState.hold)
+      if path.bound is None or entering:
+        path.bound = action_accel
+      matched = envelope.closing_speed <= APPROACH_CLOSING_SPEED and robust_lead_decel <= 0.05
+      coast_cap = envelope.selected_lead_speed + math.sqrt(2.0 * profile_config.comfort_decel * COAST_MATCH_USABLE_GAP)
+      coast_to_match = (robust_lead_decel <= 0.05 and envelope.closing_speed <= COAST_MATCH_CLOSING_SPEED
+                        and filtered_cap > coast_cap)
+      if matched or coast_to_match:
+        target_decel = 0.0
+      elif braking_zone:
+        target_decel = min(max(robust_required + REQUIRED_DECEL_MARGIN, robust_lead_decel), ROUTINE_DECEL_MAX)
+      else:
+        target_decel = profile_config.glide_decel
+      target = -target_decel
+      bound_relief = has_lead and path.bound < 0.0 and target > path.bound + 1e-9
+      if bound_relief and path.bound_relief_frames == 0:
+        path.bound_relief_required_frames = (self.shallow_brake_relief_frames
+                                             if path.bound >= SHALLOW_BRAKE_BOUND else RELIEF_CONFIRM_FRAMES)
+      path.bound_relief_frames = path.bound_relief_frames + 1 if bound_relief else 0
+      if not bound_relief:
+        self._clear_bound_relief(path)
+      if bound_relief and path.bound_relief_frames < path.bound_relief_required_frames:
+        target = path.bound
+      path.bound = self._move(path.bound, target, CAP_RELAX_JERK if target > path.bound else CAP_TIGHTEN_JERK, self.dt)
+      path.state = AccelControllerState.hold if matched or coast_to_match else AccelControllerState.restrict
+      path.relief_frames = 0
+      return False
+
+    if path.state in (AccelControllerState.restrict, AccelControllerState.hold):
+      self._clear_bound_relief(path)
+      relief = not has_lead or moving_away
+      path.relief_frames = path.relief_frames + 1 if relief else 0
+      path.bound = min(path.bound if path.bound is not None else action_accel, 0.0)
+      if path.relief_frames < RELIEF_CONFIRM_FRAMES:
+        path.state = AccelControllerState.hold
+        return False
+      path.state = AccelControllerState.free
+      path.relief_frames = 0
+      return False
+
+    if path.bound is None:
+      path.bound = positive_accel_max
+    else:
+      path.bound = self._move(path.bound, positive_accel_max, PROFILE_TRANSITION_JERK, self.dt)
+    self._clear_bound_relief(path)
+    path.state = AccelControllerState.free
+    return False
+
+  @staticmethod
+  def _build_accel_ceiling(bound: float, v_ego: float, planner_accel: float, action_time: float) -> tuple[float, ...] | None:
+    if bound >= ACCEL_MAX - 1e-9:
+      return None
+    a0 = float(np.clip(planner_accel, ACCEL_MIN, ACCEL_MAX))
+    if bound > 0.0:
+      ceiling = np.full(len(T_IDXS), min(bound + POSITIVE_MPC_HEADROOM, ACCEL_MAX))
+    elif bound == 0.0:
+      ceiling = np.maximum(0.0, a0 - HORIZON_DOWN_JERK * T_IDXS)
+    else:
+      descent = np.maximum(bound, a0 - HORIZON_DOWN_JERK * T_IDXS)
+      reach_time = max((a0 - bound) / HORIZON_DOWN_JERK, 0.0)
+      release_time = max(action_time + HORIZON_HOLD_TIME, reach_time + HORIZON_HOLD_TIME)
+      recovery = np.clip(bound + HORIZON_UP_JERK * np.maximum(T_IDXS - release_time, 0.0), bound, 0.0)
+      ceiling = np.where(T_IDXS <= release_time, descent, np.maximum(descent, recovery))
+      budget = HORIZON_SPEED_BUDGET * max(v_ego, 0.0)
+      negative_area = float(np.trapezoid(-np.minimum(ceiling, 0.0), T_IDXS))
+      if negative_area > budget and negative_area > 1e-9:
+        ceiling = np.where(ceiling < 0.0, ceiling * budget / negative_area, ceiling)
+    ceiling = np.clip(ceiling, ACCEL_MIN, ACCEL_MAX)
+    ceiling[0] = max(ceiling[0], a0)
+    return tuple(float(value) for value in ceiling)
+
+  @staticmethod
+  def _valid_context(base_speed: float, v_ego: float, a_ego: float, planner_accel: float, action_accel: float,
+                     positive_accel_max: float, delay: float, engaged: bool, cruise_initialized: bool, controller_fault: bool) -> bool:
+    values = (base_speed, v_ego, a_ego, planner_accel, action_accel, positive_accel_max, delay)
+    return (engaged and cruise_initialized and not controller_fault and base_speed >= 0.0 and v_ego >= -VEGO_NOISE_TOLERANCE
+            and delay >= 0.0 and all(math.isfinite(value) for value in values))
+
+  def _update_freshness(self, path: _ControllerPath, radar_fresh: bool) -> bool:
+    if radar_fresh:
+      path.stale_frames = 0
+      return True
+    path.stale_frames += 1
+    if path.stale_frames < self.radar_stale_frames and (path.bound is not None or path.urgent):
+      return False
+    path.reset()
+    return False
+
+  def reset(self) -> None:
+    self.live.reset()
+    self.shadow.reset()
+
+  def update(self, radar_state, *, base_speed: float, v_ego: float, a_ego: float, profile: int | AccelProfile, follow_personality,
+             enabled: bool, acc_selected: bool, engaged: bool, cruise_initialized: bool, planner_accel: float, action_accel: float,
+             stock_accel_max: float, previous_should_stop: bool, controller_fault: bool = False,
+             radar_fresh: bool = True) -> AccelControllerResult:
+    selected_profile = self._profile(profile)
+    sanitized_v_ego = max(v_ego, 0.0) if math.isfinite(v_ego) and v_ego >= -VEGO_NOISE_TOLERANCE else v_ego
+    profile_accel_max = self.get_profile_accel_max(selected_profile, sanitized_v_ego)
+    try:
+      stock_accel_max = float(stock_accel_max)
+    except (OverflowError, TypeError, ValueError):
+      stock_accel_max = math.nan
+    positive_accel_max = (max(0.0, min(profile_accel_max, stock_accel_max, ACCEL_MAX))
+                          if math.isfinite(profile_accel_max) and math.isfinite(stock_accel_max) else math.nan)
+    valid_context = self._valid_context(base_speed, sanitized_v_ego, a_ego, planner_accel, action_accel, positive_accel_max,
+                                        self._delay(), engaged, cruise_initialized, controller_fault)
+    envelope = (self.calculate_energy_envelope(radar_state, sanitized_v_ego, a_ego, selected_profile, follow_personality)
+                if valid_context and radar_fresh else EnergyEnvelope(lead_status=self._radar_has_lead(radar_state)))
+
+    shadow_fresh = self._update_freshness(self.shadow, radar_fresh) if valid_context else False
+    if valid_context and radar_fresh:
+      self._update_path(self.shadow, envelope, base_speed, sanitized_v_ego, action_accel, positive_accel_max,
+                        selected_profile, previous_should_stop)
+      shadow_active = True
+    elif valid_context and not shadow_fresh and (self.shadow.bound is not None or self.shadow.urgent):
+      shadow_active = True
+    else:
+      self.shadow.reset()
+      shadow_active = False
+
+    live_context = valid_context and bool(enabled) and bool(acc_selected)
+    live_fresh = self._update_freshness(self.live, radar_fresh) if live_context else False
+    if live_context and radar_fresh:
+      stock_mode = self._update_path(self.live, envelope, base_speed, sanitized_v_ego, action_accel,
+                                     positive_accel_max, selected_profile, previous_should_stop)
+      live_active = True
+    elif live_context and not live_fresh and (self.live.bound is not None or self.live.urgent):
+      stock_mode = self.live.urgent
+      live_active = True
+    else:
+      self.live.reset()
+      stock_mode = False
+      live_active = False
+
+    if live_active and not stock_mode and self.live.bound is not None:
+      effective_accel_max = float(np.clip(self.live.bound, ACCEL_MIN, ACCEL_MAX))
+      if self.live.bound_relief_frames and self.live.lead_speed_relief:
+        effective_accel_max = min(effective_accel_max, action_accel + RELIEF_MPC_JERK * self.dt)
+      mpc_accel_max = self._build_accel_ceiling(effective_accel_max, sanitized_v_ego, planner_accel, self._delay())
+    else:
+      effective_accel_max = math.inf
+      mpc_accel_max = None
+
+    return AccelControllerResult(
+      target_speed=0.0 if live_active and self.live.state == AccelControllerState.stopHold else base_speed,
+      enabled=bool(enabled), active=live_active, shadow_active=shadow_active,
+      launching=live_active and self.live.departing_from_stop, stock_mode=stock_mode, profile=selected_profile,
+      profile_accel_max=profile_accel_max if live_active else math.inf,
+      positive_accel_max=positive_accel_max if live_active else math.inf, effective_accel_max=effective_accel_max,
+      mpc_accel_max=mpc_accel_max,
+      state=self.live.state, shadow_state=self.shadow.state, base_speed=base_speed, raw_energy_cap=envelope.cap,
+      live_filtered_cap=self.live.filtered_cap if live_active else math.inf,
+      shadow_filtered_cap=self.shadow.filtered_cap if shadow_active else math.inf, selected_lead=envelope.selected_lead,
+      selected_lead_speed=envelope.selected_lead_speed, usable_gap=envelope.usable_gap,
+      closing_speed=envelope.closing_speed, required_decel=envelope.required_decel,
+    )
+
+  @staticmethod
+  def _radar_has_lead(radar_state) -> bool:
+    try:
+      return bool(radar_state.leadOne.status or radar_state.leadTwo.status)
+    except (AttributeError, TypeError, ValueError):
+      return True

sunnypilot/selfdrive/controls/lib/accel_personality/constants.py

@@ -0,0 +1,75 @@
+from dataclasses import dataclass
+from enum import IntEnum
+
+
+class AccelProfile(IntEnum):
+  eco = 0
+  normal = 1
+  sport = 2
+
+
+@dataclass(frozen=True)
+class ProfileConfig:
+  comfort_decel: float
+  anticipation_margin: float
+  glide_decel: float
+
+
+PROFILE_CONFIGS = {
+  AccelProfile.eco: ProfileConfig(comfort_decel=0.25, anticipation_margin=0.15, glide_decel=0.12),
+  AccelProfile.normal: ProfileConfig(comfort_decel=0.35, anticipation_margin=1.00, glide_decel=0.16),
+  AccelProfile.sport: ProfileConfig(comfort_decel=0.50, anticipation_margin=2.00, glide_decel=0.20),
+}
+
+ACCEL_PROFILE_MAX_BP = [0.0, 3.0, 10.0, 25.0, 40.0]
+ACCEL_PROFILE_MAX_V = {
+  AccelProfile.eco: [1.55, 1.25, 0.85, 0.40, 0.20],
+  AccelProfile.normal: [1.70, 1.40, 1.05, 0.55, 0.35],
+  AccelProfile.sport: [2.00, 1.90, 1.70, 0.90, 0.60],
+}
+
+CAP_FILTER_FRAMES = 5
+RELIEF_CONFIRM_FRAMES = 5
+STOP_HOLD_EXIT_FRAMES = 4
+STOP_HOLD_EGO_SPEED = 0.30
+STOPPED_LEAD_SPEED = 0.30
+STOP_HOLD_EXIT_SPEED = 0.80
+STOP_HOLD_CREEP_SPEED = 0.15
+STOP_HOLD_CREEP_DISTANCE = 0.30
+STOP_HOLD_CREEP_ABORT_FRAMES = 4
+STOP_GAP_RESERVE = 0.75
+STOP_GAP_RESERVE_LEAD_SPEED = 2.0
+STOP_GAP_RESERVE_DECEL_BP = (0.30, 0.80)
+MPC_SEED_RISE_RATE = 6.0
+APPROACH_MIN_SPEED = 2.0
+APPROACH_CLOSING_SPEED = 0.15
+BRAKE_CAP_MARGIN = 0.50
+APPROACH_LEAD_DECEL = 0.10
+APPROACH_LEAD_SPEED_MARGIN = 0.50
+RELIEF_CAP_MARGIN = 0.35
+COAST_MATCH_CLOSING_SPEED = 2.50
+COAST_MATCH_USABLE_GAP = 4.0
+REQUIRED_DECEL_MARGIN = 0.03
+ROUTINE_DECEL_MAX = 1.0
+CAP_TIGHTEN_JERK = 0.60
+CAP_RELAX_JERK = 0.80
+SHALLOW_BRAKE_BOUND = -0.25
+SHALLOW_BRAKE_RELIEF_TIME = 1.75
+RELIEF_MPC_JERK = 3.20
+RELIEF_LEAD_SPEED_STEP = 0.05
+DROPOUT_ACTION_ACCEL_MARGIN = 0.08
+PROFILE_TRANSITION_JERK = 1.50
+POSITIVE_MPC_HEADROOM = 0.02
+URGENT_CLOSING_SPEED = 12.0
+URGENT_REQUIRED_DECEL = 1.0
+URGENT_TTC = 3.2
+URGENT_TTC_MIN_CLOSING = 1.0
+URGENT_RELEASE_ACCEL = -0.20
+HORIZON_DOWN_JERK = 2.0
+HORIZON_UP_JERK = 1.0
+HORIZON_HOLD_TIME = 0.50
+HORIZON_SPEED_BUDGET = 0.80
+RADAR_STALE_TIMEOUT = 0.50
+MAX_LEAD_ACCEL_TAU = 10.0
+MIN_LEAD_SPEED = -1.0
+VEGO_NOISE_TOLERANCE = 0.10

sunnypilot/selfdrive/controls/lib/accel_personality/tests/test_accel_controller.py

@@ -0,0 +1,445 @@
+import math
+from types import SimpleNamespace
+
+import numpy as np
+import pytest
+
+from cereal import log
+from opendbc.car.interfaces import ACCEL_MAX, ACCEL_MIN
+from openpilot.common.realtime import DT_MDL
+from openpilot.selfdrive.controls.lib.longitudinal_mpc_lib.long_mpc import STOP_DISTANCE, T_IDXS, LongitudinalMpc, get_T_FOLLOW
+from openpilot.sunnypilot.selfdrive.controls.lib.accel_personality.accel_controller import (
+  ACCEL_PROFILE_MAX_BP,
+  ACCEL_PROFILE_MAX_V,
+  CAP_FILTER_FRAMES,
+  HORIZON_SPEED_BUDGET,
+  POSITIVE_MPC_HEADROOM,
+  PROFILE_CONFIGS,
+  PROFILE_TRANSITION_JERK,
+  RADAR_STALE_TIMEOUT,
+  RELIEF_CONFIRM_FRAMES,
+  SHALLOW_BRAKE_BOUND,
+  STOP_GAP_RESERVE,
+  STOP_HOLD_CREEP_ABORT_FRAMES,
+  STOP_HOLD_EXIT_FRAMES,
+  AccelController,
+  AccelControllerState,
+  AccelProfile,
+)
+
+
+def make_lead(*, status=False, d_rel=0.0, v_lead_k=0.0, a_lead_k=0.0, a_lead_tau=1.5):
+  return SimpleNamespace(status=status, dRel=d_rel, vLeadK=v_lead_k, aLeadK=a_lead_k, aLeadTau=a_lead_tau)
+
+
+def make_radar(lead_one=None, lead_two=None):
+  return SimpleNamespace(leadOne=lead_one or make_lead(), leadTwo=lead_two or make_lead())
+
+
+def make_controller(delay=0.10):
+  return AccelController(SimpleNamespace(longitudinalActuatorDelay=delay))
+
+
+def update(controller, radar_state=None, **overrides):
+  args = {
+    "base_speed": 25.0,
+    "v_ego": 10.0,
+    "a_ego": 0.0,
+    "profile": AccelProfile.normal,
+    "follow_personality": log.LongitudinalPersonality.standard,
+    "enabled": True,
+    "acc_selected": True,
+    "engaged": True,
+    "cruise_initialized": True,
+    "planner_accel": 0.0,
+    "action_accel": 0.0,
+    "stock_accel_max": ACCEL_MAX,
+    "previous_should_stop": False,
+  }
+  args.update(overrides)
+  return controller.update(radar_state or make_radar(), **args)
+
+
+def restrictive_radar():
+  return make_radar(make_lead(status=True, d_rel=25.0, v_lead_k=8.0, a_lead_k=-0.5))
+
+
+class TestProfiles:
+  def test_lookup_table_is_explicit_and_tunable(self):
+    assert ACCEL_PROFILE_MAX_BP == [0.0, 3.0, 10.0, 25.0, 40.0]
+    assert ACCEL_PROFILE_MAX_V == {
+      AccelProfile.eco: [1.55, 1.25, 0.85, 0.40, 0.20],
+      AccelProfile.normal: [1.70, 1.40, 1.05, 0.55, 0.35],
+      AccelProfile.sport: [2.00, 1.90, 1.70, 0.90, 0.60],
+    }
+
+  @pytest.mark.parametrize("profile", list(AccelProfile))
+  def test_lookup_interpolates_and_stays_inside_global_limit(self, profile):
+    for speed, expected in zip(ACCEL_PROFILE_MAX_BP, ACCEL_PROFILE_MAX_V[profile], strict=True):
+      assert AccelController.get_profile_accel_max(profile, speed) == expected
+    limits = [AccelController.get_profile_accel_max(profile, speed) for speed in np.linspace(-1.0, 50.0, 201)]
+    assert all(0.0 <= limit <= ACCEL_MAX for limit in limits)
+
+  @pytest.mark.parametrize("speed", [0.0, 3.0, 10.0, 25.0, 40.0])
+  def test_profile_order_is_distinct(self, speed):
+    limits = [AccelController.get_profile_accel_max(profile, speed) for profile in AccelProfile]
+    assert limits[0] < limits[1] < limits[2]
+
+  @pytest.mark.parametrize("profile", list(AccelProfile))
+  def test_clear_road_applies_profile_immediately(self, profile):
+    result = update(make_controller(), v_ego=0.0, profile=profile)
+    expected = ACCEL_PROFILE_MAX_V[profile][0]
+    assert result.active and result.state == AccelControllerState.free
+    assert result.target_speed == result.base_speed == 25.0
+    assert result.positive_accel_max == expected
+    assert result.effective_accel_max == expected
+    if expected == ACCEL_MAX:
+      assert result.mpc_accel_max is None
+    else:
+      np.testing.assert_array_equal(result.mpc_accel_max, min(expected + POSITIVE_MPC_HEADROOM, ACCEL_MAX))
+
+  def test_turn_or_throttle_limit_intersects_profile(self):
+    result = update(make_controller(), profile=AccelProfile.sport, stock_accel_max=0.0)
+    assert result.positive_accel_max == 0.0
+    assert result.effective_accel_max == 0.0
+    np.testing.assert_array_equal(result.mpc_accel_max, 0.0)
+
+  def test_profile_switch_changes_ceiling_without_a_step(self):
+    controller = make_controller()
+    sport = update(controller, profile=AccelProfile.sport, v_ego=10.0)
+    eco = update(controller, profile=AccelProfile.eco, v_ego=10.0)
+    assert sport.effective_accel_max > eco.effective_accel_max > eco.positive_accel_max
+    assert sport.effective_accel_max - eco.effective_accel_max == pytest.approx(PROFILE_TRANSITION_JERK * DT_MDL)
+
+  def test_invalid_profile_defaults_to_normal(self):
+    result = update(make_controller(), profile=999)
+    assert result.profile == AccelProfile.normal
+
+
+class TestEnergyEnvelope:
+  def test_relative_pace_energy_formula(self):
+    controller = make_controller()
+    lead = make_lead(status=True, d_rel=50.0, v_lead_k=8.0)
+    envelope = controller.calculate_energy_envelope(make_radar(lead), 10.0, 0.0, AccelProfile.normal)
+    delay = controller._delay()
+    lead_xv = LongitudinalMpc.extrapolate_lead(lead.dRel, lead.vLeadK, lead.aLeadK, lead.aLeadTau)
+    x_lead = float(np.interp(delay, T_IDXS, lead_xv[:, 0]))
+    v_lead = float(np.interp(delay, T_IDXS, lead_xv[:, 1]))
+    x_ego, _ = controller._project_ego(10.0, 0.0, delay)
+    gap = max(x_lead - x_ego - STOP_DISTANCE - get_T_FOLLOW(log.LongitudinalPersonality.standard) * v_lead, 0.0)
+    expected = v_lead + math.sqrt(2.0 * PROFILE_CONFIGS[AccelProfile.normal].comfort_decel * gap)
+    assert envelope.cap == pytest.approx(expected)
+    assert envelope.cap != pytest.approx(math.sqrt(v_lead**2 + 2.0 * PROFILE_CONFIGS[AccelProfile.normal].comfort_decel * gap))
+
+  def test_profile_order_controls_approach_timing(self):
+    controller = make_controller()
+    radar = make_radar(make_lead(status=True, d_rel=50.0, v_lead_k=8.0))
+    caps = [controller.calculate_energy_envelope(radar, 10.0, 0.0, profile).cap for profile in AccelProfile]
+    assert caps[0] < caps[1] < caps[2]
+
+  def test_stopped_lead_reserve_only_reduces_comfort_gap(self):
+    controller = make_controller()
+    lead = make_lead(status=True, d_rel=20.0, v_lead_k=0.0)
+    envelope = controller.calculate_energy_envelope(make_radar(lead), 2.0, 0.0, AccelProfile.normal)
+    expected = math.sqrt(2.0 * PROFILE_CONFIGS[AccelProfile.normal].comfort_decel * envelope.usable_gap)
+
+    assert envelope.required_decel < 0.30
+    assert envelope.safety_usable_gap - envelope.usable_gap == pytest.approx(STOP_GAP_RESERVE)
+    assert envelope.cap == pytest.approx(expected)
+    assert envelope.departure_cap > envelope.cap
+
+  def test_stop_reserve_fades_out_of_urgent_braking(self):
+    controller = make_controller()
+    lead = make_lead(status=True, d_rel=20.0, v_lead_k=0.0)
+    envelope = controller.calculate_energy_envelope(make_radar(lead), 20.0, 0.0, AccelProfile.normal)
+
+    assert envelope.required_decel > 0.80
+    assert envelope.usable_gap == envelope.safety_usable_gap
+    assert envelope.cap == envelope.departure_cap
+    assert controller._ttc(envelope) == pytest.approx(envelope.safety_usable_gap / envelope.closing_speed)
+
+  def test_more_restrictive_lead_is_selected(self):
+    radar = make_radar(make_lead(status=True, d_rel=70.0, v_lead_k=12.0), make_lead(status=True, d_rel=25.0, v_lead_k=8.0))
+    envelope = make_controller().calculate_energy_envelope(radar, 10.0, 0.0, AccelProfile.normal)
+    assert envelope.selected_lead == 1
+
+  @pytest.mark.parametrize("field,value", [("aLeadK", math.nan), ("aLeadK", math.inf), ("aLeadTau", math.nan), ("aLeadTau", -1.0)])
+  def test_nonessential_invalid_lead_fields_are_sanitized(self, field, value):
+    lead = make_lead(status=True, d_rel=30.0, v_lead_k=8.0)
+    setattr(lead, field, value)
+    envelope = make_controller().calculate_energy_envelope(make_radar(lead), 10.0, 0.0, AccelProfile.normal)
+    assert envelope.selected_lead == 0
+    assert math.isfinite(envelope.cap)
+
+  @pytest.mark.parametrize("field,value", [("dRel", math.nan), ("dRel", -1.0), ("vLeadK", math.nan), ("vLeadK", -2.0)])
+  def test_invalid_geometry_is_not_used(self, field, value):
+    lead = make_lead(status=True, d_rel=30.0, v_lead_k=8.0)
+    setattr(lead, field, value)
+    envelope = make_controller().calculate_energy_envelope(make_radar(lead), 10.0, 0.0, AccelProfile.normal)
+    assert envelope.selected_lead == -1
+    assert envelope.lead_status
+
+  def test_raw_radar_is_never_mutated(self):
+    lead = make_lead(status=True, d_rel=30.0, v_lead_k=8.0, a_lead_k=-15.0, a_lead_tau=math.nan)
+    before = vars(lead).copy()
+    make_controller().calculate_energy_envelope(make_radar(lead), 10.0, 0.0, AccelProfile.normal)
+    assert vars(lead) == before
+
+
+class TestAccelControllerState:
+  def test_five_frame_median_needs_three_restrictive_samples(self):
+    controller = make_controller()
+    results = [update(controller, restrictive_radar()) for _ in range(CAP_FILTER_FRAMES)]
+    assert math.isinf(results[1].live_filtered_cap)
+    assert math.isfinite(results[2].live_filtered_cap)
+
+  def test_routine_approach_builds_safe_finite_horizon_ceiling(self):
+    controller = make_controller()
+    result = None
+    for _ in range(CAP_FILTER_FRAMES):
+      result = update(controller, restrictive_radar())
+    assert result is not None and result.state == AccelControllerState.restrict
+    ceiling = np.asarray(result.mpc_accel_max)
+    assert ceiling.shape == T_IDXS.shape
+    assert np.all(np.isfinite(ceiling))
+    assert np.all((ceiling >= ACCEL_MIN) & (ceiling <= ACCEL_MAX))
+    assert ceiling[0] >= 0.0
+    assert np.min(ceiling) < -0.05 and ceiling[-1] == pytest.approx(0.0)
+    assert np.trapezoid(-np.minimum(ceiling, 0.0), T_IDXS) <= HORIZON_SPEED_BUDGET * 10.0 + 1e-9
+
+  def test_ongoing_mpc_braking_does_not_ratchet_the_controller(self):
+    controller = make_controller()
+    for _ in range(CAP_FILTER_FRAMES):
+      previous = update(controller, restrictive_radar())
+    result = update(controller, restrictive_radar(), action_accel=-1.2, planner_accel=-1.0)
+    assert result.effective_accel_max >= previous.effective_accel_max - 0.60 * DT_MDL - 1e-9
+
+  def test_two_dropouts_cannot_release_restriction(self):
+    controller = make_controller()
+    for _ in range(CAP_FILTER_FRAMES):
+      restricted = update(controller, restrictive_radar())
+    results = [update(controller) for _ in range(2)]
+    assert all(result.active and result.effective_accel_max <= 0.0 for result in results)
+    assert all(result.effective_accel_max >= restricted.effective_accel_max for result in results)
+
+  def test_relief_requires_consecutive_confirmation(self):
+    controller = make_controller()
+    for _ in range(CAP_FILTER_FRAMES):
+      update(controller, restrictive_radar())
+    moving_away = make_radar(make_lead(status=True, d_rel=45.0, v_lead_k=13.0))
+    early = [update(controller, moving_away) for _ in range(RELIEF_CONFIRM_FRAMES - 1)]
+    assert all(result.state == AccelControllerState.hold and result.effective_accel_max <= 0.0 for result in early)
+    released = update(controller, moving_away)
+    assert released.state == AccelControllerState.free
+    assert released.effective_accel_max <= 0.0
+    accelerating = update(controller, moving_away)
+    assert released.effective_accel_max < accelerating.effective_accel_max <= accelerating.positive_accel_max
+
+  def test_shallow_brake_relief_uses_long_confirmation_without_delaying_tightening(self):
+    controller = make_controller()
+    controller.live.state = AccelControllerState.restrict
+    controller.live.bound = SHALLOW_BRAKE_BOUND + 0.05
+    matched = make_radar(make_lead(status=True, d_rel=20.0, v_lead_k=10.0))
+
+    held = [update(controller, matched) for _ in range(controller.shallow_brake_relief_frames - 1)]
+    assert all(result.effective_accel_max == pytest.approx(SHALLOW_BRAKE_BOUND + 0.05) for result in held)
+    relaxed = update(controller, matched)
+    assert relaxed.effective_accel_max > held[-1].effective_accel_max
+
+    for _ in range(CAP_FILTER_FRAMES):
+      tightened = update(controller, restrictive_radar())
+    assert tightened.effective_accel_max < relaxed.effective_accel_max
+
+  def test_strong_brake_relief_keeps_existing_confirmation(self):
+    controller = make_controller()
+    controller.live.state = AccelControllerState.restrict
+    controller.live.bound = SHALLOW_BRAKE_BOUND - 0.25
+    matched = make_radar(make_lead(status=True, d_rel=20.0, v_lead_k=10.0))
+
+    held = [update(controller, matched) for _ in range(RELIEF_CONFIRM_FRAMES - 1)]
+    assert all(result.effective_accel_max == pytest.approx(SHALLOW_BRAKE_BOUND - 0.25) for result in held)
+    relaxed = update(controller, matched)
+    assert relaxed.effective_accel_max > held[-1].effective_accel_max
+    continuing = [update(controller, matched) for _ in range(8)]
+    assert all(current.effective_accel_max > previous.effective_accel_max
+               for previous, current in zip([relaxed, *continuing[:-1]], continuing, strict=True))
+
+  def test_urgent_frame_uses_exact_stock_path(self):
+    urgent = make_radar(make_lead(status=True, d_rel=18.0, v_lead_k=0.0))
+    result = update(make_controller(), urgent, v_ego=20.0)
+    assert result.active and result.stock_mode
+    assert result.mpc_accel_max is None
+    assert math.isinf(result.effective_accel_max)
+
+  def test_urgent_relief_stays_stock_until_braking_has_recovered(self):
+    controller = make_controller()
+    urgent = make_radar(make_lead(status=True, d_rel=18.0, v_lead_k=0.0))
+    update(controller, urgent, v_ego=20.0)
+    result = update(controller, action_accel=-1.5, planner_accel=-1.2, v_ego=19.8)
+    assert result.stock_mode
+    assert result.mpc_accel_max is None
+    recovered = [update(controller, action_accel=0.0, planner_accel=0.0, v_ego=19.8) for _ in range(RELIEF_CONFIRM_FRAMES)]
+    assert all(sample.stock_mode for sample in recovered[:-1])
+    assert recovered[-1].state == AccelControllerState.free
+
+  def test_stop_hold_needs_four_departure_frames(self):
+    controller = make_controller()
+    stopped = make_radar(make_lead(status=True, d_rel=6.0, v_lead_k=0.0))
+    held = update(controller, stopped, base_speed=8.0, v_ego=0.1, previous_should_stop=True)
+    assert held.state == AccelControllerState.stopHold
+    np.testing.assert_array_equal(held.mpc_accel_max, 0.0)
+
+    departing = make_radar(make_lead(status=True, d_rel=8.0, v_lead_k=2.0))
+    confirmation = [update(controller, departing, base_speed=8.0, v_ego=0.1) for _ in range(STOP_HOLD_EXIT_FRAMES)]
+    assert all(result.effective_accel_max == 0.0 for result in confirmation[:-1])
+    launched = confirmation[-1]
+    assert launched.launching and launched.state == AccelControllerState.free
+    assert launched.effective_accel_max == launched.positive_accel_max
+
+  def test_false_creep_speed_without_range_gain_stays_held(self):
+    controller = make_controller()
+    stopped = make_radar(make_lead(status=True, d_rel=6.0, v_lead_k=0.0))
+    update(controller, stopped, base_speed=8.0, v_ego=0.1, previous_should_stop=True)
+
+    false_creep = make_radar(make_lead(status=True, d_rel=6.0, v_lead_k=0.30))
+    held = [update(controller, false_creep, base_speed=8.0, v_ego=0.1) for _ in range(20)]
+    assert all(result.state == AccelControllerState.stopHold and not result.launching for result in held)
+
+  def test_invalid_lead_geometry_cannot_confirm_departure(self):
+    controller = make_controller()
+    stopped = make_radar(make_lead(status=True, d_rel=6.0, v_lead_k=0.0))
+    update(controller, stopped, base_speed=8.0, v_ego=0.1, previous_should_stop=True)
+
+    invalid = make_radar(make_lead(status=True, d_rel=math.nan, v_lead_k=0.30))
+    held = [update(controller, invalid, base_speed=8.0, v_ego=0.1) for _ in range(2 * STOP_HOLD_EXIT_FRAMES)]
+    assert all(result.state == AccelControllerState.stopHold and not result.launching for result in held)
+
+  def test_short_range_drop_and_restore_cannot_fake_creep_departure(self):
+    controller = make_controller()
+    stopped = make_radar(make_lead(status=True, d_rel=6.0, v_lead_k=0.0))
+    update(controller, stopped, base_speed=8.0, v_ego=0.1, previous_should_stop=True)
+
+    low_range = make_radar(make_lead(status=True, d_rel=5.0, v_lead_k=0.0))
+    for _ in range(2):
+      update(controller, low_range, base_speed=8.0, v_ego=0.1)
+    restored = make_radar(make_lead(status=True, d_rel=6.0, v_lead_k=0.30))
+    results = [update(controller, restored, base_speed=8.0, v_ego=0.1) for _ in range(8)]
+
+    assert all(result.state == AccelControllerState.stopHold and not result.launching for result in results)
+
+  def test_slow_creep_with_confirmed_range_gain_releases(self):
+    controller = make_controller()
+    stopped = make_radar(make_lead(status=True, d_rel=6.0, v_lead_k=0.0))
+    update(controller, stopped, base_speed=8.0, v_ego=0.1, previous_should_stop=True)
+
+    results = []
+    for frame in range(20):
+      creep = make_radar(make_lead(status=True, d_rel=6.0 + 0.05 * frame, v_lead_k=0.30))
+      results.append(update(controller, creep, base_speed=8.0, v_ego=0.1))
+
+    launched = [frame for frame, result in enumerate(results) if result.launching]
+    assert launched and launched[0] >= STOP_HOLD_EXIT_FRAMES
+    assert all(result.state == AccelControllerState.stopHold for result in results[:launched[0]])
+
+  def test_slow_creep_survives_lead_slot_switching(self):
+    controller = make_controller()
+    stopped = make_radar(make_lead(status=True, d_rel=6.0), make_lead(status=True, d_rel=6.1))
+    update(controller, stopped, base_speed=8.0, v_ego=0.1, previous_should_stop=True)
+
+    results = []
+    for frame in range(24):
+      distance = 6.0 + 0.04 * frame
+      offset = 0.05 if frame % 2 else 0.0
+      lead_one = make_lead(status=True, d_rel=distance + offset, v_lead_k=0.30)
+      lead_two = make_lead(status=True, d_rel=distance + 0.05 - offset, v_lead_k=0.30)
+      results.append(update(controller, make_radar(lead_one, lead_two), base_speed=8.0, v_ego=0.1))
+
+    assert any(result.launching for result in results)
+
+  def test_departure_that_stops_again_returns_to_hold(self):
+    controller = make_controller()
+    stopped = make_radar(make_lead(status=True, d_rel=6.0, v_lead_k=0.0))
+    update(controller, stopped, base_speed=8.0, v_ego=0.1, previous_should_stop=True)
+    departing = make_radar(make_lead(status=True, d_rel=8.0, v_lead_k=2.0))
+    for _ in range(STOP_HOLD_EXIT_FRAMES):
+      launched = update(controller, departing, base_speed=8.0, v_ego=0.1)
+    assert launched.launching
+
+    stalled = make_radar(make_lead(status=True, d_rel=6.0, v_lead_k=0.11))
+    settling = [update(controller, stalled, base_speed=8.0, v_ego=0.1) for _ in range(STOP_HOLD_CREEP_ABORT_FRAMES)]
+    assert all(result.launching for result in settling[:-1])
+    assert settling[-1].state == AccelControllerState.stopHold and not settling[-1].launching
+
+  def test_invalid_geometry_after_departure_returns_to_hold(self):
+    controller = make_controller()
+    stopped = make_radar(make_lead(status=True, d_rel=6.0, v_lead_k=0.0))
+    update(controller, stopped, base_speed=8.0, v_ego=0.1, previous_should_stop=True)
+    departing = make_radar(make_lead(status=True, d_rel=8.0, v_lead_k=2.0))
+    for _ in range(STOP_HOLD_EXIT_FRAMES):
+      launched = update(controller, departing, base_speed=8.0, v_ego=0.1)
+    assert launched.launching
+
+    invalid = make_radar(make_lead(status=True, d_rel=math.nan, v_lead_k=0.30))
+    settling = [update(controller, invalid, base_speed=8.0, v_ego=0.1) for _ in range(STOP_HOLD_CREEP_ABORT_FRAMES)]
+    assert settling[-1].state == AccelControllerState.stopHold and not settling[-1].launching
+
+  def test_stale_radar_freezes_then_discards_live_state(self):
+    controller = make_controller()
+    for _ in range(CAP_FILTER_FRAMES):
+      restricted = update(controller, restrictive_radar())
+    hold_frames = math.ceil(RADAR_STALE_TIMEOUT / DT_MDL) - 1
+    frozen = [update(controller, radar_fresh=False) for _ in range(hold_frames)]
+    assert all(result.active and result.effective_accel_max == restricted.effective_accel_max for result in frozen)
+    timed_out = update(controller, radar_fresh=False)
+    assert not timed_out.active and timed_out.mpc_accel_max is None
+
+  def test_stale_radar_preserves_urgent_stock_passthrough_until_timeout(self):
+    controller = make_controller()
+    urgent = make_radar(make_lead(status=True, d_rel=18.0, v_lead_k=0.0))
+    result = update(controller, urgent, v_ego=20.0)
+    assert result.active and result.shadow_active and result.stock_mode
+
+    hold_frames = math.ceil(RADAR_STALE_TIMEOUT / DT_MDL) - 1
+    frozen = [update(controller, radar_fresh=False, v_ego=20.0) for _ in range(hold_frames)]
+    assert all(sample.active and sample.shadow_active and sample.stock_mode for sample in frozen)
+    assert all(sample.state == AccelControllerState.hold and sample.mpc_accel_max is None for sample in frozen)
+
+    timed_out = update(controller, radar_fresh=False, v_ego=20.0)
+    assert not timed_out.active and not timed_out.shadow_active and not timed_out.stock_mode
+
+  @pytest.mark.parametrize("override", [
+    {"enabled": False}, {"acc_selected": False}, {"engaged": False}, {"cruise_initialized": False}, {"controller_fault": True},
+  ])
+  def test_bypass_never_actuates(self, override):
+    result = update(make_controller(), restrictive_radar(), **override)
+    assert not result.active
+    assert result.target_speed == result.base_speed
+    assert result.mpc_accel_max is None
+    assert math.isinf(result.effective_accel_max)
+
+  def test_shadow_history_never_enters_live_actuation(self):
+    controller = make_controller()
+    for _ in range(CAP_FILTER_FRAMES):
+      shadow = update(controller, restrictive_radar(), enabled=False)
+    assert shadow.shadow_state == AccelControllerState.restrict
+    live = update(controller)
+    assert live.state == AccelControllerState.free
+    assert live.effective_accel_max > 0.0
+
+
+@pytest.mark.parametrize("v_ego", [0.0, 0.2, 0.5, 1.0, 2.0, 10.0, 40.0])
+@pytest.mark.parametrize("bound", [-3.5, -2.0, -1.0, -0.1, 0.0, 0.8, 2.0])
+def test_accel_ceiling_properties(v_ego, bound):
+  result = AccelController._build_accel_ceiling(bound, v_ego, planner_accel=0.3, action_time=0.25)
+  if bound >= ACCEL_MAX:
+    assert result is None
+    return
+  ceiling = np.asarray(result)
+  assert ceiling.shape == T_IDXS.shape
+  assert np.all(np.isfinite(ceiling))
+  assert np.all((ceiling >= ACCEL_MIN) & (ceiling <= ACCEL_MAX))
+  assert ceiling[0] >= 0.3 - 1e-9
+  if bound > 0.0:
+    np.testing.assert_array_equal(ceiling, min(bound + POSITIVE_MPC_HEADROOM, ACCEL_MAX))
+  if bound < 0.0:
+    assert np.trapezoid(-np.minimum(ceiling, 0.0), T_IDXS) <= HORIZON_SPEED_BUDGET * v_ego + 1e-9

sunnypilot/selfdrive/controls/lib/accel_personality/tests/test_accel_controller_interfaces.py

@@ -0,0 +1,246 @@
+import math
+from types import SimpleNamespace
+
+import numpy as np
+import pytest
+
+from cereal import custom, messaging
+from opendbc.car.interfaces import ACCEL_MAX, ACCEL_MIN
+from openpilot.selfdrive.controls.lib.longitudinal_mpc_lib.long_mpc import N, LongitudinalMpc
+from openpilot.sunnypilot.selfdrive.controls.lib.accel_personality import AccelControllerState, AccelProfile
+from openpilot.sunnypilot.selfdrive.controls.lib.longitudinal_planner import LongitudinalPlannerSP, LongitudinalPlanSource
+
+
+def radar_state():
+  return messaging.new_message('radarState').radarState
+
+
+def test_legacy_profile_enum_keeps_toyota_importable():
+  expected = {"eco": 0, "normal": 1, "sport": 2}
+  assert custom.LongitudinalPlanSP.AccelerationPersonality.schema.enumerants == expected
+  assert custom.LongitudinalPlanSP.AccelController.Profile.schema.enumerants == expected
+  from opendbc.car.toyota.carstate import AccelPersonality, CarState
+  assert AccelPersonality.schema.enumerants == expected
+  assert CarState.__module__ == "opendbc.car.toyota.carstate"
+
+
+def test_stock_mpc_parameters_are_unchanged():
+  mpc = LongitudinalMpc()
+  mpc.set_cur_state(10.0, 0.0)
+  mpc.run = lambda: None
+  mpc.update(radar_state(), 30.0)
+  np.testing.assert_array_equal(mpc.params[:, 0], ACCEL_MIN)
+  np.testing.assert_array_equal(mpc.params[:, 1], ACCEL_MAX)
+
+
+def test_positive_scalar_changes_only_acceleration_ceiling():
+  radar = radar_state()
+  mpc = LongitudinalMpc()
+  mpc.set_cur_state(10.0, 0.0)
+  mpc.run = lambda: None
+  mpc.update(radar, 30.0)
+  stock = mpc.params.copy()
+  stock_source = mpc.source
+  mpc.update(radar, 30.0, accel_max=0.8)
+  np.testing.assert_array_equal(mpc.params[:, 0], stock[:, 0])
+  np.testing.assert_array_equal(mpc.params[:, 1], 0.8)
+  np.testing.assert_array_equal(mpc.params[:, 2:], stock[:, 2:])
+  assert mpc.source == stock_source
+
+
+def test_negative_finite_horizon_ceiling_is_applied_exactly():
+  ceiling = np.linspace(0.2, -0.8, N + 1)
+  mpc = LongitudinalMpc()
+  mpc.set_cur_state(10.0, 0.2)
+  mpc.run = lambda: None
+  mpc.update(radar_state(), 30.0, accel_max=ceiling)
+  np.testing.assert_allclose(mpc.params[:, 1], ceiling)
+  np.testing.assert_array_equal(mpc.params[:, 0], ACCEL_MIN)
+
+
+@pytest.mark.parametrize("accel_max", [np.inf, np.nan, -0.4, ACCEL_MIN, np.ones(N), "invalid"])
+def test_invalid_or_negative_scalar_limit_is_exact_stock(accel_max):
+  radar = radar_state()
+  mpc = LongitudinalMpc()
+  mpc.set_cur_state(10.0, 0.0)
+  mpc.run = lambda: None
+  mpc.update(radar, 30.0)
+  stock = mpc.params.copy()
+  mpc.update(radar, 30.0, accel_max=accel_max)
+  np.testing.assert_array_equal(mpc.params, stock)
+
+
+def test_custom_ceiling_keeps_raw_lead_obstacle_and_source_authoritative():
+  radar = radar_state()
+  radar.leadOne.status = True
+  radar.leadOne.dRel = 30.0
+  radar.leadOne.vLead = 5.0
+  radar.leadOne.aLeadK = 0.0
+  radar.leadOne.aLeadTau = 1.0
+  before = (radar.leadOne.dRel, radar.leadOne.vLead, radar.leadOne.aLeadK)
+  mpc = LongitudinalMpc()
+  mpc.set_cur_state(20.0, 0.0)
+  mpc.run = lambda: None
+  mpc.update(radar, 30.0)
+  stock = mpc.params.copy()
+  stock_source = mpc.source
+  mpc.update(radar, 30.0, accel_max=np.linspace(0.0, -0.5, N + 1))
+  np.testing.assert_array_equal(mpc.params[:, 0], stock[:, 0])
+  np.testing.assert_array_equal(mpc.params[:, 2:], stock[:, 2:])
+  assert mpc.source == stock_source
+  assert (radar.leadOne.dRel, radar.leadOne.vLead, radar.leadOne.aLeadK) == before
+
+
+def test_retry_seed_is_bounded_and_nonnegative_in_speed():
+  planner = LongitudinalPlannerSP.__new__(LongitudinalPlannerSP)
+  planner.mpc = LongitudinalMpc()
+  states = []
+  planner.mpc.solver = SimpleNamespace(set=lambda _stage, field, value: states.append(np.asarray(value)) if field == 'x' else None)
+  planner.mpc.set_cur_state(0.0, ACCEL_MIN)
+  planner._seed_mpc_current_state()
+  states = np.asarray(states)
+  assert len(states) == N + 1
+  assert np.all(np.diff(states[:, 0]) >= 0.0)
+  assert np.all(states[:, 1] >= 0.0)
+  assert np.all((states[:, 2] >= ACCEL_MIN) & (states[:, 2] <= ACCEL_MAX))
+
+
+def test_last_solve_failure_survives_internal_reset():
+  mpc = LongitudinalMpc()
+  mpc.last_solution_status = 3
+  mpc.reset()
+  assert mpc.solution_status == 0
+  assert mpc.last_solution_status == 3
+
+
+@pytest.mark.parametrize(("controller_active", "expected_accel"), [(True, -1.2), (False, None)])
+def test_e2e_to_acc_handoff_preserves_braking_only_when_controller_is_active(controller_active, expected_accel):
+  planner = LongitudinalPlannerSP.__new__(LongitudinalPlannerSP)
+  planner._previous_is_e2e = True
+  planner.accel_personality_enabled = controller_active
+  planner.accel_controller_fault_latched = False
+  planner.is_e2e = lambda _sm: False
+  planner.accel_controller = SimpleNamespace(reset=lambda: None)
+  planner.mpc = SimpleNamespace(
+    a_prev=np.zeros(N + 1), crash_cnt=0, v_solution=np.zeros(N + 1), a_solution=np.zeros(N + 1),
+    j_solution=np.zeros(N), last_solution_status=0,
+  )
+  planner.update_accel_controller = lambda *_args, **_kwargs: setattr(
+    planner, "accel_controller_result",
+    SimpleNamespace(enabled=controller_active, active=controller_active, shadow_active=controller_active,
+                    stock_mode=not controller_active, target_speed=20.0, mpc_accel_max=None,
+                    state=AccelControllerState.free, effective_accel_max=0.8 if controller_active else np.inf),
+  )
+  calls = []
+  planner._run_mpc = lambda *_args, **kwargs: calls.append(kwargs)
+
+  is_e2e = planner.update_accel_controller_mpc(
+    {}, 20.0, 20.0, True, reset_state=False, cruise_initialized=True, planner_accel=0.1,
+    previous_output_accel=-1.2, available_accel_max=ACCEL_MAX, previous_should_stop=False, force_decel=False,
+  )
+
+  assert not is_e2e
+  assert calls[0]["current_accel"] == expected_accel
+  assert calls[0]["seed_target"] is None
+
+
+def test_failed_e2e_to_acc_handoff_retries_without_custom_state():
+  planner = LongitudinalPlannerSP.__new__(LongitudinalPlannerSP)
+  planner._previous_is_e2e = True
+  planner.accel_personality_enabled = True
+  planner.accel_controller_fault_latched = False
+  planner.is_e2e = lambda _sm: False
+  reset_calls = []
+  planner.accel_controller = SimpleNamespace(reset=lambda: reset_calls.append(True))
+  planner.mpc = SimpleNamespace(
+    a_prev=np.zeros(N + 1), crash_cnt=0, v_solution=np.zeros(N + 1), a_solution=np.zeros(N + 1),
+    j_solution=np.zeros(N), last_solution_status=0,
+  )
+  planner.update_accel_controller = lambda *_args, **_kwargs: setattr(
+    planner, "accel_controller_result",
+    SimpleNamespace(enabled=True, active=True, shadow_active=True, stock_mode=True, target_speed=15.0, mpc_accel_max=None,
+                    state=AccelControllerState.hold, effective_accel_max=np.inf),
+  )
+  calls = []
+  positional_calls = []
+
+  def run_mpc(*args, **kwargs):
+    positional_calls.append(args)
+    calls.append(kwargs)
+    planner.mpc.last_solution_status = 1 if len(calls) == 1 else 0
+
+  planner._run_mpc = run_mpc
+  planner.update_accel_controller_mpc(
+    {}, 20.0, 20.0, True, reset_state=False, cruise_initialized=True, planner_accel=0.1,
+    previous_output_accel=-1.2, available_accel_max=ACCEL_MAX, previous_should_stop=False, force_decel=False,
+  )
+
+  assert [call.get("current_accel") for call in calls] == [-1.2, None]
+  assert [args[1] for args in positional_calls] == [15.0, 20.0]
+  assert len(positional_calls[1]) == 3
+  assert calls[0]["seed_target"] is None
+  assert calls[1]["seed"]
+  assert "current_accel" not in calls[1]
+  assert "retry_state" in calls[1]
+  assert reset_calls == [True]
+  assert planner.accel_controller_fault_latched
+
+
+def test_e2e_to_acc_handoff_never_turns_braking_into_acceleration():
+  planner = LongitudinalPlannerSP.__new__(LongitudinalPlannerSP)
+  planner._previous_is_e2e = True
+  planner.accel_personality_enabled = True
+  planner.accel_controller_fault_latched = False
+  planner.is_e2e = lambda _sm: False
+  planner.accel_controller = SimpleNamespace(reset=lambda: None)
+  planner.mpc = SimpleNamespace(
+    a_prev=np.zeros(N + 1), crash_cnt=0, v_solution=np.zeros(N + 1), a_solution=np.zeros(N + 1),
+    j_solution=np.zeros(N), last_solution_status=0,
+  )
+  planner.update_accel_controller = lambda *_args, **_kwargs: setattr(
+    planner, "accel_controller_result",
+    SimpleNamespace(enabled=True, active=True, shadow_active=True, stock_mode=True, target_speed=20.0, mpc_accel_max=None,
+                    state=AccelControllerState.free, effective_accel_max=np.inf),
+  )
+  calls = []
+  planner._run_mpc = lambda *_args, **kwargs: calls.append(kwargs)
+
+  planner.update_accel_controller_mpc(
+    {}, 20.0, 20.0, True, reset_state=False, cruise_initialized=True, planner_accel=-0.7,
+    previous_output_accel=0.3, available_accel_max=ACCEL_MAX, previous_should_stop=False, force_decel=False,
+  )
+
+  assert calls[0]["current_accel"] == -0.7
+
+
+def test_shadow_telemetry_publishes_controller_fields():
+  planner = LongitudinalPlannerSP.__new__(LongitudinalPlannerSP)
+  planner.source = LongitudinalPlanSource.cruise
+  planner.output_v_target = 20.0
+  planner.output_a_target = 0.0
+  planner.events_sp = SimpleNamespace(to_msg=list)
+  planner.dec = SimpleNamespace(mode=lambda: "acc", enabled=lambda: False, active=lambda: False)
+  planner.accel_controller_result = SimpleNamespace(
+    enabled=True, active=False, shadow_active=True, profile=AccelProfile.normal, state=AccelControllerState.inactive,
+    shadow_state=AccelControllerState.restrict, base_speed=20.0, raw_energy_cap=15.0, live_filtered_cap=np.inf,
+    shadow_filtered_cap=12.5, selected_lead=1, usable_gap=30.0, closing_speed=5.0, required_decel=0.4,
+    profile_accel_max=np.inf, effective_accel_max=np.inf,
+  )
+  planner.scc = SimpleNamespace(
+    vision=SimpleNamespace(state=0, output_v_target=20.0, output_a_target=0.0, current_lat_acc=0.0, max_pred_lat_acc=0.0,
+                           is_enabled=False, is_active=False),
+    map=SimpleNamespace(state=0, output_v_target=20.0, output_a_target=0.0, is_enabled=False, is_active=False),
+  )
+  planner.resolver = SimpleNamespace(speed_limit=0.0, speed_limit_last=0.0, speed_limit_final=0.0, speed_limit_final_last=0.0,
+                                     speed_limit_valid=False, speed_limit_last_valid=False, speed_limit_offset=0.0,
+                                     distance=0.0, source=custom.LongitudinalPlanSP.SpeedLimit.Source.none)
+  planner.sla = SimpleNamespace(state=custom.LongitudinalPlanSP.SpeedLimit.AssistState.disabled, is_enabled=False, is_active=False,
+                                output_v_target=20.0, output_a_target=0.0)
+  planner.e2e_alerts_helper = SimpleNamespace(green_light_alert=False, lead_depart_alert=False)
+  sent = {}
+  planner.publish_longitudinal_plan_sp(SimpleNamespace(all_checks=lambda service_list: True),
+                                       SimpleNamespace(send=lambda service, message: sent.update({service: message})))
+  telemetry = sent["longitudinalPlanSP"].longitudinalPlanSP.accelController
+  assert telemetry.vTargetShadow == pytest.approx(12.5)
+  assert telemetry.aMaxProfile == math.inf
+  assert telemetry.aMaxEffective == math.inf

sunnypilot/selfdrive/controls/lib/dec/constants.py

@@ -1,17 +1,43 @@
+from openpilot.common.realtime import DT_MDL
+
+
 class WMACConstants:
-  # Lead detection parameters
-  LEAD_WINDOW_SIZE = 6  # Stable detection window
-  LEAD_PROB = 0.45  # Balanced threshold for lead detection
+  TRAJECTORY_SIZE = 33
+  PARAM_READ_FRAMES = max(1, int(round(1.0 / DT_MDL)))
 
-  # Slow down detection parameters
-  SLOW_DOWN_WINDOW_SIZE = 5  # Responsive but stable
-  SLOW_DOWN_PROB = 0.3  # Balanced threshold for slow down scenarios
+  EMERGENCY_HOLD_FRAMES = max(1, int(round(0.75 / DT_MDL)))
+  MIN_MODE_DURATION = {'acc': max(1, int(round(0.6 / DT_MDL))), 'blended': max(1, int(round(0.5 / DT_MDL)))}
+  ENTER_BLENDED_FRAMES = max(1, int(round(0.4 / DT_MDL)))
+  EXIT_BLENDED_FRAMES = max(1, int(round(0.35 / DT_MDL)))
+  STANDSTILL_FRAMES = max(1, int(round(0.2 / DT_MDL)))
 
-  # Optimized slow down distance curve - smooth and progressive
+  LEAD_PROB = 0.45
+  LEAD_EXIT_PROB = 0.25
+  LEAD_RISE_RATE = 1.0
+  LEAD_FALL_RATE = 0.35
+  RADAR_LEAD_ACC_PROB = 0.5
+  RADAR_LEAD_ACC_EXIT_PROB = 0.4
+  RADAR_LEAD_ACC_RISE_RATE = 1.0
+  RADAR_LEAD_ACC_FALL_RATE = 0.25
+
+  SLOW_DOWN_PROB = 0.5
+  SLOW_DOWN_EXIT_PROB = 0.4
+  SLOW_DOWN_RISE_RATE = 0.65
+  SLOW_DOWN_FALL_RATE = 0.15
   SLOW_DOWN_BP = [0., 10., 20., 30., 40., 50., 55., 60.]
   SLOW_DOWN_DIST = [32., 46., 64., 86., 108., 130., 145., 165.]
+  URGENT_SLOW_DOWN_PROB = 0.85
 
-  # Slowness detection parameters
-  SLOWNESS_WINDOW_SIZE = 10  # Stable slowness detection
-  SLOWNESS_PROB = 0.55  # Clear threshold for slowness
-  SLOWNESS_CRUISE_OFFSET = 1.025  # Conservative cruise speed offset
+  MODEL_DECEL_START = -0.5
+  MODEL_DECEL_RANGE = 2.0
+  ENDPOINT_URGENCY_GAIN = 1.3
+  CRITICAL_ENDPOINT_FACTOR = 0.3
+  CRITICAL_URGENCY_GAIN = 1.5
+  SPEED_URGENCY_MIN = 25.0
+  SPEED_URGENCY_RANGE = 80.0
+
+  SLOWNESS_PROB = 0.55
+  SLOWNESS_EXIT_PROB = 0.45
+  SLOWNESS_RISE_RATE = 0.35
+  SLOWNESS_FALL_RATE = 0.5
+  SLOWNESS_CRUISE_OFFSET = 1.025

sunnypilot/selfdrive/controls/lib/dec/dec.py

@@ -6,129 +6,116 @@ See the LICENSE.md file in the root directory for more details.
 """
 # Version = 2025-6-30
 
-from cereal import messaging
-from opendbc.car import structs
-from numpy import interp
-from openpilot.common.params import Params
-from openpilot.common.realtime import DT_MDL
-from openpilot.sunnypilot.selfdrive.controls.lib.dec.constants import WMACConstants
 from typing import Literal
 
-# d-e2e, from modeldata.h
-TRAJECTORY_SIZE = 33
-SET_MODE_TIMEOUT = 15
+from cereal import messaging
+from numpy import interp
+from opendbc.car import structs
+from openpilot.common.params import Params
+from openpilot.sunnypilot.selfdrive.controls.lib.dec.constants import WMACConstants
 
-# Define the valid mode types
 ModeType = Literal['acc', 'blended']
 
 
-class SmoothKalmanFilter:
-  """Enhanced Kalman filter with smoothing for stable decision making."""
+def clip01(value: float) -> float:
+  return max(0.0, min(1.0, float(value)))
 
-  def __init__(self, initial_value=0, measurement_noise=0.1, process_noise=0.01,
-               alpha=1.0, smoothing_factor=0.85):
-    self.x = initial_value
-    self.P = 1.0
-    self.R = measurement_noise
-    self.Q = process_noise
-    self.alpha = alpha
-    self.smoothing_factor = smoothing_factor
-    self.initialized = False
-    self.history = []
-    self.max_history = 10
-    self.confidence = 0.0
 
-  def add_data(self, measurement):
-    if len(self.history) >= self.max_history:
-      self.history.pop(0)
-    self.history.append(measurement)
+class SmoothedSignal:
+  def __init__(self, rise_rate: float, fall_rate: float, initial_value: float = 0.0):
+    self.rise_rate = clip01(rise_rate)
+    self.fall_rate = clip01(fall_rate)
+    self.value = clip01(initial_value)
 
-    if not self.initialized:
-      self.x = measurement
-      self.initialized = True
-      self.confidence = 0.1
-      return
+  def update(self, measurement: float) -> float:
+    measurement = clip01(measurement)
+    rate = self.rise_rate if measurement > self.value else self.fall_rate
+    self.value += (measurement - self.value) * rate
+    return self.value
 
-    self.P = self.alpha * self.P + self.Q
+  def reset(self, value: float = 0.0) -> None:
+    self.value = clip01(value)
 
-    K = self.P / (self.P + self.R)
-    effective_K = K * (1.0 - self.smoothing_factor) + self.smoothing_factor * 0.1
 
-    innovation = measurement - self.x
-    self.x = self.x + effective_K * innovation
-    self.P = (1 - effective_K) * self.P
+class HysteresisSignal:
+  def __init__(self, enter_threshold: float, exit_threshold: float, rise_rate: float, fall_rate: float):
+    self.enter_threshold = clip01(enter_threshold)
+    self.exit_threshold = clip01(exit_threshold)
+    self.filter = SmoothedSignal(rise_rate, fall_rate)
+    self.active = False
 
-    if abs(innovation) < 0.1:
-      self.confidence = min(1.0, self.confidence + 0.05)
-    else:
-      self.confidence = max(0.1, self.confidence - 0.02)
+  def update(self, measurement: float) -> bool:
+    value = self.filter.update(measurement)
+    threshold = self.exit_threshold if self.active else self.enter_threshold
+    self.active = value > threshold
+    return self.active
 
-  def get_value(self):
-    return self.x if self.initialized else None
+  def reset(self) -> None:
+    self.filter.reset()
+    self.active = False
 
-  def get_confidence(self):
-    return self.confidence
-
-  def reset_data(self):
-    self.initialized = False
-    self.history = []
-    self.confidence = 0.0
+  @property
+  def value(self) -> float:
+    return self.filter.value
 
 
 class ModeTransitionManager:
-  """Manages smooth transitions between driving modes with hysteresis."""
-
   def __init__(self):
     self.current_mode: ModeType = 'acc'
-    self.mode_confidence = {'acc': 1.0, 'blended': 0.0}
-    self.transition_timeout = 0
-    self.min_mode_duration = 10
     self.mode_duration = 0
-    self.emergency_override = False
+    self._pending_mode: ModeType = 'acc'
+    self._pending_count = 0
+    self._blended_hold_frames = 0
 
-  def request_mode(self, mode: ModeType, confidence: float = 1.0, emergency: bool = False):
-    # Emergency override for critical situations (stops, collisions)
-    if emergency:
-      self.emergency_override = True
-      self.current_mode = mode
-      self.transition_timeout = SET_MODE_TIMEOUT
-      self.mode_duration = 0
+  def request_mode(self, mode: ModeType, immediate: bool = False, hold_frames: int = 0, cancel_hold: bool = False) -> None:
+    if immediate:
+      self._blended_hold_frames = max(self._blended_hold_frames, hold_frames) if mode == 'blended' else 0
+      self._pending_mode = mode
+      self._pending_count = 0
+      self._switch_mode(mode)
       return
 
-    self.mode_confidence[mode] = min(1.0, self.mode_confidence[mode] + 0.1 * confidence)
-    for m in self.mode_confidence:
-      if m != mode:
-        self.mode_confidence[m] = max(0.0, self.mode_confidence[m] - 0.05)
+    if cancel_hold and mode == 'acc':
+      self._blended_hold_frames = 0
 
-    # Require minimum duration in current mode (unless emergency)
-    if self.mode_duration < self.min_mode_duration and not self.emergency_override:
+    if self._blended_hold_frames > 0:
+      mode = 'blended'
+
+    if mode == self.current_mode:
+      self._pending_mode = mode
+      self._pending_count = 0
       return
 
-    # Hysteresis: higher threshold for mode changes
-    confidence_threshold = 0.6 if mode != self.current_mode else 0.3  # Lower threshold for faster response
+    if mode != self._pending_mode:
+      self._pending_mode = mode
+      self._pending_count = 1
+    else:
+      self._pending_count += 1
 
-    if self.mode_confidence[mode] > confidence_threshold:
-      if mode != self.current_mode and self.transition_timeout == 0:
-        self.transition_timeout = SET_MODE_TIMEOUT
-        self.current_mode = mode
-        self.mode_duration = 0
+    if self.mode_duration < WMACConstants.MIN_MODE_DURATION[self.current_mode]:
+      return
 
-  def update(self):
-    if self.transition_timeout > 0:
-      self.transition_timeout -= 1
+    required_count = WMACConstants.ENTER_BLENDED_FRAMES if mode == 'blended' else WMACConstants.EXIT_BLENDED_FRAMES
+    if self._pending_count >= required_count:
+      self._switch_mode(mode)
+
+  def update(self) -> None:
+    if self._blended_hold_frames > 0:
+      self._blended_hold_frames -= 1
     self.mode_duration += 1
 
-    # Reset emergency override after some time
-    if self.emergency_override and self.mode_duration > 20:
-      self.emergency_override = False
-
-    # Gradual confidence decay
-    for mode in self.mode_confidence:
-      self.mode_confidence[mode] *= 0.98
-
   def get_mode(self) -> ModeType:
     return self.current_mode
 
+  def _switch_mode(self, mode: ModeType) -> None:
+    if mode == self.current_mode:
+      return
+
+    self.current_mode = mode
+    self.mode_duration = 0
+    self._pending_mode = mode
+    self._pending_count = 0
+
 
 class DynamicExperimentalController:
   def __init__(self, CP: structs.CarParams, mpc, params=None):
@@ -142,35 +129,33 @@ class DynamicExperimentalController:
 
     self._mode_manager = ModeTransitionManager()
 
-    # Smooth filters for stable decision making with faster response for critical scenarios
-    self._lead_filter = SmoothKalmanFilter(
-      measurement_noise=0.15,
-      process_noise=0.05,
-      alpha=1.02,
-      smoothing_factor=0.8
+    self._lead_tracker = HysteresisSignal(
+      enter_threshold=WMACConstants.LEAD_PROB,
+      exit_threshold=WMACConstants.LEAD_EXIT_PROB,
+      rise_rate=WMACConstants.LEAD_RISE_RATE,
+      fall_rate=WMACConstants.LEAD_FALL_RATE,
+    )
+    self._radar_acc_lead_tracker = HysteresisSignal(
+      enter_threshold=WMACConstants.RADAR_LEAD_ACC_PROB,
+      exit_threshold=WMACConstants.RADAR_LEAD_ACC_EXIT_PROB,
+      rise_rate=WMACConstants.RADAR_LEAD_ACC_RISE_RATE,
+      fall_rate=WMACConstants.RADAR_LEAD_ACC_FALL_RATE,
+    )
+    self._slow_down_tracker = HysteresisSignal(
+      enter_threshold=WMACConstants.SLOW_DOWN_PROB,
+      exit_threshold=WMACConstants.SLOW_DOWN_EXIT_PROB,
+      rise_rate=WMACConstants.SLOW_DOWN_RISE_RATE,
+      fall_rate=WMACConstants.SLOW_DOWN_FALL_RATE,
+    )
+    self._slowness_tracker = HysteresisSignal(
+      enter_threshold=WMACConstants.SLOWNESS_PROB,
+      exit_threshold=WMACConstants.SLOWNESS_EXIT_PROB,
+      rise_rate=WMACConstants.SLOWNESS_RISE_RATE,
+      fall_rate=WMACConstants.SLOWNESS_FALL_RATE,
     )
 
-    self._slow_down_filter = SmoothKalmanFilter(
-      measurement_noise=0.1,
-      process_noise=0.1,
-      alpha=1.05,
-      smoothing_factor=0.7
-    )
-
-    self._slowness_filter = SmoothKalmanFilter(
-      measurement_noise=0.1,
-      process_noise=0.06,
-      alpha=1.015,
-      smoothing_factor=0.92
-    )
-
-    self._mpc_fcw_filter = SmoothKalmanFilter(
-      measurement_noise=0.2,
-      process_noise=0.1,
-      alpha=1.1,
-      smoothing_factor=0.5
-    )
     self._has_lead_filtered = False
+    self._has_radar_acc_lead = False
     self._has_slow_down = False
     self._has_slowness = False
     self._has_mpc_fcw = False
@@ -179,13 +164,14 @@ class DynamicExperimentalController:
     self._has_standstill = False
     self._mpc_fcw_crash_cnt = 0
     self._standstill_count = 0
-    # debug
+
     self._endpoint_x = float('inf')
     self._expected_distance = 0.0
     self._trajectory_valid = False
+    self._raw_urgency = 0.0
 
   def _read_params(self) -> None:
-    if self._frame % int(1. / DT_MDL) == 0:
+    if self._frame % WMACConstants.PARAM_READ_FRAMES == 0:
       self._enabled = self._params.get_bool("DynamicExperimentalControl")
 
   def mode(self) -> str:
@@ -198,191 +184,115 @@ class DynamicExperimentalController:
     return self._active
 
   def set_mpc_fcw_crash_cnt(self) -> None:
-    """Set MPC FCW crash count"""
     self._mpc_fcw_crash_cnt = self._mpc.crash_cnt
 
   def _update_calculations(self, sm: messaging.SubMaster) -> None:
     car_state = sm['carState']
-    lead_one = sm['radarState'].leadOne
+    radar_state = sm['radarState']
+    lead_one = radar_state.leadOne
+    lead_two = radar_state.leadTwo
     md = sm['modelV2']
 
     self._v_ego_kph = car_state.vEgo * 3.6
     self._v_cruise_kph = car_state.vCruise
     self._has_standstill = car_state.standstill
 
-    # standstill detection
     if self._has_standstill:
-      self._standstill_count = min(20, self._standstill_count + 1)
+      self._standstill_count = min(WMACConstants.STANDSTILL_FRAMES * 3, self._standstill_count + 1)
     else:
       self._standstill_count = max(0, self._standstill_count - 1)
 
-    # Lead detection
-    self._lead_filter.add_data(float(lead_one.status))
-    lead_value = self._lead_filter.get_value() or 0.0
-    self._has_lead_filtered = lead_value > WMACConstants.LEAD_PROB
-
-    # MPC FCW detection
-    fcw_filtered_value = self._mpc_fcw_filter.get_value() or 0.0
-    self._mpc_fcw_filter.add_data(float(self._mpc_fcw_crash_cnt > 0))
-    self._has_mpc_fcw = fcw_filtered_value > 0.5
-
-    # Slow down detection
+    self._has_lead_filtered = self._lead_tracker.update(float(lead_one.status))
+    radar_acc_lead_score = max(self._radar_acc_lead_score(lead_one), self._radar_acc_lead_score(lead_two))
+    self._has_radar_acc_lead = self._radar_acc_lead_tracker.update(radar_acc_lead_score)
+    self._has_mpc_fcw = self._mpc_fcw_crash_cnt > 0
     self._calculate_slow_down(md)
 
-    # Slowness detection
-    if not (self._standstill_count > 5) and not self._has_slow_down:
+    if self._standstill_count > WMACConstants.STANDSTILL_FRAMES or self._has_slow_down:
+      self._slowness_tracker.reset()
+      self._has_slowness = False
+    else:
       current_slowness = float(self._v_ego_kph <= (self._v_cruise_kph * WMACConstants.SLOWNESS_CRUISE_OFFSET))
-      self._slowness_filter.add_data(current_slowness)
-      slowness_value = self._slowness_filter.get_value() or 0.0
+      self._has_slowness = self._slowness_tracker.update(current_slowness)
 
-      # Hysteresis for slowness
-      threshold = WMACConstants.SLOWNESS_PROB * (0.8 if self._has_slowness else 1.1)
-      self._has_slowness = slowness_value > threshold
-
-  def _calculate_slow_down(self, md):
-    """Calculate urgency based on trajectory endpoint vs expected distance."""
-
-    # Reset to safe defaults
-    urgency = 0.0
+  def _calculate_slow_down(self, md) -> None:
     self._endpoint_x = float('inf')
+    self._expected_distance = 0.0
     self._trajectory_valid = False
 
-    #Require exact trajectory size
-    position_valid = len(md.position.x) == TRAJECTORY_SIZE
-    orientation_valid = len(md.orientation.x) == TRAJECTORY_SIZE
+    urgency = self._model_action_urgency(md)
+    position_valid = len(md.position.x) == WMACConstants.TRAJECTORY_SIZE
 
-    if not (position_valid and orientation_valid):
-      # Invalid trajectory - this itself might indicate a stop scenario
-      # Apply moderate urgency for incomplete trajectories at speed
-      if self._v_ego_kph > 20.0:
-        urgency = 0.3
+    if position_valid:
+      self._trajectory_valid = True
+      self._endpoint_x = md.position.x[WMACConstants.TRAJECTORY_SIZE - 1]
+      self._expected_distance = interp(self._v_ego_kph, WMACConstants.SLOW_DOWN_BP, WMACConstants.SLOW_DOWN_DIST)
+      urgency = max(urgency, self._endpoint_urgency(self._endpoint_x, self._expected_distance))
 
-      self._slow_down_filter.add_data(urgency)
-      urgency_filtered = self._slow_down_filter.get_value() or 0.0
-      self._has_slow_down = urgency_filtered > WMACConstants.SLOW_DOWN_PROB
-      self._urgency = urgency_filtered
-      return
+    self._raw_urgency = clip01(urgency)
+    self._has_slow_down = self._slow_down_tracker.update(self._raw_urgency)
+    self._urgency = self._slow_down_tracker.value
 
-    # We have a valid full trajectory
-    self._trajectory_valid = True
+  def _radar_acc_lead_score(self, lead_one) -> float:
+    radar_track_id = int(getattr(lead_one, 'radarTrackId', -1))
+    return float(lead_one.status and (bool(getattr(lead_one, 'radar', False)) or radar_track_id >= 0))
 
-    # Use the exact endpoint (33rd point, index 32)
-    endpoint_x = md.position.x[TRAJECTORY_SIZE - 1]
-    self._endpoint_x = endpoint_x
+  def _model_action_urgency(self, md) -> float:
+    action = getattr(md, 'action', None)
+    if action is None:
+      return 0.0
 
-    # Get expected distance based on current speed using tuned constants
-    expected_distance = interp(self._v_ego_kph,
-                               WMACConstants.SLOW_DOWN_BP,
-                               WMACConstants.SLOW_DOWN_DIST)
-    self._expected_distance = expected_distance
+    urgency = 1.0 if getattr(action, 'shouldStop', False) else 0.0
+    desired_accel = getattr(action, 'desiredAcceleration', 0.0)
+    if desired_accel < WMACConstants.MODEL_DECEL_START:
+      urgency = max(urgency, min(1.0, (WMACConstants.MODEL_DECEL_START - desired_accel) / WMACConstants.MODEL_DECEL_RANGE))
+    return urgency
 
-    # Calculate urgency based on trajectory shortage
-    if endpoint_x < expected_distance:
-      shortage = expected_distance - endpoint_x
-      shortage_ratio = shortage / expected_distance
+  def _endpoint_urgency(self, endpoint_x: float, expected_distance: float) -> float:
+    if endpoint_x >= expected_distance:
+      return 0.0
 
-      # Base urgency on shortage ratio
-      urgency = min(1.0, shortage_ratio * 2.0)
+    shortage_ratio = (expected_distance - endpoint_x) / expected_distance
+    urgency = min(1.0, shortage_ratio * WMACConstants.ENDPOINT_URGENCY_GAIN)
 
-      # Increase urgency for very short trajectories (imminent stops)
-      critical_distance = expected_distance * 0.3
-      if endpoint_x < critical_distance:
-        urgency = min(1.0, urgency * 2.0)
+    if endpoint_x < expected_distance * WMACConstants.CRITICAL_ENDPOINT_FACTOR:
+      urgency = min(1.0, urgency * WMACConstants.CRITICAL_URGENCY_GAIN)
 
-      # Speed-based urgency adjustment
-      if self._v_ego_kph > 25.0:
-        speed_factor = 1.0 + (self._v_ego_kph - 25.0) / 80.0
-        urgency = min(1.0, urgency * speed_factor)
+    if self._v_ego_kph > WMACConstants.SPEED_URGENCY_MIN:
+      speed_factor = 1.0 + (self._v_ego_kph - WMACConstants.SPEED_URGENCY_MIN) / WMACConstants.SPEED_URGENCY_RANGE
+      urgency = min(1.0, urgency * speed_factor)
 
-    # Apply filtering but with less smoothing for stops
-    self._slow_down_filter.add_data(urgency)
-    urgency_filtered = self._slow_down_filter.get_value() or 0.0
+    return urgency
 
-    # Update state with lower threshold for better stop detection
-    self._has_slow_down = urgency_filtered > (WMACConstants.SLOW_DOWN_PROB * 0.8)
-    self._urgency = urgency_filtered
+  def _desired_mode(self) -> tuple[ModeType, bool]:
+    if not self._CP.radarUnavailable and self._has_radar_acc_lead:
+      return 'acc', True
 
-  def _radarless_mode(self) -> None:
-    """Radarless mode decision logic with emergency handling."""
-
-    # EMERGENCY: MPC FCW - immediate blended mode
     if self._has_mpc_fcw:
-      self._mode_manager.request_mode('blended', confidence=1.0, emergency=True)
-      return
+      return 'blended', True
 
-    # Standstill: use blended
-    if self._standstill_count > 3:
-      self._mode_manager.request_mode('blended', confidence=0.9)
-      return
+    standstill = self._standstill_count > WMACConstants.STANDSTILL_FRAMES
+    urgent_slow_down = self._has_slow_down and self._raw_urgency > WMACConstants.URGENT_SLOW_DOWN_PROB
 
-    # Slow down scenarios: emergency for high urgency, normal for lower urgency
-    if self._has_slow_down:
-      if self._urgency > 0.7:
-        # Emergency: immediate blended mode for high urgency stops
-        self._mode_manager.request_mode('blended', confidence=1.0, emergency=True)
-      else:
-        # Normal: blended with urgency-based confidence
-        confidence = min(1.0, self._urgency * 1.5)
-        self._mode_manager.request_mode('blended', confidence=confidence)
-      return
+    if self._CP.radarUnavailable:
+      if standstill or self._has_slow_down:
+        return 'blended', urgent_slow_down
+      return 'acc', False
 
-    # Driving slow: use ACC (but not if actively slowing down)
-    if self._has_slowness and not self._has_slow_down:
-      self._mode_manager.request_mode('acc', confidence=0.8)
-      return
+    if standstill or self._has_slow_down:
+      return 'blended', urgent_slow_down
 
-    # Default: ACC
-    self._mode_manager.request_mode('acc', confidence=0.7)
-
-  def _radar_mode(self) -> None:
-    """Radar mode with emergency handling."""
-
-    # EMERGENCY: MPC FCW - immediate blended mode
-    if self._has_mpc_fcw:
-      self._mode_manager.request_mode('blended', confidence=1.0, emergency=True)
-      return
-
-    # If lead detected and not in standstill: always use ACC
-    if self._has_lead_filtered and not (self._standstill_count > 3):
-      self._mode_manager.request_mode('acc', confidence=1.0)
-      return
-
-    # Slow down scenarios: emergency for high urgency, normal for lower urgency
-    if self._has_slow_down:
-      if self._urgency > 0.7:
-        # Emergency: immediate blended mode for high urgency stops
-        self._mode_manager.request_mode('blended', confidence=1.0, emergency=True)
-      else:
-        # Normal: blended with urgency-based confidence
-        confidence = min(1.0, self._urgency * 1.3)
-        self._mode_manager.request_mode('blended', confidence=confidence)
-      return
-
-    # Standstill: use blended
-    if self._standstill_count > 3:
-      self._mode_manager.request_mode('blended', confidence=0.9)
-      return
-
-    # Driving slow: use ACC (but not if actively slowing down)
-    if self._has_slowness and not self._has_slow_down:
-      self._mode_manager.request_mode('acc', confidence=0.8)
-      return
-
-    # Default: ACC
-    self._mode_manager.request_mode('acc', confidence=0.7)
+    return 'acc', False
 
   def update(self, sm: messaging.SubMaster) -> None:
     self._read_params()
-
     self.set_mpc_fcw_crash_cnt()
-
     self._update_calculations(sm)
 
-    if self._CP.radarUnavailable:
-      self._radarless_mode()
-    else:
-      self._radar_mode()
-
+    mode, immediate = self._desired_mode()
+    self._mode_manager.request_mode(mode, immediate=immediate, hold_frames=WMACConstants.EMERGENCY_HOLD_FRAMES,
+                                    cancel_hold=not self._CP.radarUnavailable and self._has_radar_acc_lead)
     self._mode_manager.update()
+
     self._active = sm['selfdriveState'].experimentalMode and self._enabled
     self._frame += 1

sunnypilot/selfdrive/controls/lib/dec/tests/pytest_dynamic_controller.py

@@ -1,94 +0,0 @@
-import pytest
-
-from openpilot.sunnypilot.selfdrive.controls.lib.dec.dec import DynamicExperimentalController
-
-class MockLeadOne:
-  def __init__(self, status=0.0):
-    self.status = status
-
-class MockRadarState:
-  def __init__(self, status=0.0):
-    self.leadOne = MockLeadOne(status=status)
-
-class MockCarState:
-  def __init__(self, vEgo=0.0, vCruise=0.0, standstill=False):
-    self.vEgo = vEgo
-    self.vCruise = vCruise
-    self.standstill = standstill
-
-class MockModelData:
-  def __init__(self, valid=True):
-    size = 33 if valid else 10  # incomplete if invalid
-    self.position = type("Pos", (), {"x": [0.0] * size})()
-    self.orientation = type("Ori", (), {"x": [0.0] * size})()
-
-class MockSelfDriveState:
-  def __init__(self, experimentalMode=False):
-    self.experimentalMode = experimentalMode
-
-class MockParams:
-  def get_bool(self, name):
-    return True
-
-@pytest.fixture
-def default_sm():
-  sm = {
-    'carState': MockCarState(vEgo=10.0, vCruise=20.0),
-    'radarState': MockRadarState(status=1.0),
-    'modelV2': MockModelData(valid=True),
-    'selfdriveState': MockSelfDriveState(experimentalMode=True),
-  }
-  return sm
-
-@pytest.fixture
-def mock_cp():
-  class CP:
-    radarUnavailable = False
-  return CP()
-
-@pytest.fixture
-def mock_mpc():
-  class MPC:
-    crash_cnt = 0
-  return MPC()
-
-# Fake Kalman Filter that always returns a given value
-class FakeKalman:
-  def __init__(self, value=1.0):
-    self.value = value
-  def add_data(self, v): pass
-  def get_value(self): return self.value
-  def get_confidence(self): return 1.0
-  def reset_data(self): pass
-
-def test_initial_mode_is_acc(mock_cp, mock_mpc):
-  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
-  assert controller.mode() == "acc"
-
-def test_standstill_triggers_blended(mock_cp, mock_mpc, default_sm):
-  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
-  default_sm['carState'].standstill = True
-  for _ in range(10):
-    controller.update(default_sm)
-  assert controller.mode() == "blended"
-
-def test_emergency_blended_on_fcw(mock_cp, mock_mpc, default_sm):
-  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
-  mock_mpc.crash_cnt = 1  # simulate FCW
-  for _ in range(2):
-    controller.update(default_sm)
-  assert controller.mode() == "blended"
-
-def test_radarless_slowdown_triggers_blended(mock_cp, mock_mpc, default_sm):
-  mock_cp.radarUnavailable = True
-  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
-
-  # Force conditions to simulate slowdown
-  controller._slow_down_filter = FakeKalman(value=1.0)  # Ensure urgency triggers slowdown
-  controller._v_ego_kph = 35.0
-  default_sm['modelV2'] = MockModelData(valid=False)  # Incomplete trajectory
-
-  for _ in range(3):
-    controller.update(default_sm)
-
-  assert controller.mode() == "blended"

sunnypilot/selfdrive/controls/lib/dec/tests/test_dynamic_controller.py

@@ -0,0 +1,313 @@
+import pytest
+
+from openpilot.sunnypilot.selfdrive.controls.lib.dec.dec import DynamicExperimentalController, HysteresisSignal
+
+
+class MockLeadOne:
+  def __init__(self, status=0.0, dRel=30.0, vRel=0.0, radar=False, radarTrackId=-1):
+    self.status = status
+    self.dRel = dRel
+    self.vRel = vRel
+    self.radar = radar
+    self.radarTrackId = radarTrackId
+
+
+class MockRadarState:
+  def __init__(self, status=0.0, dRel=30.0, vRel=0.0, radar=False, radarTrackId=-1, leadTwo=None):
+    self.leadOne = MockLeadOne(status=status, dRel=dRel, vRel=vRel, radar=radar, radarTrackId=radarTrackId)
+    self.leadTwo = leadTwo if leadTwo is not None else MockLeadOne()
+
+
+class MockCarState:
+  def __init__(self, vEgo=0.0, vCruise=0.0, standstill=False):
+    self.vEgo = vEgo
+    self.vCruise = vCruise
+    self.standstill = standstill
+
+
+class MockAction:
+  def __init__(self, desiredAcceleration=0.0, shouldStop=False):
+    self.desiredAcceleration = desiredAcceleration
+    self.shouldStop = shouldStop
+
+
+class MockModelData:
+  def __init__(self, valid=True, endpoint_x=200.0, orientation_valid=None, desired_acceleration=0.0, should_stop=False):
+    position_size = 33 if valid else 10
+    orientation_size = position_size if orientation_valid is None else (33 if orientation_valid else 10)
+    position_x = [0.0] * position_size
+    if position_x:
+      position_x[-1] = endpoint_x
+    self.position = type("Pos", (), {"x": position_x})()
+    self.orientation = type("Ori", (), {"x": [0.0] * orientation_size})()
+    self.acceleration = type("Accel", (), {"x": [0.0] * position_size})()
+    self.action = MockAction(desired_acceleration, should_stop)
+
+
+class MockSelfDriveState:
+  def __init__(self, experimentalMode=False):
+    self.experimentalMode = experimentalMode
+
+
+class MockParams:
+  def get_bool(self, name):
+    return True
+
+
+@pytest.fixture
+def default_sm():
+  sm = {
+    'carState': MockCarState(vEgo=10.0, vCruise=20.0),
+    'radarState': MockRadarState(status=1.0, radar=True, radarTrackId=7),
+    'modelV2': MockModelData(valid=True),
+    'selfdriveState': MockSelfDriveState(experimentalMode=True),
+  }
+  return sm
+
+
+@pytest.fixture
+def mock_cp():
+  class CP:
+    radarUnavailable = False
+  return CP()
+
+
+@pytest.fixture
+def mock_mpc():
+  class MPC:
+    crash_cnt = 0
+  return MPC()
+
+
+def test_initial_mode_is_acc(mock_cp, mock_mpc):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  assert controller.mode() == "acc"
+
+
+def test_standstill_triggers_blended(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=0.0)
+  default_sm['carState'].standstill = True
+  for _ in range(20):
+    controller.update(default_sm)
+  assert controller.mode() == "blended"
+
+
+def test_emergency_blended_on_fcw(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=0.0)
+  mock_mpc.crash_cnt = 1
+  controller.update(default_sm)
+  assert controller.mode() == "blended"
+
+
+def test_radarless_slowdown_triggers_blended(mock_cp, mock_mpc, default_sm):
+  mock_cp.radarUnavailable = True
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=0.0)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0)
+
+  controller.update(default_sm)
+
+  assert controller.mode() == "blended"
+
+
+def test_valid_position_with_missing_orientation_can_trigger_slowdown(mock_cp, mock_mpc, default_sm):
+  mock_cp.radarUnavailable = True
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=0.0)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0, orientation_valid=False)
+
+  controller.update(default_sm)
+
+  assert controller._trajectory_valid
+  assert controller.mode() == "blended"
+
+
+def test_incomplete_position_does_not_trigger_slowdown(mock_cp, mock_mpc, default_sm):
+  mock_cp.radarUnavailable = True
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=0.0)
+  default_sm['modelV2'] = MockModelData(valid=False, endpoint_x=0.0)
+
+  for _ in range(3):
+    controller.update(default_sm)
+
+  assert not controller._trajectory_valid
+  assert not controller._has_slow_down
+  assert controller.mode() == "acc"
+
+
+def test_slowdown_hysteresis_prevents_threshold_chatter():
+  signal = HysteresisSignal(enter_threshold=0.5, exit_threshold=0.4, rise_rate=1.0, fall_rate=1.0)
+
+  assert signal.update(0.55)
+  assert signal.update(0.45)
+  assert not signal.update(0.35)
+
+
+def test_model_should_stop_triggers_blended_without_valid_trajectory(mock_cp, mock_mpc, default_sm):
+  mock_cp.radarUnavailable = True
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=0.0)
+  default_sm['modelV2'] = MockModelData(valid=False, should_stop=True)
+
+  controller.update(default_sm)
+
+  assert not controller._trajectory_valid
+  assert controller.mode() == "blended"
+
+
+def test_radar_lead_keeps_acc_over_model_slowdown(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=1.0, radar=True, radarTrackId=7)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0)
+
+  for _ in range(3):
+    controller.update(default_sm)
+
+  assert controller._has_slow_down
+  assert controller._has_radar_acc_lead
+  assert controller.mode() == "acc"
+
+
+def test_far_radar_lead_always_uses_acc(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=1.0, dRel=120.0, vRel=0.0, radar=True)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0)
+
+  controller.update(default_sm)
+
+  assert controller._has_lead_filtered
+  assert controller._has_radar_acc_lead
+  assert controller.mode() == "acc"
+
+
+def test_radar_acquisition_immediately_returns_blended_to_acc(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=0.0)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0)
+  controller.update(default_sm)
+  assert controller.mode() == "blended"
+
+  default_sm['radarState'] = MockRadarState(status=1.0, dRel=120.0, radar=True, radarTrackId=7)
+  controller.update(default_sm)
+
+  assert controller._has_radar_acc_lead
+  assert controller.mode() == "acc"
+
+  default_sm['radarState'] = MockRadarState(status=0.0)
+  default_sm['modelV2'] = MockModelData(valid=True)
+  for _ in range(20):
+    controller.update(default_sm)
+  assert controller.mode() == "acc"
+
+
+def test_close_vision_only_lead_can_use_blended(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=1.0, dRel=30.0, vRel=-5.0)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0)
+  controller.update(default_sm)
+
+  assert not controller._has_radar_acc_lead
+  assert controller.mode() == "blended"
+
+
+def test_second_radar_lead_forces_acc(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  lead_two = MockLeadOne(status=1.0, dRel=120.0, radar=True, radarTrackId=8)
+  default_sm['radarState'] = MockRadarState(status=1.0, dRel=30.0, vRel=-5.0, leadTwo=lead_two)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0)
+  controller.update(default_sm)
+
+  assert controller._has_radar_acc_lead
+  assert controller.mode() == "acc"
+
+
+def test_second_vision_only_lead_does_not_force_acc(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  lead_two = MockLeadOne(status=1.0, dRel=20.0, vRel=-10.0)
+  default_sm['radarState'] = MockRadarState(status=0.0, leadTwo=lead_two)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0)
+  controller.update(default_sm)
+
+  assert not controller._has_radar_acc_lead
+  assert controller.mode() == "blended"
+
+
+def test_inactive_lead_with_radar_marker_does_not_force_acc(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=0.0, radar=True, radarTrackId=7)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0)
+  controller.update(default_sm)
+
+  assert not controller._has_radar_acc_lead
+  assert controller.mode() == "blended"
+
+
+def test_radarless_car_ignores_marked_radar_track(mock_cp, mock_mpc, default_sm):
+  mock_cp.radarUnavailable = True
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=1.0, radar=True, radarTrackId=7)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0)
+  controller.update(default_sm)
+
+  assert controller._has_radar_acc_lead
+  assert controller.mode() == "blended"
+
+
+def test_closing_far_radar_lead_returns_to_acc(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=1.0, dRel=120.0, vRel=-25.0, radarTrackId=7)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0)
+
+  for _ in range(20):
+    controller.update(default_sm)
+
+  assert controller._has_radar_acc_lead
+  assert controller.mode() == "acc"
+
+
+def test_radar_lead_keeps_acc_over_fcw_and_standstill(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=1.0, radar=True, radarTrackId=7)
+  default_sm['carState'].standstill = True
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0, should_stop=True)
+  mock_mpc.crash_cnt = 1
+
+  for _ in range(10):
+    controller.update(default_sm)
+
+  assert controller._has_lead_filtered
+  assert controller._has_mpc_fcw
+  assert controller.mode() == "acc"
+
+
+def test_lead_flicker_hold_prevents_one_frame_mode_flip(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=1.0, radar=True, radarTrackId=7)
+  controller.update(default_sm)
+
+  default_sm['radarState'] = MockRadarState(status=0.0)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0)
+  controller.update(default_sm)
+
+  assert controller._has_lead_filtered
+  assert controller.mode() == "acc"
+
+
+def test_radar_lead_dropout_guard_expires(mock_cp, mock_mpc, default_sm):
+  controller = DynamicExperimentalController(mock_cp, mock_mpc, params=MockParams())
+  default_sm['radarState'] = MockRadarState(status=1.0, radar=True, radarTrackId=7)
+  controller.update(default_sm)
+
+  default_sm['radarState'] = MockRadarState(status=0.0)
+  default_sm['modelV2'] = MockModelData(valid=True, endpoint_x=0.0)
+  for _ in range(3):
+    controller.update(default_sm)
+    assert controller._has_radar_acc_lead
+    assert controller.mode() == "acc"
+
+  controller.update(default_sm)
+  assert not controller._has_radar_acc_lead
+  assert controller.mode() == "blended"

sunnypilot/selfdrive/controls/lib/longitudinal_planner.py

@@ -5,10 +5,19 @@ This file is part of sunnypilot and is licensed under the MIT License.
 See the LICENSE.md file in the root directory for more details.
 """
 
+import numpy as np
+
 from cereal import messaging, custom
 from opendbc.car import structs
+from opendbc.car.interfaces import ACCEL_MIN, ACCEL_MAX
 from openpilot.common.constants import CV
+from openpilot.common.params import Params
+from openpilot.common.realtime import DT_MDL
 from openpilot.selfdrive.car.cruise import V_CRUISE_MAX
+from openpilot.selfdrive.controls.lib.longitudinal_mpc_lib.long_mpc import N, T_IDXS
+from openpilot.sunnypilot import get_sanitize_int_param
+from openpilot.sunnypilot.selfdrive.controls.lib.accel_personality import AccelController, AccelControllerState, AccelProfile
+from openpilot.sunnypilot.selfdrive.controls.lib.accel_personality.constants import MPC_SEED_RISE_RATE
 from openpilot.sunnypilot.selfdrive.controls.lib.dec.dec import DynamicExperimentalController
 from openpilot.sunnypilot.selfdrive.controls.lib.e2e_alerts_helper import E2EAlertsHelper
 from openpilot.sunnypilot.selfdrive.controls.lib.smart_cruise_control.smart_cruise_control import SmartCruiseControl
@@ -22,7 +31,8 @@ LongitudinalPlanSource = custom.LongitudinalPlanSP.LongitudinalPlanSource
 
 
 class LongitudinalPlannerSP:
-  def __init__(self, CP: structs.CarParams, CP_SP: structs.CarParamsSP, mpc):
+  def __init__(self, CP: structs.CarParams, CP_SP: structs.CarParamsSP, mpc, dt: float = DT_MDL):
+    self.params = Params()
     self.events_sp = EventsSP()
     self.resolver = SpeedLimitResolver()
     self.dec = DynamicExperimentalController(CP, mpc)
@@ -32,10 +42,28 @@ class LongitudinalPlannerSP:
     self.generation = int(model_bundle.generation) if (model_bundle := get_active_bundle()) else None
     self.source = LongitudinalPlanSource.cruise
     self.e2e_alerts_helper = E2EAlertsHelper()
+    self.accel_controller = AccelController(CP, dt=dt)
+    self.accel_controller_result = None
+    self.accel_controller_fault_latched = False
+    self._previous_is_e2e = False
+
+    self._param_read_frames = max(1, int(round(0.25 / dt)))
+    self._param_frame = 0
+    self.accel_personality_enabled = False
+    self.accel_personality = int(AccelProfile.normal)
 
     self.output_v_target = 0.
     self.output_a_target = 0.
 
+  def _read_accel_controller_params(self) -> None:
+    if self._param_frame % self._param_read_frames == 0:
+      self.accel_personality_enabled = self.params.get_bool("AccelPersonalityEnabled")
+      self.accel_personality = get_sanitize_int_param(
+        "AccelPersonality", int(AccelProfile.eco), int(AccelProfile.sport), self.params,
+      )
+
+    self._param_frame += 1
+
   def is_e2e(self, sm: messaging.SubMaster) -> bool:
     experimental_mode = sm['selfdriveState'].experimentalMode
     if not self.dec.active():
@@ -73,7 +101,108 @@ class LongitudinalPlannerSP:
     self.output_v_target, self.output_a_target = targets[self.source]
     return self.output_v_target, self.output_a_target
 
+  @staticmethod
+  def _radar_fresh(sm: messaging.SubMaster) -> bool:
+    try:
+      return bool(sm.updated['radarState'] and sm.valid['radarState'] and sm.alive['radarState'])
+    except (AttributeError, KeyError, TypeError):
+      return True
+
+  def update_accel_controller(self, sm: messaging.SubMaster, base_speed: float, engaged: bool, cruise_initialized: bool,
+                              acc_selected: bool, planner_accel: float, action_accel: float, stock_accel_max: float,
+                              previous_should_stop: bool, controller_fault: bool = False) -> float:
+    self.accel_controller_result = self.accel_controller.update(
+      sm['radarState'], base_speed=base_speed, v_ego=sm['carState'].vEgo, a_ego=sm['carState'].aEgo,
+      profile=self.accel_personality, follow_personality=sm['selfdriveState'].personality,
+      enabled=self.accel_personality_enabled, acc_selected=acc_selected, engaged=engaged, cruise_initialized=cruise_initialized,
+      planner_accel=planner_accel, action_accel=action_accel, stock_accel_max=stock_accel_max,
+      previous_should_stop=previous_should_stop, controller_fault=controller_fault, radar_fresh=self._radar_fresh(sm),
+    )
+    return self.accel_controller_result.target_speed
+
+  def _run_mpc(self, sm: messaging.SubMaster, v_cruise: float, prev_accel_constraint: bool, accel_max=None, *, seed=False,
+               seed_target=None, seed_rise_rate=MPC_SEED_RISE_RATE, retry_state=None, current_accel=None) -> None:
+    if retry_state is not None:
+      self.mpc.a_prev = retry_state[0].copy()
+      self.mpc.crash_cnt = retry_state[1]
+    self.mpc.set_weights(prev_accel_constraint, personality=sm['selfdriveState'].personality)
+    mpc_accel = self.a_desired if current_accel is None else float(np.clip(current_accel, ACCEL_MIN, ACCEL_MAX))
+    self.mpc.set_cur_state(self.v_desired_filter.x, mpc_accel)
+    if seed or seed_target is not None:
+      self._seed_mpc_current_state(seed_target, seed_rise_rate)
+    self.mpc.update(sm['radarState'], v_cruise, personality=sm['selfdriveState'].personality, accel_max=accel_max)
+
+  def _seed_mpc_current_state(self, accel_target=None, rise_rate=MPC_SEED_RISE_RATE) -> None:
+    target = float(np.clip(self.mpc.x0[2] if accel_target is None else accel_target, ACCEL_MIN, ACCEL_MAX))
+    desired_accel = target * np.ones(N + 1) if accel_target is None else np.minimum(self.mpc.x0[2] + rise_rate * T_IDXS, target)
+    acceleration = np.zeros(N + 1)
+    velocity = np.zeros(N + 1)
+    position = np.zeros(N + 1)
+    jerk = np.zeros(N)
+    acceleration[0] = self.mpc.x0[2]
+    velocity[0] = max(self.mpc.x0[1], 0.0)
+    position[0] = self.mpc.x0[0]
+    for idx in range(1, N + 1):
+      dt = T_IDXS[idx] - T_IDXS[idx - 1]
+      min_accel = 0.0 if velocity[idx - 1] <= 1e-3 and acceleration[idx - 1] < 0.0 else -2.0 * velocity[idx - 1] / dt - acceleration[idx - 1]
+      acceleration[idx] = np.clip(max(desired_accel[idx], min_accel), ACCEL_MIN, ACCEL_MAX)
+      jerk[idx - 1] = (acceleration[idx] - acceleration[idx - 1]) / dt
+      position[idx] = max(position[idx - 1], position[idx - 1] + velocity[idx - 1] * dt + 0.5 * acceleration[idx - 1] * dt**2
+                          + jerk[idx - 1] * dt**3 / 6.0)
+      velocity[idx] = max(0.0, velocity[idx - 1] + 0.5 * (acceleration[idx - 1] + acceleration[idx]) * dt)
+    for idx in range(N + 1):
+      self.mpc.solver.set(idx, 'x', np.array([position[idx], velocity[idx], acceleration[idx]]))
+    for idx in range(N):
+      self.mpc.solver.set(idx, 'u', np.array([jerk[idx]]))
+
+  def update_accel_controller_mpc(self, sm: messaging.SubMaster, base_v_cruise: float, mpc_v_cruise: float,
+                                  prev_accel_constraint: bool, *, reset_state: bool, cruise_initialized: bool,
+                                  planner_accel: float, previous_output_accel: float, available_accel_max: float,
+                                  previous_should_stop: bool, force_decel: bool):
+    is_e2e = self.is_e2e(sm)
+    was_e2e = self._previous_is_e2e
+    if reset_state or not self.accel_personality_enabled:
+      self.accel_controller_fault_latched = False
+
+    self.update_accel_controller(
+      sm, base_v_cruise, engaged=not reset_state and not force_decel, cruise_initialized=cruise_initialized,
+      acc_selected=not is_e2e, planner_accel=planner_accel, action_accel=previous_output_accel,
+      stock_accel_max=available_accel_max, previous_should_stop=previous_should_stop,
+      controller_fault=self.accel_controller_fault_latched,
+    )
+    result = self.accel_controller_result
+    handoff_context = result.enabled and result.shadow_active and not force_decel and not self.accel_controller_fault_latched
+    transition_from_e2e = handoff_context and was_e2e and not is_e2e and result.active
+    handoff_accel = (min(planner_accel, previous_output_accel)
+                     if transition_from_e2e and result.active and np.isfinite(previous_output_accel) else None)
+    self._previous_is_e2e = is_e2e and handoff_context
+    controller_actuating = result.active and not result.stock_mode and not force_decel
+    accel_max = result.mpc_accel_max if controller_actuating else None
+    free_profile_limit = controller_actuating and result.state == AccelControllerState.free and result.effective_accel_max > 0.0
+    seed_target = result.effective_accel_max if free_profile_limit and handoff_accel is None else None
+    custom_mpc = handoff_accel is not None or (controller_actuating and (accel_max is not None or seed_target is not None))
+    retry_state = (self.mpc.a_prev.copy(), self.mpc.crash_cnt)
+    controller_v_cruise = min(mpc_v_cruise, result.target_speed)
+    self._run_mpc(sm, controller_v_cruise, prev_accel_constraint, accel_max, seed_target=seed_target, current_accel=handoff_accel)
+
+    finite_solution = all(np.all(np.isfinite(solution)) for solution in (self.mpc.v_solution, self.mpc.a_solution, self.mpc.j_solution))
+    custom_failed = custom_mpc and (self.mpc.last_solution_status != 0 or not finite_solution)
+    if custom_failed:
+      self.accel_controller_fault_latched = True
+      self.accel_controller.reset()
+      self._run_mpc(sm, mpc_v_cruise, prev_accel_constraint, seed=True, retry_state=retry_state)
+      self.update_accel_controller(
+        sm, base_v_cruise, engaged=not reset_state and not force_decel, cruise_initialized=cruise_initialized,
+        acc_selected=not is_e2e, planner_accel=planner_accel, action_accel=previous_output_accel,
+        stock_accel_max=available_accel_max, previous_should_stop=previous_should_stop, controller_fault=True,
+      )
+    if custom_failed and self.mpc.last_solution_status != 0:
+      self.mpc.a_prev, self.mpc.crash_cnt = retry_state
+
+    return is_e2e
+
   def update(self, sm: messaging.SubMaster) -> None:
+    self._read_accel_controller_params()
     self.events_sp.clear()
     self.dec.update(sm)
     self.e2e_alerts_helper.update(sm, self.events_sp)
@@ -95,6 +224,25 @@ class LongitudinalPlannerSP:
     dec.enabled = self.dec.enabled()
     dec.active = self.dec.active()
 
+    if self.accel_controller_result is not None:
+      result = self.accel_controller_result
+      accel_controller = longitudinalPlanSP.accelController
+      accel_controller.enabled = result.enabled
+      accel_controller.active = result.active
+      accel_controller.shadowOnly = result.shadow_active and not result.active
+      accel_controller.profile = int(result.profile)
+      accel_controller.state = int(result.state if result.active else result.shadow_state)
+      accel_controller.vTargetBase = float(result.base_speed)
+      accel_controller.vTargetRaw = float(result.raw_energy_cap)
+      accel_controller.vTargetFiltered = float(result.live_filtered_cap)
+      accel_controller.vTargetShadow = float(result.shadow_filtered_cap)
+      accel_controller.leadIndex = result.selected_lead
+      accel_controller.usableGap = float(result.usable_gap)
+      accel_controller.closingSpeed = float(result.closing_speed)
+      accel_controller.requiredDecel = float(result.required_decel)
+      accel_controller.aMaxProfile = float(result.profile_accel_max)
+      accel_controller.aMaxEffective = float(result.effective_accel_max)
+
     # Smart Cruise Control
     smartCruiseControl = longitudinalPlanSP.smartCruiseControl
     # Vision Control

sunnypilot/selfdrive/controls/lib/relc.py

@@ -0,0 +1,84 @@
+"""
+Copyright (c) 2021-, rav4kumar, Haibin Wen, sunnypilot, and a number of other contributors.
+
+This file is part of sunnypilot and is licensed under the MIT License.
+See the LICENSE.md file in the root directory for more details.
+"""
+import numpy as np
+
+from openpilot.common.constants import CV
+from openpilot.common.realtime import DT_MDL
+from openpilot.common.params import Params
+
+NEARSIDE_PROB = 0.2
+EDGE_PROB = 0.35
+EDGE_REACTION_TIME = 1.0
+EDGE_CLEAR_TIME = 0.3
+MIN_SPEED = 20 * CV.MPH_TO_MS
+
+
+class RoadEdgeLaneChangeController:
+  def __init__(self, desire_helper):
+    self.DH = desire_helper
+    self.params = Params()
+    self.enabled = self.params.get_bool("RoadEdgeLaneChangeEnabled")
+    self.param_read_counter = 0
+    self.left_edge_detected = False
+    self.right_edge_detected = False
+    self.left_edge_timer = 0.0
+    self.right_edge_timer = 0.0
+    self.left_clear_timer = 0.0
+    self.right_clear_timer = 0.0
+
+  def read_params(self) -> None:
+    self.enabled = self.params.get_bool("RoadEdgeLaneChangeEnabled")
+
+  def update_params(self) -> None:
+    if self.param_read_counter % 50 == 0:
+      self.read_params()
+    self.param_read_counter += 1
+
+  def reset(self) -> None:
+    self.left_edge_detected = False
+    self.right_edge_detected = False
+    self.left_edge_timer = 0.0
+    self.right_edge_timer = 0.0
+    self.left_clear_timer = 0.0
+    self.right_clear_timer = 0.0
+
+  def update(self, road_edge_stds, lane_line_probs, v_ego: float) -> None:
+    self.update_params()
+
+    if not self.enabled or v_ego < MIN_SPEED:
+      self.reset()
+      return
+
+    left_edge_prob = np.clip(1.0 - road_edge_stds[0], 0.0, 1.0)
+    right_edge_prob = np.clip(1.0 - road_edge_stds[1], 0.0, 1.0)
+    left_lane_prob = lane_line_probs[0]
+    right_lane_prob = lane_line_probs[3]
+
+    left_cond = left_edge_prob > EDGE_PROB and left_lane_prob < NEARSIDE_PROB and right_lane_prob >= left_lane_prob
+    right_cond = right_edge_prob > EDGE_PROB and right_lane_prob < NEARSIDE_PROB and left_lane_prob >= right_lane_prob
+
+    if left_cond:
+      self.left_edge_timer = min(self.left_edge_timer + DT_MDL, EDGE_REACTION_TIME + EDGE_CLEAR_TIME)
+      self.left_clear_timer = 0.0
+      if self.left_edge_timer > EDGE_REACTION_TIME:
+        self.left_edge_detected = True
+    else:
+      self.left_clear_timer += DT_MDL
+      if self.left_clear_timer > EDGE_CLEAR_TIME:
+        self.left_edge_timer = 0.0
+        self.left_edge_detected = False
+
+    if right_cond:
+      self.right_edge_timer = min(self.right_edge_timer + DT_MDL, EDGE_REACTION_TIME + EDGE_CLEAR_TIME)
+      self.right_clear_timer = 0.0
+      if self.right_edge_timer > EDGE_REACTION_TIME:
+        self.right_edge_detected = True
+    else:
+      self.right_clear_timer += DT_MDL
+      if self.right_clear_timer > EDGE_CLEAR_TIME:
+        self.right_edge_timer = 0.0
+        self.right_edge_detected = False

sunnypilot/selfdrive/controls/lib/tests/test_accel_controller_closed_loop.py

@@ -0,0 +1,990 @@
+from collections.abc import Callable
+from dataclasses import dataclass
+import gc
+
+import numpy as np
+import pytest
+
+from opendbc.car.interfaces import ACCEL_MAX, ACCEL_MIN
+from openpilot.common.realtime import DT_MDL
+from openpilot.selfdrive.controls.lib.longitudinal_mpc_lib.long_mpc import STOP_DISTANCE, get_T_FOLLOW
+from openpilot.selfdrive.controls.lib.longitudinal_planner import get_max_accel
+from openpilot.selfdrive.test.longitudinal_maneuvers.plant import PRIUS_TSS2_ROUTE_MODEL, LeadObservation, Plant
+from openpilot.sunnypilot.selfdrive.controls.lib.accel_personality import AccelControllerState, AccelProfile
+
+ROUTINE_GAP_TOLERANCE = 0.10
+
+
+@dataclass
+class ClosedLoopTrace:
+  time: np.ndarray
+  speed: np.ndarray
+  distance: np.ndarray
+  distance_lead: np.ndarray
+  a_target: np.ndarray
+  acceleration: np.ndarray
+  should_stop: np.ndarray
+  fcw: np.ndarray
+  source: list
+  active: np.ndarray
+  shadow_active: np.ndarray
+  launching: np.ndarray
+  target_speed: np.ndarray
+  stock_mode: np.ndarray
+  raw_cap: np.ndarray
+  filtered_cap: np.ndarray
+  selected_lead: np.ndarray
+  profile_accel_max: np.ndarray
+  effective_accel_max: np.ndarray
+  state: np.ndarray
+  required_decel: np.ndarray
+  controller_fault: np.ndarray
+  controller_fault_latched: np.ndarray
+  mpc_accel_max: np.ndarray
+  actuator_command: np.ndarray
+  solver_status: np.ndarray
+  solver_failures: int
+  solver_failure_times: list[float]
+
+
+def _configure_plant(plant: Plant, *, enabled: bool, profile: int = 1, dec_enabled: bool = False) -> None:
+  plant.planner.accel_personality_enabled = enabled
+  plant.planner.accel_personality = profile
+  plant.planner._read_accel_controller_params = lambda: None
+  plant.planner.dec._enabled = dec_enabled
+  plant.planner.dec._read_params = lambda: None
+
+
+def _run(
+  *,
+  duration: float,
+  controller_enabled: bool,
+  profile: int = 1,
+  v_lead: float | Callable[[float], float] = 0.0,
+  v_cruise: float = 30.0,
+  dec_enabled: bool = False,
+  **plant_kwargs,
+) -> ClosedLoopTrace:
+  gc.collect()
+  plant = Plant(**plant_kwargs)
+  _configure_plant(plant, enabled=controller_enabled, profile=profile, dec_enabled=dec_enabled)
+  plant.v_lead_prev = float(v_lead) if isinstance(v_lead, (int, float)) else float(v_lead(0.0))
+  solver_failures = 0
+  solver_failure_times = []
+  original_mpc_reset = plant.planner.mpc.reset
+
+  def count_failed_solve() -> None:
+    nonlocal solver_failures
+    if plant.planner.mpc.solution_status != 0:
+      solver_failures += 1
+      solver_failure_times.append(plant.current_time)
+    original_mpc_reset()
+
+  plant.planner.mpc.reset = count_failed_solve
+  rows = []
+  sources = []
+  while plant.current_time < duration:
+    lead_speed = float(v_lead) if isinstance(v_lead, (int, float)) else v_lead(plant.current_time)
+    controller_fault = plant.planner.mpc.last_solution_status != 0
+    result = plant.step(v_lead=lead_speed, v_cruise=v_cruise)
+    controller = plant.planner.accel_controller_result
+    rows.append(
+      (
+        plant.current_time,
+        result["speed"],
+        result["distance"],
+        result["distance_lead"],
+        result["a_target"],
+        result["realized_acceleration"],
+        result["should_stop"],
+        result["fcw"],
+        controller.active,
+        controller.shadow_active,
+        controller.launching,
+        controller.target_speed,
+        controller.stock_mode,
+        controller.raw_energy_cap,
+        controller.live_filtered_cap,
+        controller.selected_lead,
+        controller.profile_accel_max,
+        controller.effective_accel_max,
+        controller.state,
+        controller.required_decel,
+        controller_fault,
+        plant.planner.accel_controller_fault_latched,
+        min(controller.mpc_accel_max) if controller.mpc_accel_max is not None else np.nan,
+        result["actuator_command"],
+        plant.planner.mpc.last_solution_status,
+      )
+    )
+    sources.append(result["mpc_source"])
+
+  data = np.asarray(rows, dtype=float)
+  trace = ClosedLoopTrace(
+    time=data[:, 0],
+    speed=data[:, 1],
+    distance=data[:, 2],
+    distance_lead=data[:, 3],
+    a_target=data[:, 4],
+    acceleration=data[:, 5],
+    should_stop=data[:, 6].astype(bool),
+    fcw=data[:, 7].astype(bool),
+    source=sources,
+    active=data[:, 8].astype(bool),
+    shadow_active=data[:, 9].astype(bool),
+    launching=data[:, 10].astype(bool),
+    target_speed=data[:, 11],
+    stock_mode=data[:, 12].astype(bool),
+    raw_cap=data[:, 13],
+    filtered_cap=data[:, 14],
+    selected_lead=data[:, 15].astype(int),
+    profile_accel_max=data[:, 16],
+    effective_accel_max=data[:, 17],
+    state=data[:, 18].astype(int),
+    required_decel=data[:, 19],
+    controller_fault=data[:, 20].astype(bool),
+    controller_fault_latched=data[:, 21].astype(bool),
+    mpc_accel_max=data[:, 22],
+    actuator_command=data[:, 23],
+    solver_status=data[:, 24].astype(int),
+    solver_failures=solver_failures,
+    solver_failure_times=solver_failure_times,
+  )
+  plant.planner.mpc.reset = original_mpc_reset
+  gc.collect()
+  return trace
+
+
+def _first_time_below(trace: ClosedLoopTrace, threshold: float) -> float:
+  indices = np.flatnonzero(trace.a_target <= threshold)
+  assert len(indices), f"never reached {threshold} m/s²"
+  return float(trace.time[indices[0]])
+
+
+def _sustained_time_below(trace: ClosedLoopTrace, threshold: float, *, after: float = 0.5, duration: float = 0.5) -> float:
+  required_frames = round(duration / DT_MDL)
+  below = (trace.time >= after) & (trace.a_target <= threshold)
+  sustained = np.convolve(below.astype(int), np.ones(required_frames, dtype=int), mode="valid") == required_frames
+  indices = np.flatnonzero(sustained)
+  assert len(indices), f"never sustained {threshold} m/s² for {duration} s"
+  return float(trace.time[indices[0]])
+
+
+def _command_jerk(trace: ClosedLoopTrace, after: float = 0.0) -> np.ndarray:
+  indices = np.flatnonzero(trace.time >= after)
+  assert len(indices) >= 2
+  return np.diff(trace.a_target[indices]) / DT_MDL
+
+
+def _filtered_realized_jerk(trace: ClosedLoopTrace, after: float = 1.0, min_speed: float = 0.0) -> np.ndarray:
+  filtered_acceleration = np.convolve(trace.acceleration, np.ones(3) / 3.0, mode="valid")
+  samples = (trace.time[2:-1] >= after) & (trace.speed[2:-1] >= min_speed)
+  return (np.diff(filtered_acceleration) / DT_MDL)[samples]
+
+
+def _has_brake_coast_brake(values: np.ndarray, brake: float = -0.8, coast: float = -0.35, frames: int = 2) -> bool:
+  phase = 0
+  for index in range(len(values) - frames + 1):
+    window = values[index : index + frames]
+    if np.all(window <= brake):
+      if phase == 2:
+        return True
+      phase = 1
+    elif phase == 1 and np.all(window >= coast):
+      phase = 2
+  return False
+
+
+def _has_propulsion_after_braking(values: np.ndarray, propulsion: float = 0.2, brake: float = -0.2, frames: int = 2) -> bool:
+  braking = False
+  for index in range(len(values) - frames + 1):
+    window = values[index : index + frames]
+    if np.all(window <= brake):
+      braking = True
+    elif braking and np.all(window >= propulsion):
+      return True
+  return False
+
+
+def _has_propulsion_brake_cycle(values: np.ndarray, propulsion: float = 0.2, brake: float = -0.2, frames: int = 2) -> bool:
+  phases = []
+  for index in range(len(values) - frames + 1):
+    window = values[index : index + frames]
+    phase = 1 if np.all(window >= propulsion) else -1 if np.all(window <= brake) else 0
+    if phase and (not phases or phase != phases[-1]):
+      phases.append(phase)
+      if len(phases) >= 3 and phases[-1] == phases[-3]:
+        return True
+  return False
+
+
+@pytest.mark.parametrize(
+  ("plant_kwargs", "expect_shadow"),
+  [
+    ({"enabled": False, "lead_relevancy": True, "speed": 20.0, "distance_lead": 70.0}, False),
+    ({"e2e": True, "lead_relevancy": False, "speed": 20.0}, True),
+  ],
+  ids=("disengaged", "e2e-shadow"),
+)
+def test_non_actuating_modes_match_clean_base(plant_kwargs, expect_shadow):
+  common = dict(duration=2.0, v_lead=14.0, **plant_kwargs)
+  baseline = _run(controller_enabled=False, **common)
+  trace = _run(controller_enabled=True, **common)
+
+  np.testing.assert_allclose(trace.a_target, baseline.a_target, atol=1e-6, rtol=0.0)
+  np.testing.assert_array_equal(trace.should_stop, baseline.should_stop)
+  np.testing.assert_array_equal(trace.fcw, baseline.fcw)
+  assert trace.source == baseline.source
+  assert not trace.active.any()
+  np.testing.assert_array_equal(trace.shadow_active, np.full_like(trace.active, expect_shadow))
+
+
+def test_disabled_profiles_match_clean_base():
+  common = dict(duration=2.0, controller_enabled=False, lead_relevancy=True, speed=20.0, distance_lead=70.0, v_lead=14.0)
+  traces = [_run(profile=profile, **common) for profile in range(3)]
+  for trace in traces[1:]:
+    np.testing.assert_allclose(trace.a_target, traces[0].a_target, atol=1e-6, rtol=0.0)
+    np.testing.assert_array_equal(trace.should_stop, traces[0].should_stop)
+    assert trace.source == traces[0].source
+  assert all(np.isinf(trace.effective_accel_max).all() for trace in traces)
+
+
+@pytest.mark.parametrize("lead_relevancy", (False, True), ids=("clear-road", "lead"))
+def test_force_decel_matches_controller_off(lead_relevancy):
+  common = dict(duration=2.0, force_decel=True, lead_relevancy=lead_relevancy, speed=20.0,
+                distance_lead=70.0, v_lead=14.0, profile=0)
+  baseline = _run(controller_enabled=False, **common)
+  trace = _run(controller_enabled=True, **common)
+  np.testing.assert_allclose(trace.a_target, baseline.a_target, atol=1e-6, rtol=0.0)
+  np.testing.assert_array_equal(trace.should_stop, baseline.should_stop)
+  np.testing.assert_array_equal(trace.fcw, baseline.fcw)
+  assert trace.source == baseline.source
+
+
+def test_e2e_to_radar_acc_handoff_keeps_braking_continuous():
+  plant = Plant(
+    lead_relevancy=True, speed=10.0, distance_lead=30.0, actuator_delay=0.15, actuator_lag=0.20,
+    model_action_fn=lambda current_time, _v_ego, _a_ego: (-1.0 if current_time < 2.0 else 0.0, False),
+  )
+  _configure_plant(plant, enabled=True)
+  rows = []
+  while plant.current_time < 2.4:
+    plant.e2e = plant.current_time < 2.0
+    result = plant.step(v_lead=8.0, v_cruise=20.0)
+    rows.append((plant.current_time, result["a_target"], plant.planner.mpc.last_solution_status,
+                 plant.planner.accel_controller_result.active))
+
+  time_values, acceleration, solver_status, active = np.asarray(rows, dtype=float).T
+  transition = np.flatnonzero(time_values > 2.0)[0]
+  assert acceleration[transition] - acceleration[transition - 1] < 0.15
+  assert np.max(np.diff(acceleration[transition - 1:]) / DT_MDL) < 3.0
+  assert not solver_status[transition:].any()
+  assert active[transition]
+
+
+def test_active_controller_is_pre_mpc_and_preserves_stock_lead_authority():
+  plant = Plant(lead_relevancy=False, speed=0.0, actuator_delay=0.15, actuator_lag=0.20)
+  _configure_plant(plant, enabled=True, profile=0)
+  result = plant.step(v_cruise=15.0)
+  controller = plant.planner.accel_controller_result
+
+  assert controller.mpc_accel_max is not None
+  np.testing.assert_allclose(plant.planner.mpc.params[:, 1], controller.mpc_accel_max)
+  assert np.all((plant.planner.mpc.params[:, 1] >= 0.0) & (plant.planner.mpc.params[:, 1] <= ACCEL_MAX))
+  assert ACCEL_MIN <= result["a_target"] <= get_max_accel(plant.speed)
+  for _ in range(100):
+    result = plant.step(v_cruise=15.0)
+    if plant.speed >= 0.30:
+      break
+  assert plant.speed >= 0.30
+  controller = plant.planner.accel_controller_result
+  assert controller.mpc_accel_max is not None
+  np.testing.assert_allclose(plant.planner.mpc.params[:, 1], controller.mpc_accel_max)
+  assert np.all((plant.planner.mpc.params[:, 1] >= 0.0) & (plant.planner.mpc.params[:, 1] <= ACCEL_MAX))
+
+  lead_plant = Plant(lead_relevancy=True, speed=0.0, distance_lead=6.0, actuator_delay=0.15, actuator_lag=0.20)
+  _configure_plant(lead_plant, enabled=True, profile=0)
+  lead_plant.step(v_lead=0.0, v_cruise=15.0)
+  controller = lead_plant.planner.accel_controller_result
+  assert controller.target_speed == 0.0
+  np.testing.assert_array_equal(controller.mpc_accel_max, 0.0)
+  np.testing.assert_array_equal(lead_plant.planner.mpc.params[:, 1], 0.0)
+
+
+def test_clear_road_launch_is_immediate_and_profiles_separate():
+  common = dict(
+    duration=6.0,
+    controller_enabled=True,
+    lead_relevancy=False,
+    speed=0.0,
+    v_cruise=15.0,
+    actuator_delay=0.15,
+    actuator_lag=0.20,
+  )
+  traces = [_run(profile=profile, **common) for profile in range(3)]
+
+  for trace in traces:
+    positive = np.flatnonzero(trace.a_target > 0.05)
+    moving = np.flatnonzero(trace.speed > 0.01)
+    assert len(positive) and trace.time[positive[0]] <= 4 * DT_MDL
+    assert len(moving) and trace.time[moving[0]] <= 1.0
+    assert np.all(trace.effective_accel_max[trace.active] > 0.0)
+    assert not np.any(trace.a_target < -0.05)
+    assert trace.solver_failures == 0
+
+  onset_times = [float(trace.time[np.flatnonzero(trace.a_target > 0.05)[0]]) for trace in traces]
+  assert max(onset_times) - min(onset_times) <= DT_MDL
+
+
+def test_profiles_have_distinct_moving_speed_preshape():
+  traces = [
+    _run(
+      duration=18.0,
+      controller_enabled=True,
+      profile=profile,
+      lead_relevancy=False,
+      speed=0.0,
+      v_cruise=30.0,
+      actuator_delay=0.15,
+      actuator_lag=0.20,
+    )
+    for profile in range(3)
+  ]
+  samples = [np.flatnonzero(trace.speed >= 10.0)[0] for trace in traces]
+  configured = [float(trace.profile_accel_max[index]) for trace, index in zip(traces, samples, strict=True)]
+  effective = [float(trace.effective_accel_max[index]) for trace, index in zip(traces, samples, strict=True)]
+  assert configured[0] < configured[1] < configured[2]
+  assert effective[0] < effective[1] < effective[2]
+  speed_grid = np.linspace(5.0, 16.0, 45)
+  moving_acceleration = [np.interp(speed_grid, trace.speed, trace.a_target) for trace in traces]
+  assert np.all(moving_acceleration[1] - moving_acceleration[0] > 0.10)
+  assert np.all(moving_acceleration[2] - moving_acceleration[1] > 0.05)
+  assert all(trace.solver_failures == 0 for trace in traces)
+
+
+def test_runtime_profile_switch_is_distinct_and_smooth():
+  plant = Plant(lead_relevancy=False, speed=0.0, actuator_delay=0.15, actuator_lag=0.20)
+  _configure_plant(plant, enabled=True, profile=AccelProfile.sport)
+  while plant.speed < 10.0 and plant.current_time < 15.0:
+    plant.step(v_cruise=30.0)
+  assert plant.speed >= 10.0
+  switch_start = plant.current_time
+  rows = []
+  while plant.current_time < switch_start + 5.0:
+    elapsed = plant.current_time - switch_start
+    profile = AccelProfile.sport if elapsed < 1.0 or elapsed >= 3.0 else AccelProfile.eco
+    plant.planner.accel_personality = profile
+    result = plant.step(v_cruise=30.0)
+    controller = plant.planner.accel_controller_result
+    rows.append((plant.current_time - switch_start, profile, result["a_target"], controller.effective_accel_max,
+                 plant.planner.mpc.last_solution_status, plant.planner.accel_controller_fault_latched))
+
+  data = np.asarray(rows, dtype=float)
+  time_values, profiles, acceleration, effective_max, solver_status, fault_latched = data.T
+  settled_eco = (profiles == AccelProfile.eco) & (time_values >= 2.0) & (time_values < 3.0)
+  settled_sport = (profiles == AccelProfile.sport) & (time_values >= 4.0)
+  assert np.max(effective_max[settled_eco]) < np.min(effective_max[settled_sport])
+  assert np.mean(acceleration[settled_eco]) + 0.15 < np.mean(acceleration[settled_sport])
+  switch_window = ((time_values[1:] >= 0.5) & (time_values[1:] <= 1.5)) | ((time_values[1:] >= 2.5) & (time_values[1:] <= 3.5))
+  assert np.max(np.abs(np.diff(acceleration)[switch_window] / DT_MDL)) < 3.0
+  assert np.min(acceleration) >= -0.05
+  assert not solver_status.any()
+  assert not fault_latched.any()
+
+
+def test_clear_road_acceleration_crosses_lut_without_solver_failure():
+  trace = _run(
+    duration=12.0,
+    controller_enabled=True,
+    profile=1,
+    lead_relevancy=False,
+    speed=0.0,
+    v_cruise=22.352,
+    actuator_delay=0.15,
+    actuator_lag=0.20,
+  )
+  assert np.max(trace.speed) > 10.0
+  assert trace.solver_failures == 0
+  assert np.all(trace.effective_accel_max[trace.active] > 0.0)
+
+
+def test_prius_route_model_launches_without_a_dead_pedal():
+  trace = _run(
+    duration=3.0,
+    controller_enabled=True,
+    profile=1,
+    lead_relevancy=False,
+    speed=0.0,
+    v_cruise=22.352,
+    actuator_model=PRIUS_TSS2_ROUTE_MODEL,
+  )
+  positive = np.flatnonzero(trace.a_target > 0.05)
+  moving = np.flatnonzero(trace.speed > 0.05)
+  assert len(positive) and trace.time[positive[0]] <= 4 * DT_MDL
+  assert len(moving) and trace.time[moving[0]] <= 1.0
+  assert trace.solver_failures == 0
+
+
+@pytest.mark.parametrize(
+  ("actuator_delay", "actuator_lag"),
+  [(0.10, 0.20), (0.15, 0.25), (0.20, 0.20), (0.25, 0.30), (0.30, 0.35)],
+  ids=("toyota", "honda", "gm", "hyundai", "ford"),
+)
+def test_stopped_lead_requires_four_departure_frames_and_launches_promptly(actuator_delay, actuator_lag):
+  departure_time = 1.0
+
+  def lead_speed(current_time: float) -> float:
+    return 0.0 if current_time < departure_time else 2.0
+
+  trace = _run(
+    duration=2.5,
+    controller_enabled=True,
+    lead_relevancy=True,
+    speed=0.0,
+    distance_lead=6.0,
+    v_lead=lead_speed,
+    v_cruise=8.0,
+    actuator_delay=actuator_delay,
+    actuator_lag=actuator_lag,
+  )
+
+  first_three = (trace.time > departure_time) & (trace.time <= departure_time + 3 * DT_MDL + 1e-9)
+  assert np.max(trace.speed[first_three]) < 1e-3
+  assert not trace.launching[first_three].any()
+  departure_release = np.flatnonzero((trace.time >= departure_time) & trace.launching)
+  assert len(departure_release) and trace.time[departure_release[0]] >= departure_time + 3 * DT_MDL
+  moving = np.flatnonzero((trace.time >= departure_time) & (trace.speed > 0.05))
+  assert len(moving) and trace.time[moving[0]] <= departure_time + 4 * DT_MDL + 1.0
+  assert np.min(trace.effective_accel_max[departure_release[0] : moving[0] + 1]) > 1.5
+  assert not _has_brake_coast_brake(trace.a_target[trace.time >= departure_time])
+  assert trace.solver_failures == 0
+
+
+def test_creeping_lead_departure_is_prompt_and_does_not_lurch():
+  departure_time = 1.0
+
+  def lead_speed(current_time: float) -> float:
+    if current_time < departure_time:
+      return 0.0
+    if current_time < departure_time + 0.5:
+      return 1.6 * (current_time - departure_time)
+    return min(2.5, 0.8 + 0.7 * (current_time - departure_time - 0.5))
+
+  def observe(_current_time: float, lead_name: str, truth: LeadObservation) -> LeadObservation | None:
+    return None if lead_name == "leadTwo" else truth | {"aLeadK": 0.0, "radarTrackId": 2133, "radar": True}
+
+  common = dict(
+    duration=6.0,
+    profile=0,
+    lead_relevancy=True,
+    speed=0.0,
+    distance_lead=3.6,
+    v_lead=lead_speed,
+    v_cruise=22.352,
+    lead_observation_fn=observe,
+    actuator_delay=0.15,
+    actuator_lag=0.20,
+  )
+  baseline = _run(controller_enabled=False, **common)
+  trace = _run(controller_enabled=True, **common)
+  after_departure = trace.time >= departure_time
+  lead_speeds = np.array([lead_speed(max(0.0, current_time - DT_MDL)) for current_time in trace.time])
+  baseline_moving = np.flatnonzero((baseline.time >= departure_time) & (baseline.speed > 0.05))
+  moving = np.flatnonzero(after_departure & (trace.speed > 0.05))
+  assert len(baseline_moving) and len(moving)
+  assert trace.time[moving[0]] <= baseline.time[baseline_moving[0]]
+  assert np.all(trace.speed[after_departure] <= lead_speeds[after_departure] + 0.20)
+  assert not _has_brake_coast_brake(trace.a_target[after_departure])
+  assert np.min(trace.distance_lead - trace.distance) >= np.min(baseline.distance_lead - baseline.distance) - 1e-3
+  assert trace.solver_failures == 0
+
+
+def test_stop_hold_ignores_two_frame_total_lead_dropout():
+  def observe(current_time: float, _lead_name: str, truth: LeadObservation) -> LeadObservation | None:
+    return None if 1.0 <= current_time < 1.1 else truth
+
+  trace = _run(
+    duration=2.0,
+    controller_enabled=True,
+    lead_relevancy=True,
+    speed=0.0,
+    distance_lead=6.0,
+    v_lead=0.0,
+    v_cruise=8.0,
+    lead_observation_fn=observe,
+    actuator_delay=0.15,
+    actuator_lag=0.20,
+  )
+  assert np.max(trace.speed) < 1e-3
+  assert np.max(trace.effective_accel_max[np.isfinite(trace.effective_accel_max)]) == 0.0
+  assert trace.solver_failures == 0
+
+
+def test_low_speed_stopped_lead_never_accelerates_during_stop_hold():
+  def lead_speed(current_time: float) -> float:
+    return max(0.0, 1.9 - 1.16 * current_time)
+
+  def observe(current_time: float, lead_name: str, truth: LeadObservation) -> LeadObservation | None:
+    if lead_name == "leadTwo":
+      return None
+    moving = lead_speed(current_time) > 0.0
+    return truth | {"vLeadK": truth["vLeadK"] if moving else -0.01, "aLeadK": -1.16 if moving else 0.0, "radarTrackId": 7, "radar": True}
+
+  common = dict(
+    duration=6.0,
+    profile=0,
+    lead_relevancy=True,
+    speed=4.5,
+    distance_lead=18.0,
+    v_lead=lead_speed,
+    v_cruise=23.056,
+    lead_observation_fn=observe,
+    actuator_delay=0.15,
+    actuator_lag=0.20,
+  )
+  baseline = _run(controller_enabled=False, **common)
+  trace = _run(controller_enabled=True, **common)
+
+  urgent_demand = (trace.required_decel >= 0.45) & (trace.speed >= 0.30) & ~trace.should_stop
+  stop_hold = trace.state == int(AccelControllerState.stopHold)
+  assert urgent_demand.any() and stop_hold.any()
+  assert np.max(trace.a_target[urgent_demand]) < 0.0
+  hold_indices = np.flatnonzero(stop_hold)
+  assert np.max(trace.acceleration[stop_hold]) < 0.25
+  assert np.max(trace.speed[stop_hold]) < 0.30
+  assert trace.distance[hold_indices[-1]] - trace.distance[hold_indices[0]] < 0.05
+  assert not _has_brake_coast_brake(trace.a_target[trace.time >= 1.0])
+  assert np.min(trace.a_target) >= np.min(baseline.a_target) - ROUTINE_GAP_TOLERANCE
+  assert np.min(trace.distance_lead - trace.distance) >= np.min(baseline.distance_lead - baseline.distance) - ROUTINE_GAP_TOLERANCE
+  assert not trace.fcw.any()
+  assert trace.solver_failures == 0
+
+
+def test_moving_lead_dropout_and_false_relief_do_not_release_pace():
+  def observe(current_time: float, _lead_name: str, truth: LeadObservation) -> LeadObservation | None:
+    if 2.0 <= current_time < 2.1:
+      return None
+    if 3.0 <= current_time < 3.1:
+      return {"dRel": truth["dRel"] + 5.0}
+    return truth
+
+  common = dict(
+    duration=5.0,
+    lead_relevancy=True,
+    speed=22.0,
+    distance_lead=85.0,
+    v_lead=14.0,
+    lead_observation_fn=observe,
+    actuator_delay=0.20,
+    actuator_lag=0.25,
+  )
+  trace = _run(controller_enabled=True, **common)
+  for start in (2.0, 3.0):
+    before = trace.effective_accel_max[np.flatnonzero(trace.time < start)[-1]]
+    guard = (trace.time >= start) & (trace.time < start + 0.2) & trace.active
+    assert np.all(trace.effective_accel_max[guard] <= before + 0.02)
+  assert not _has_brake_coast_brake(trace.a_target[trace.time >= 1.0])
+  assert not _has_propulsion_after_braking(trace.a_target[trace.time >= 1.0])
+  assert np.max(np.abs(_command_jerk(trace, after=1.0))) < 3.0
+  assert trace.solver_failures == 0
+
+
+@pytest.mark.parametrize(
+  ("actuator_delay", "actuator_lag"),
+  [(0.10, 0.20), (0.15, 0.25), (0.20, 0.20), (0.25, 0.30), (0.30, 0.35)],
+  ids=("toyota", "honda", "gm", "hyundai", "ford"),
+)
+def test_confirmed_finite_relief_transitions_smoothly(actuator_delay, actuator_lag):
+  def lead_speed(current_time: float) -> float:
+    return 8.0 if current_time < 5.0 else min(15.0, 8.0 + 3.5 * (current_time - 5.0))
+
+  common = dict(
+    duration=9.0,
+    profile=1,
+    lead_relevancy=True,
+    speed=12.0,
+    distance_lead=50.0,
+    v_lead=lead_speed,
+    v_cruise=20.0,
+    actuator_delay=actuator_delay,
+    actuator_lag=actuator_lag,
+  )
+  trace = _run(controller_enabled=True, **common)
+
+  released = np.flatnonzero((trace.time >= 5.0) & (trace.state == int(AccelControllerState.free)))
+  assert len(released)
+  reached_profile = np.flatnonzero((np.arange(len(trace.time)) >= released[0]) &
+                                   (trace.effective_accel_max >= trace.profile_accel_max - 1e-6))
+  assert len(reached_profile)
+  rising = (np.arange(len(trace.time)) >= released[0]) & (np.arange(len(trace.time)) <= reached_profile[0])
+  assert rising.any()
+  assert np.all(np.diff(trace.effective_accel_max[rising]) >= -1e-9)
+  assert not _has_brake_coast_brake(trace.a_target[trace.time >= 5.0])
+  assert not _has_propulsion_brake_cycle(trace.a_target[trace.time >= 5.0])
+  assert np.max(np.abs(_command_jerk(trace, after=5.0))) < 3.0
+  assert trace.solver_failures == 0
+
+
+def test_low_speed_far_lead_acquisition_does_not_fault_or_lurch():
+  acquisition_time = 5.0
+
+  def observe(current_time: float, _lead_name: str, truth: LeadObservation) -> LeadObservation | None:
+    return None if current_time < acquisition_time else truth
+
+  common = dict(
+    duration=8.0,
+    profile=0,
+    lead_relevancy=True,
+    speed=0.0,
+    distance_lead=180.0,
+    v_lead=3.0,
+    v_cruise=30.0,
+    lead_observation_fn=observe,
+    actuator_delay=0.15,
+    actuator_lag=0.20,
+  )
+  trace = _run(controller_enabled=True, **common)
+
+  acquired = (trace.time >= acquisition_time) & (trace.selected_lead >= 0)
+  response = trace.time >= acquisition_time
+  jerk_response = trace.time[1:] >= acquisition_time
+  assert acquired.any()
+  assert not trace.controller_fault[response].any()
+  assert not trace.solver_status.any()
+  assert not trace.controller_fault_latched.any()
+  assert trace.solver_failures == 0
+  assert np.max(np.abs(np.diff(trace.a_target)[jerk_response] / DT_MDL)) < 3.0
+  assert not _has_brake_coast_brake(trace.a_target[response])
+  assert not _has_propulsion_after_braking(trace.a_target[response])
+
+
+def test_alternating_range_glitch_has_bounded_jerk_and_no_reversal():
+  glitch_start = 5.0
+  glitch_end = 5.5
+
+  def observe(current_time: float, _lead_name: str, truth: LeadObservation) -> LeadObservation:
+    if glitch_start <= current_time < glitch_end:
+      frame = round(current_time / DT_MDL)
+      return truth | {"dRel": truth["dRel"] + (5.0 if frame % 2 else 0.0)}
+    return truth
+
+  common = dict(
+    duration=10.0,
+    lead_relevancy=True,
+    speed=8.0,
+    distance_lead=20.0,
+    v_lead=1.5,
+    actuator_delay=0.20,
+    actuator_lag=0.25,
+  )
+  control = _run(controller_enabled=True, **common)
+  baseline = _run(controller_enabled=False, lead_observation_fn=observe, **common)
+  trace = _run(controller_enabled=True, lead_observation_fn=observe, **common)
+  window = (trace.time[1:] >= glitch_start) & (trace.time[1:] < glitch_end)
+  assert np.max(np.abs(np.diff(trace.a_target)[window] / DT_MDL)) < 3.0
+  response = (trace.time >= glitch_start) & (trace.time < glitch_end + 1.0)
+  assert np.max(np.abs((trace.a_target - control.a_target)[response])) < 0.07
+  np.testing.assert_array_equal(trace.should_stop[response], baseline.should_stop[response])
+  np.testing.assert_array_equal(trace.fcw[response], baseline.fcw[response])
+  assert not _has_brake_coast_brake(trace.a_target[response])
+  assert not _has_propulsion_after_braking(trace.a_target[response])
+  assert np.min(trace.distance_lead - trace.distance) >= np.min(baseline.distance_lead - baseline.distance) - ROUTINE_GAP_TOLERANCE
+  assert trace.solver_failures == 0
+
+
+@pytest.mark.parametrize(
+  ("actuator_delay", "actuator_lag"),
+  [(0.10, 0.20), (0.15, 0.25), (0.20, 0.20), (0.25, 0.30), (0.30, 0.35)],
+  ids=("toyota", "honda", "gm", "hyundai", "ford"),
+)
+def test_slow_lead_approach_is_smooth_across_actuator_dynamics(actuator_delay, actuator_lag):
+  lead_speed = 10.0
+  trace = _run(
+    duration=70.0,
+    controller_enabled=True,
+    profile=1,
+    lead_relevancy=True,
+    speed=20.0,
+    distance_lead=100.0,
+    v_lead=lead_speed,
+    v_cruise=30.0,
+    actuator_delay=actuator_delay,
+    actuator_lag=actuator_lag,
+  )
+  desired_gap = STOP_DISTANCE + get_T_FOLLOW() * lead_speed
+  gap = trace.distance_lead - trace.distance
+  closing_speed = trace.speed - lead_speed
+  closing = closing_speed > 0.1
+  meaningful_closing = closing_speed > 0.3
+  settled = trace.time >= trace.time[-1] - 3.0
+  moving = (trace.time[1:] >= 0.5) & (trace.speed[1:] >= 2.0) & ~trace.should_stop[1:] & ~trace.should_stop[:-1]
+  assert np.max(np.abs(np.diff(trace.a_target)[moving] / DT_MDL)) < 3.0
+  assert not _has_brake_coast_brake(trace.a_target[trace.time >= 1.0])
+  assert not _has_propulsion_brake_cycle(trace.a_target[trace.time >= 1.0])
+  assert np.max(trace.a_target[meaningful_closing]) <= 0.2
+  assert np.percentile(np.abs(_filtered_realized_jerk(trace)), 95) < 0.35
+  assert np.min(trace.a_target) >= -1.1
+  assert np.min(trace.acceleration) >= -1.1
+  assert np.min(gap) >= desired_gap - 1.6
+  assert np.min(gap[closing] / closing_speed[closing]) >= 2.0
+  assert abs(np.median(trace.speed[settled]) - lead_speed) <= 0.5
+  assert desired_gap - 1.6 <= np.median(gap[settled]) <= desired_gap + 6.0
+  assert not trace.fcw.any()
+  assert not trace.should_stop.any()
+  assert not trace.solver_status.any()
+  assert not trace.controller_fault_latched.any()
+  assert trace.solver_failures == 0
+
+
+def test_decelerating_moving_lead_stays_smooth_and_safe():
+  def lead_speed(current_time: float) -> float:
+    if current_time < 2.0:
+      return 15.0
+    progress = min((current_time - 2.0) / 6.0, 1.0)
+    return 15.0 - 5.0 * (3.0 * progress**2 - 2.0 * progress**3)
+
+  common = dict(
+    duration=14.0,
+    profile=1,
+    lead_relevancy=True,
+    speed=20.0,
+    distance_lead=110.0,
+    v_lead=lead_speed,
+    v_cruise=30.0,
+    actuator_delay=0.20,
+    actuator_lag=0.25,
+  )
+  baseline = _run(controller_enabled=False, **common)
+  trace = _run(controller_enabled=True, **common)
+  lead_decelerating = (trace.time >= 2.0) & (trace.time <= 8.0) & trace.active
+  settled = trace.time >= 8.0
+  assert np.any(trace.effective_accel_max[lead_decelerating] < 0.0)
+  assert not trace.should_stop.any()
+  assert np.max(np.abs(_command_jerk(trace, after=1.0))) < 4.25
+  baseline_p95 = np.percentile(np.abs(_filtered_realized_jerk(baseline)), 95)
+  trace_p95 = np.percentile(np.abs(_filtered_realized_jerk(trace)), 95)
+  assert trace_p95 <= max(0.20, baseline_p95 + 0.02)
+  assert not _has_brake_coast_brake(trace.a_target[trace.time >= 1.0])
+  assert not _has_propulsion_after_braking(trace.a_target[trace.time >= 1.0])
+  assert np.max(trace.a_target[settled]) <= 0.2
+  assert np.min(trace.distance_lead - trace.distance) >= np.min(baseline.distance_lead - baseline.distance) - ROUTINE_GAP_TOLERANCE
+  assert not trace.fcw.any()
+  assert trace.solver_failures == 0
+
+
+def test_severe_closing_never_delays_stock_braking_or_reduces_clearance():
+  common = dict(
+    duration=12.0,
+    lead_relevancy=True,
+    speed=20.0,
+    distance_lead=160.0,
+    v_lead=3.5,
+    actuator_delay=0.20,
+    actuator_lag=0.20,
+  )
+  baseline = _run(controller_enabled=False, **common)
+  trace = _run(controller_enabled=True, **common)
+  for threshold in (-1.0, -2.0):
+    assert _first_time_below(trace, threshold) <= _first_time_below(baseline, threshold) + 1e-9
+  baseline_gap = baseline.distance_lead - baseline.distance
+  controlled_gap = trace.distance_lead - trace.distance
+  assert np.min(controlled_gap) >= np.min(baseline_gap) - 0.02
+  baseline_closing = baseline.speed - 3.5
+  controlled_closing = trace.speed - 3.5
+  baseline_ttc = np.min(baseline_gap[baseline_closing > 0.1] / baseline_closing[baseline_closing > 0.1])
+  controlled_ttc = np.min(controlled_gap[controlled_closing > 0.1] / controlled_closing[controlled_closing > 0.1])
+  assert controlled_ttc >= baseline_ttc - 0.02
+  assert np.min(controlled_gap) > 0.0
+  onset = (trace.time[1:] > 0.5) & (trace.time[1:] < 3.0)
+  assert np.max(np.abs(np.diff(trace.a_target)[onset] / DT_MDL)) < 4.0
+  assert trace.solver_failures == 0
+
+
+@pytest.mark.parametrize(
+  ("actuator_delay", "actuator_lag"),
+  [(0.10, 0.20), (0.15, 0.25), (0.20, 0.20), (0.25, 0.30), (0.30, 0.35)],
+  ids=("toyota", "honda", "gm", "hyundai", "ford"),
+)
+@pytest.mark.parametrize("profile", range(3), ids=("eco", "normal", "sport"))
+def test_far_lead_deceleration_starts_early_and_stays_smooth(profile, actuator_delay, actuator_lag):
+  common = dict(
+    duration=11.0,
+    lead_relevancy=True,
+    speed=25.0,
+    distance_lead=200.0,
+    v_lead=15.0,
+    actuator_delay=actuator_delay,
+    actuator_lag=actuator_lag,
+  )
+  baseline = _run(controller_enabled=False, **common)
+  trace = _run(controller_enabled=True, profile=profile, **common)
+  baseline_onset = _sustained_time_below(baseline, -0.10)
+  trace_onset = _sustained_time_below(trace, -0.10)
+  negative_bound = np.isfinite(trace.mpc_accel_max) & (trace.mpc_accel_max < -0.05)
+  assert negative_bound.any()
+  assert trace.time[np.flatnonzero(negative_bound)[0]] <= baseline_onset - 0.5
+  assert trace_onset <= baseline_onset - 0.5
+  assert trace.acceleration.min() >= baseline.acceleration.min() - 0.1
+  trace_p95 = float(np.percentile(np.abs(_filtered_realized_jerk(trace)), 95))
+  assert trace_p95 < 0.45
+  assert np.max(np.abs(_command_jerk(trace, after=0.5))) < 3.0
+  assert not _has_brake_coast_brake(trace.a_target[trace.time >= 1.0])
+  assert not _has_propulsion_brake_cycle(trace.a_target[trace.time >= 1.0])
+  assert not trace.fcw.any()
+  assert not trace.solver_status.any()
+  assert not trace.controller_fault_latched.any()
+  assert trace.solver_failures == 0
+
+
+def test_far_lead_profile_order_is_monotonic():
+  traces = [
+    _run(
+      duration=6.0,
+      controller_enabled=True,
+      profile=profile,
+      lead_relevancy=True,
+      speed=25.0,
+      distance_lead=200.0,
+      v_lead=15.0,
+      actuator_delay=0.10,
+      actuator_lag=0.20,
+    )
+    for profile in range(3)
+  ]
+  bound_onsets = [
+    float(trace.time[np.flatnonzero(np.isfinite(trace.mpc_accel_max) & (trace.mpc_accel_max < -0.05))[0]])
+    for trace in traces
+  ]
+  decel_onsets = [_sustained_time_below(trace, -0.10) for trace in traces]
+  assert bound_onsets[0] <= bound_onsets[1] <= bound_onsets[2]
+  assert decel_onsets[0] <= decel_onsets[1] <= decel_onsets[2]
+  assert traces[0].raw_cap[0] < traces[1].raw_cap[0] < traces[2].raw_cap[0]
+
+
+def test_prior_stock_solver_status_does_not_disable_clear_road_controller():
+  plant = Plant(speed=0.0, actuator_delay=0.15, actuator_lag=0.20)
+  _configure_plant(plant, enabled=True, profile=1)
+  plant.step(v_cruise=15.0)
+  assert plant.planner.accel_controller_result.active
+  assert plant.planner.mpc.last_solution_status == 0
+
+  plant.planner.mpc.last_solution_status = 3
+  plant.step(v_cruise=15.0)
+  assert plant.planner.mpc.last_solution_status == 0
+  recovered = plant.planner.accel_controller_result
+  assert recovered.active
+  assert not plant.planner.accel_controller_fault_latched
+  assert np.isfinite(recovered.effective_accel_max)
+
+
+def test_prior_stock_solver_status_does_not_disable_lead_controller():
+  plant = Plant(lead_relevancy=True, speed=25.0, distance_lead=200.0, actuator_delay=0.15, actuator_lag=0.20)
+  _configure_plant(plant, enabled=True, profile=1)
+  plant.v_lead_prev = 15.0
+  for _ in range(30):
+    plant.step(v_lead=15.0, v_cruise=30.0)
+
+  assert plant.planner.accel_controller_result.effective_accel_max < 0.0
+  assert plant.planner.mpc.last_solution_status == 0
+
+  plant.planner.mpc.last_solution_status = 3
+  result = plant.step(v_lead=15.0, v_cruise=30.0)
+  assert plant.planner.mpc.last_solution_status == 0
+  assert plant.planner.accel_controller_result.active
+  assert not plant.planner.accel_controller_fault_latched
+  assert result["a_target"] <= 0.2
+
+
+@pytest.mark.parametrize(("profile", "speed", "expects_ceiling"), ((0, 10.0, True), (2, 0.0, False)), ids=("ceiling", "seed-only"))
+def test_failed_custom_solve_restores_stock_state_and_counts_fcw_once(profile, speed, expects_ceiling):
+  plant = Plant(lead_relevancy=False, speed=speed, actuator_delay=0.15, actuator_lag=0.20)
+  _configure_plant(plant, enabled=True, profile=profile)
+  saved_a_prev = np.full_like(plant.planner.mpc.a_prev, -0.25)
+  accepted_a_prev = np.full_like(saved_a_prev, 0.15)
+  plant.planner.mpc.a_prev = saved_a_prev.copy()
+  plant.planner.mpc.crash_cnt = 2.0
+  if not expects_ceiling:
+    plant.planner.accel_controller._build_accel_ceiling = lambda *_args: None
+  calls = []
+
+  def update_mpc(_radar_state, _v_cruise, personality, accel_max=None):
+    calls.append((personality, accel_max))
+    if len(calls) == 1:
+      plant.planner.mpc.last_solution_status = plant.planner.mpc.solution_status = 4
+      plant.planner.mpc.a_prev = np.zeros_like(saved_a_prev)
+      plant.planner.mpc.crash_cnt = 0.0
+    else:
+      np.testing.assert_array_equal(plant.planner.mpc.a_prev, saved_a_prev)
+      assert plant.planner.mpc.crash_cnt == 2.0
+      plant.planner.mpc.last_solution_status = plant.planner.mpc.solution_status = 0
+      plant.planner.mpc.a_prev = accepted_a_prev.copy()
+      plant.planner.mpc.crash_cnt += 1.0
+
+  plant.planner.mpc.update = update_mpc
+  result = plant.step(v_cruise=30.0)
+  assert len(calls) == 2
+  assert (calls[0][1] is not None) == expects_ceiling and calls[1][1] is None
+  assert plant.planner.accel_controller_fault_latched
+  assert not plant.planner.accel_controller_result.active
+  assert plant.planner.mpc.crash_cnt == 3.0
+  np.testing.assert_array_equal(plant.planner.mpc.a_prev, accepted_a_prev)
+  assert result["fcw"] == (speed > 0.0)
+
+
+@pytest.mark.parametrize("mode", ("disabled", "e2e"))
+def test_stock_solver_recovery_is_not_warm_seeded_when_controller_cannot_actuate(mode):
+  plant = Plant(lead_relevancy=False, speed=10.0, actuator_delay=0.15, actuator_lag=0.20, e2e=mode == "e2e")
+  _configure_plant(plant, enabled=mode != "disabled", profile=0)
+  plant.planner.mpc.last_solution_status = 3
+  seeds = []
+  plant.planner._seed_mpc_current_state = lambda _target=None: seeds.append(True)
+  plant.step(v_cruise=30.0)
+  assert not seeds
+
+
+@pytest.mark.parametrize("pre_frames", (1, 2))
+@pytest.mark.parametrize("mode", ("disabled", "e2e"))
+def test_early_launch_transition_returns_to_stock_without_solver_fault(pre_frames, mode):
+  plant = Plant(speed=0.0, actuator_delay=0.15, actuator_lag=0.20)
+  _configure_plant(plant, enabled=True, profile=1)
+  for _ in range(pre_frames):
+    plant.step(v_cruise=15.0)
+
+  if mode == "disabled":
+    plant.planner.accel_personality_enabled = False
+    plant.planner._read_accel_controller_params = lambda: None
+  else:
+    plant.e2e = True
+
+  for _ in range(4):
+    plant.step(v_cruise=15.0)
+    controller = plant.planner.accel_controller_result
+    assert not controller.active
+    assert controller.mpc_accel_max is None
+    assert plant.planner.mpc.last_solution_status == 0
+    np.testing.assert_array_equal(plant.planner.mpc.params[:, 1], ACCEL_MAX)
+
+
+@pytest.mark.parametrize("profile", range(3), ids=("eco", "normal", "sport"))
+@pytest.mark.parametrize("mode", ("disabled", "e2e"))
+def test_launch_transition_after_crossing_standstill_threshold(profile, mode):
+  plant = Plant(speed=0.29, actuator_delay=0.15, actuator_lag=0.20)
+  _configure_plant(plant, enabled=True, profile=profile)
+  plant.acceleration = 0.5
+  plant.planner.a_desired = 0.5
+  plant.step(v_cruise=15.0)
+  assert plant.speed > 0.30
+
+  if mode == "disabled":
+    plant.planner.accel_personality_enabled = False
+    plant.planner._read_accel_controller_params = lambda: None
+  else:
+    plant.e2e = True
+
+  for _ in range(4):
+    plant.step(v_cruise=15.0)
+    controller = plant.planner.accel_controller_result
+    assert not controller.active
+    assert controller.mpc_accel_max is None
+    assert plant.planner.mpc.last_solution_status == 0
+    np.testing.assert_array_equal(plant.planner.mpc.params[:, 1], ACCEL_MAX)

sunnypilot/selfdrive/controls/lib/tests/test_lane_turn_desire.py

@@ -5,6 +5,8 @@ from openpilot.common.params import Params
 from openpilot.selfdrive.controls.lib.desire_helper import DesireHelper
 from openpilot.sunnypilot.selfdrive.controls.lib.lane_turn_desire import LaneTurnController, LANE_CHANGE_SPEED_MIN
 from openpilot.sunnypilot.selfdrive.controls.lib.auto_lane_change import AutoLaneChangeMode
+from openpilot.sunnypilot.selfdrive.controls.lib.relc import RoadEdgeLaneChangeController
+
 
 TurnDirection = custom.ModelDataV2SP.TurnDirection
 
@@ -107,7 +109,11 @@ def set_lane_turn_params():
 ])
 def test_desire_helper_integration(carstate, lateral_active, lane_change_prob, expected_desire, set_lane_turn_params):
   dh = DesireHelper()
+  relc = RoadEdgeLaneChangeController(dh)
+  relc.enabled = True
   dh.alc.lane_change_set_timer = AutoLaneChangeMode.NUDGE
   for _ in range(10):
-    dh.update(carstate, lateral_active, lane_change_prob)
+    dh.update(carstate, lateral_active, lane_change_prob,
+              left_edge_detected=relc.left_edge_detected, right_edge_detected=relc.right_edge_detected)
   assert dh.desire == expected_desire  # The first four tests were unit tests to test the controller, where this tests the integration in desire helpers
+

sunnypilot/selfdrive/controls/lib/tests/test_relc.py

@@ -0,0 +1,99 @@
+"""
+Copyright (c) 2021-, rav4kumar, Haibin Wen, sunnypilot, and a number of other contributors.
+
+This file is part of sunnypilot and is licensed under the MIT License.
+See the LICENSE.md file in the root directory for more details.
+"""
+import pytest
+
+from openpilot.common.realtime import DT_MDL
+from openpilot.selfdrive.controls.lib.desire_helper import DesireHelper
+from openpilot.sunnypilot.selfdrive.controls.lib.relc import (
+  RoadEdgeLaneChangeController, EDGE_REACTION_TIME, EDGE_CLEAR_TIME, MIN_SPEED,
+)
+
+V_HIGH = MIN_SPEED + 2.0
+V_LOW = MIN_SPEED - 1.0
+
+
+@pytest.fixture
+def relc(mocker):
+  mock_params = mocker.patch("openpilot.sunnypilot.selfdrive.controls.lib.relc.Params")
+  mock_params.return_value.get_bool.return_value = True
+  controller = RoadEdgeLaneChangeController(DesireHelper())
+  controller.enabled = True
+  return controller
+
+
+def drive(controller, road_edge_stds, lane_line_probs, seconds, v_ego=V_HIGH):
+  for _ in range(int(seconds / DT_MDL) + 1):
+    controller.update(road_edge_stds, lane_line_probs, v_ego)
+
+
+@pytest.mark.parametrize("road_edge_stds,lane_line_probs,attr", [
+  ([0.0, 0.9], [0.0, 0.8, 0.8, 0.8], "left_edge_detected"),
+  ([0.9, 0.0], [0.8, 0.8, 0.8, 0.0], "right_edge_detected"),
+])
+def test_edge_detection(relc, road_edge_stds, lane_line_probs, attr):
+  drive(relc, road_edge_stds, lane_line_probs, EDGE_REACTION_TIME + 0.1)
+  assert getattr(relc, attr)
+
+
+def test_edge_detection_requires_time(relc):
+  drive(relc, [0.0, 0.9], [0.0, 0.8, 0.8, 0.8], EDGE_REACTION_TIME - 0.05)
+  assert not relc.left_edge_detected
+
+
+def test_both_edges_detected(relc):
+  drive(relc, [0.0, 0.0], [0.0, 0.8, 0.8, 0.0], EDGE_REACTION_TIME + 0.1)
+  assert relc.left_edge_detected
+  assert relc.right_edge_detected
+
+
+def test_noise_doesnt_clear(relc):
+  edge = ([0.0, 0.9], [0.0, 0.8, 0.8, 0.8])
+  clear = ([0.9, 0.9], [0.8, 0.8, 0.8, 0.8])
+
+  drive(relc, *edge, EDGE_REACTION_TIME + 0.1)
+  assert relc.left_edge_detected
+
+  relc.update(*clear, V_HIGH)
+  relc.update(*edge, V_HIGH)
+  assert relc.left_edge_detected
+
+
+def test_clears_after_window(relc):
+  edge = ([0.0, 0.9], [0.0, 0.8, 0.8, 0.8])
+  clear = ([0.9, 0.9], [0.8, 0.8, 0.8, 0.8])
+
+  drive(relc, *edge, EDGE_REACTION_TIME + 0.1)
+  assert relc.left_edge_detected
+
+  drive(relc, *clear, EDGE_CLEAR_TIME + 0.05)
+  assert not relc.left_edge_detected
+  assert relc.left_edge_timer == 0.0
+
+
+def test_low_speed_skips(relc):
+  drive(relc, [0.0, 0.9], [0.0, 0.8, 0.8, 0.8], EDGE_REACTION_TIME + 0.1, v_ego=V_LOW)
+  assert not relc.left_edge_detected
+  assert relc.left_edge_timer == 0.0
+
+
+def test_speed_drop_resets(relc):
+  drive(relc, [0.0, 0.9], [0.0, 0.8, 0.8, 0.8], EDGE_REACTION_TIME + 0.1)
+  assert relc.left_edge_detected
+
+  relc.update([0.0, 0.9], [0.0, 0.8, 0.8, 0.8], V_LOW)
+  assert not relc.left_edge_detected
+
+
+def test_param_off_resets(relc):
+  drive(relc, [0.0, 0.9], [0.0, 0.8, 0.8, 0.8], EDGE_REACTION_TIME + 0.1)
+  assert relc.left_edge_detected
+
+  relc.params.get_bool.return_value = False
+  relc.read_params()
+  relc.update([0.0, 0.9], [0.0, 0.8, 0.8, 0.8], V_HIGH)
+  assert not relc.left_edge_detected
+  assert not relc.right_edge_detected

sunnypilot/selfdrive/selfdrived/events.py

@@ -243,4 +243,12 @@ EVENTS_SP: dict[int, dict[str, Alert | AlertCallbackType]] = {
       AlertStatus.normal, AlertSize.none,
       Priority.MID, VisualAlert.none, AudibleAlert.prompt, 3.),
   },
+
+  EventNameSP.laneChangeRoadEdge: {
+    ET.WARNING: Alert(
+      "Lane Change Unavailable: Road Edge",
+      "",
+      AlertStatus.userPrompt, AlertSize.small,
+      Priority.LOW, VisualAlert.none, AudibleAlert.prompt, 0.1),
+  },
 }

sunnypilot/sunnylink/params_metadata.json

@@ -1,4 +1,26 @@
 {
+  "AccelPersonality": {
+    "title": "Acceleration Profile",
+    "description": "Eco slows earliest and recovers gently, Normal balances comfort and response, and Sport reacts and recovers more quickly.",
+    "options": [
+      {
+        "value": 0,
+        "label": "Eco"
+      },
+      {
+        "value": 1,
+        "label": "Normal"
+      },
+      {
+        "value": 2,
+        "label": "Sport"
+      }
+    ]
+  },
+  "AccelPersonalityEnabled": {
+    "title": "Enable Accel Controller",
+    "description": "Begin slowing early and smoothly behind lead vehicles. Stock longitudinal control retains braking and stopping authority."
+  },
   "AccessToken": {
     "title": "AccessTokenIsNice",
     "description": ""
@@ -1118,6 +1140,10 @@
     "title": "Record Front Lock",
     "description": ""
   },
+  "RoadEdgeLaneChangeEnabled": {
+    "title": "Block Lane Change: Road Edge Detection",
+    "description": ""
+  },
   "RoadName": {
     "title": "Road Name",
     "description": ""
@@ -1323,6 +1349,22 @@
     "max": 5.0,
     "step": 0.1,
     "unit": "m/s\u00b2"
+  },
+    "ToyotaAutoHold": {
+    "title": "Toyota: Auto Brake Hold FOR TSS2 HYBRID CARS",
+    "description": ""
+  },
+    "ToyotaEnhancedBsm": {
+    "title": "Toyota: Prius TSS2 BSM and some tssp",
+    "description": ""
+  },
+    "ToyotaTSS2Long": {
+    "title": "Toyota: custom longitudinal for TSS2",
+    "description": ""
+  },
+    "ToyotaDriveMode": {
+    "title": "Enable drive mode btn link",
+    "description": ""
   },
   "ToyotaEnforceStockLongitudinal": {
     "title": "Toyota: Enforce Factory Longitudinal Control",

sunnypilot/sunnylink/settings_ui.json

@@ -620,6 +620,58 @@
                 }
               ]
             },
+            {
+              "key": "AccelPersonalityEnabled",
+              "widget": "toggle",
+              "title": "Enable Accel Controller",
+              "description": "Begin slowing early and smoothly behind lead vehicles. Stock longitudinal control retains braking and stopping authority.",
+              "visibility": [
+                {
+                  "type": "capability",
+                  "field": "has_longitudinal_control",
+                  "equals": true
+                }
+              ],
+              "enablement": [
+                {
+                  "type": "capability",
+                  "field": "has_longitudinal_control",
+                  "equals": true
+                }
+              ]
+            },
+            {
+              "key": "AccelPersonality",
+              "widget": "multiple_button",
+              "title": "Acceleration Profile",
+              "description": "Eco slows earliest and recovers gently, Normal balances comfort and response, and Sport reacts and recovers more quickly.",
+              "options": [
+                {
+                  "value": 0,
+                  "label": "Eco"
+                },
+                {
+                  "value": 1,
+                  "label": "Normal"
+                },
+                {
+                  "value": 2,
+                  "label": "Sport"
+                }
+              ],
+              "enablement": [
+                {
+                  "type": "capability",
+                  "field": "has_longitudinal_control",
+                  "equals": true
+                },
+                {
+                  "type": "param",
+                  "key": "AccelPersonalityEnabled",
+                  "equals": true
+                }
+              ]
+            },
             {
               "key": "IntelligentCruiseButtonManagement",
               "widget": "toggle",
@@ -2001,6 +2053,22 @@
                   "equals": true
                 }
               ]
+            },
+            {
+              "key": "PlanplusControl",
+              "widget": "option",
+              "title": "Plan Plus Controls",
+              "description": "Adjust planplus model recentering strength. The higher this number the more aggressively the model will recover to lane center; too high and it will ping-pong.",
+              "min": 0.0,
+              "max": 2.0,
+              "step": 0.1,
+              "enablement": [
+                {
+                  "type": "param",
+                  "key": "ShowAdvancedControls",
+                  "equals": true
+                }
+              ]
             }
           ]
         },
@@ -2168,6 +2236,50 @@
       "title": "Toyota / Lexus Settings",
       "description": "",
       "items": [
+        {
+          "key": "ToyotaAutoHold",
+          "widget": "toggle",
+          "needs_onroad_cycle": true,
+          "title": "Toyota: Auto Brake Hold FOR TSS2 HYBRID CARS",
+          "enablement": [
+            {
+              "type": "not_engaged"
+            }
+          ]
+        },
+        {
+          "key": "ToyotaEnhancedBsm",
+          "widget": "toggle",
+          "needs_onroad_cycle": true,
+          "title": "Toyota: Prius TSS2 BSM and some tssp",
+          "enablement": [
+            {
+              "type": "not_engaged"
+            }
+          ]
+        },
+        {
+          "key": "ToyotaTSS2Long",
+          "widget": "toggle",
+          "needs_onroad_cycle": true,
+          "title": "Toyota: custom longitudinal for TSS2",
+          "enablement": [
+            {
+              "type": "not_engaged"
+            }
+          ]
+        },
+        {
+          "key": "ToyotaDriveMode",
+          "widget": "toggle",
+          "needs_onroad_cycle": true,
+          "title": "Enable drive mode btn link",
+          "enablement": [
+            {
+              "type": "not_engaged"
+            }
+          ]
+        },
         {
           "key": "ToyotaEnforceStockLongitudinal",
           "widget": "toggle",

sunnypilot/sunnylink/settings_ui_src/pages/cruise.yaml

@@ -43,6 +43,32 @@ sections:
       label: Relaxed
     enablement:
     - $ref: '#/macros/longitudinal'
+  - key: AccelPersonalityEnabled
+    widget: toggle
+    title: Enable Accel Controller
+    description: Begin slowing early and smoothly behind lead vehicles. Stock longitudinal control retains braking
+      and stopping authority.
+    visibility:
+    - $ref: '#/macros/longitudinal'
+    enablement:
+    - $ref: '#/macros/longitudinal'
+  - key: AccelPersonality
+    widget: multiple_button
+    title: Acceleration Profile
+    description: Eco slows earliest and recovers gently, Normal balances comfort and response, and Sport reacts
+      and recovers more quickly.
+    options:
+    - value: 0
+      label: Eco
+    - value: 1
+      label: Normal
+    - value: 2
+      label: Sport
+    enablement:
+    - $ref: '#/macros/longitudinal'
+    - type: param
+      key: AccelPersonalityEnabled
+      equals: true
   - key: IntelligentCruiseButtonManagement
     widget: toggle
     title: Intelligent Cruise Button Management (ICBM) (Alpha)

sunnypilot/sunnylink/settings_ui_src/pages/models.yaml

@@ -51,6 +51,16 @@ sections:
         key: LagdToggle
         equals: true
     - $ref: '#/macros/advanced_only'
+  - key: PlanplusControl
+    widget: option
+    title: Plan Plus Controls
+    description: Adjust planplus model recentering strength. The higher this number the more aggressively the model will recover
+      to lane center; too high and it will ping-pong.
+    min: 0.0
+    max: 2.0
+    step: 0.1
+    enablement:
+    - $ref: '#/macros/advanced_only'
 - id: lateral_control
   title: Lateral Control
   description: Neural network lateral control for supported models

sunnypilot/sunnylink/settings_ui_src/pages/vehicle.yaml

@@ -60,6 +60,30 @@ sections:
   title: Toyota / Lexus Settings
   description: ''
   items:
+  - key: ToyotaAutoHold
+    widget: toggle
+    needs_onroad_cycle: true
+    title: 'Toyota: Auto Brake Hold FOR TSS2 HYBRID CARS'
+    enablement:
+    - $ref: '#/macros/not_engaged'
+  - key: ToyotaEnhancedBsm
+    widget: toggle
+    needs_onroad_cycle: true
+    title: 'Toyota: Prius TSS2 BSM and some tssp'
+    enablement:
+    - $ref: '#/macros/not_engaged'
+  - key: ToyotaTSS2Long
+    widget: toggle
+    needs_onroad_cycle: true
+    title: 'Toyota: custom longitudinal for TSS2'
+    enablement:
+    - $ref: '#/macros/not_engaged'
+  - key: ToyotaDriveMode
+    widget: toggle
+    needs_onroad_cycle: true
+    title: Enable drive mode btn link
+    enablement:
+    - $ref: '#/macros/not_engaged'
   - key: ToyotaEnforceStockLongitudinal
     widget: toggle
     needs_onroad_cycle: true

sunnypilot/sunnylink/tests/test_settings_schema.py

@@ -272,6 +272,22 @@ class TestKnownPanels:
     nnlc_enable_keys = {r.get("key") for r in nnlc.get("enablement", []) if r.get("type") == "param"}
     assert "EnforceTorqueControl" in nnlc_enable_keys
 
+  def test_accel_controller_profile_mapping_and_enablement(self, schema):
+    cruise = next(p for p in schema["panels"] if p["id"] == "cruise")
+    items = {item["key"]: item for item in _iter_panel_items(cruise)}
+
+    assert items["AccelPersonalityEnabled"]["widget"] == "toggle"
+    assert items["AccelPersonality"]["options"] == [
+      {"value": 0, "label": "Eco"},
+      {"value": 1, "label": "Normal"},
+      {"value": 2, "label": "Sport"},
+    ]
+    assert {
+      "type": "param",
+      "key": "AccelPersonalityEnabled",
+      "equals": True,
+    } in items["AccelPersonality"]["enablement"]
+
 
 class TestKnownVehicleSettings:
   def test_hyundai_has_longitudinal_tuning(self, schema):

system/ui/lib/scroll_panel2.py

@@ -45,8 +45,9 @@ class ScrollState(Enum):
 
 
 class GuiScrollPanel2:
-  def __init__(self, horizontal: bool = True) -> None:
+  def __init__(self, horizontal: bool = True, handle_out_of_bounds: bool = True) -> None:
     self._horizontal = horizontal
+    self._handle_out_of_bounds = handle_out_of_bounds
     self._state = ScrollState.STEADY
     self._offset: rl.Vector2 = rl.Vector2(0, 0)
     self._initial_click_event: MouseEvent | None = None
@@ -98,7 +99,7 @@ class GuiScrollPanel2:
       # simple exponential return if out of bounds
       # out of bounds is handled by snapping, so skip if set
       out_of_bounds = self.get_offset() > max_offset or self.get_offset() < min_offset
-      if out_of_bounds and snap_target is None:
+      if out_of_bounds and snap_target is None and self._handle_out_of_bounds:
         target = max_offset if self.get_offset() > max_offset else min_offset
 
         dt = rl.get_frame_time() or 1e-6

system/ui/widgets/scroller.py

@@ -75,7 +75,6 @@ class _Scroller(Widget):
     self._items: list[Widget] = []
     self._horizontal = horizontal
     self._snap_items = snap_items
-    assert not self._snap_items or self._horizontal, "Snapping is only supported for horizontal scrolling"
     self._spacing = spacing
     self._pad = pad
 
@@ -191,8 +190,12 @@ class _Scroller(Widget):
     snap_target: float | None = None
     if self._snap_items and visible_items and self._scrolling_to[0] is None:
       # TODO: this doesn't handle two small buttons at the edges well
-      center_pos = self._rect.x + self._rect.width / 2
-      closest_delta_pos = min((((item.rect.x + item.rect.width / 2) - center_pos) for item in visible_items), key=abs)
+      if self._horizontal:
+        center_pos = self._rect.x + self._rect.width / 2
+        closest_delta_pos = min((((item.rect.x + item.rect.width / 2) - center_pos) for item in visible_items), key=abs)
+      else:
+        center_pos = self._rect.y + self._rect.height / 2
+        closest_delta_pos = min((((item.rect.y + item.rect.height / 2) - center_pos) for item in visible_items), key=abs)
       snap_target = self.scroll_panel.get_offset() - closest_delta_pos
 
     return self.scroll_panel.update(self._rect, content_size, snap_target=snap_target)