Force Stop: Fast Approach

Force Stop is now better at faster approach speeds. It has more authority on braking in general.

Force Stop Distance Offset: still the lever that moves the ending point sooner or later.

Radar bug fixed

cereal/custom.capnp

@@ -221,6 +221,11 @@ struct StarPilotPlan @0xf98d843bfd7004a3 {
   trackingLead @36 :Bool;
   stopSignConfirmed @37 :Bool;
   pulseGlideCoasting @38 :Bool;  # developer-only P&G phase for on-road status UI
+  # Curve Speed Controller diagnostics, for tuning and rollout validation
+  cscOverridden @39 :Bool;          # driver cancelled this curve with RES+
+  cscLearnedLatAccel @40 :Float32;  # learned comfort at the current curvature, before margin
+  cscBindingDistance @41 :Float32;  # distance to the horizon point setting the target, m
+  approachStopLength @42 :Float32;  # pre-commit distance to a detected stop, m; 0 when off
 }
 
 struct StarPilotRadarState @0xb86e6369214c01c8 {

selfdrive/controls/controlsd.py

@@ -169,6 +169,20 @@ CURVATURE_HOLD_OPPOSITE_RELEASE = 0.01  # 1/m
 CURVATURE_HOLD_CONFIRM_MIN = 0.003  # 1/m (~7 deg) of wound curvature before capture
 CURVATURE_HOLD_CONFIRM_SWEPT = 0.6  # rad of heading swept this blinker cycle; past this the push is exit-shaping, not initiation
 
+# Pull-away twitch guard. modeld divides the action head's lateral-ACCELERATION output by
+# max(1, v)^2, so its residual at pull-away (~0.02 m/s^2, the head's noise floor) reads as
+# curvature 0.015 — 38 deg of wheel — where the same value at highway speed is 0.2 deg.
+# Route 78511c37 twitched on 10 of 10 straight takeoffs. The model's own planned path is the
+# tell: it read straight there while the action demanded 6-108x more.
+TWITCH_GUARD_MAX_SPEED = 4.0      # m/s; above this the 1/v^2 amplification is gone
+TWITCH_GUARD_FADE_SPEED = 3.0     # m/s; full strength below, faded out by MAX_SPEED
+TWITCH_GUARD_PLAN_RATIO = 4.0     # allowed |action| / |plan curvature|
+TWITCH_GUARD_FLOOR = 0.002        # 1/m (~5 deg); a near-zero probe must not clamp to nothing
+TWITCH_GUARD_STRAIGHT_LO = 0.005  # 1/m; a plain ratio is too permissive near straight (3x of
+TWITCH_GUARD_STRAIGHT_HI = 0.014  # 0.003 still licenses 22 deg), so fade the allowance out too
+TWITCH_GUARD_MIN_REACH = 12.0     # m; shorter plans read straight while the action legitimately
+                                  # unwinds a turn (ce2b186c51 seg 28 t=14.6). Twitches: p5 24 m
+
 
 def _plan_circle_curvature(xs, ys, lookahead: float) -> float:
   # curvature of the circle through the origin, tangent to the car's heading, passing
@@ -226,6 +240,25 @@ def get_plan_reach(model_v2) -> float:
   return xs[-1] if len(xs) else 0.0
 
 
+def limit_curvature_to_plan(model_v2, curvature: float, v_ego: float) -> float:
+  # See TWITCH_GUARD_*. Magnitude only: the command is bounded, never reversed. FAR fit alone —
+  # the near probe swings with the car's heading error, so once a twitch has yawed the car it
+  # bends to correct it and licenses the very command that caused it (seg 10 t=53.1).
+  if v_ego >= TWITCH_GUARD_MAX_SPEED or curvature == 0.0:
+    return curvature
+  if get_plan_reach(model_v2) < TWITCH_GUARD_MIN_REACH:
+    return curvature
+  plan = abs(_plan_circle_curvature(model_v2.position.x, model_v2.position.y,
+                                    CURVATURE_HOLD_PLAN_LOOKAHEAD_FAR))
+  straightness = (plan - TWITCH_GUARD_STRAIGHT_LO) / (TWITCH_GUARD_STRAIGHT_HI - TWITCH_GUARD_STRAIGHT_LO)
+  limit = max(TWITCH_GUARD_PLAN_RATIO * plan * min(max(straightness, 0.0), 1.0), TWITCH_GUARD_FLOOR)
+  if abs(curvature) <= limit:
+    return curvature
+  fade = (TWITCH_GUARD_MAX_SPEED - v_ego) / (TWITCH_GUARD_MAX_SPEED - TWITCH_GUARD_FADE_SPEED)
+  fade = min(max(fade, 0.0), 1.0)
+  return curvature + (math.copysign(limit, curvature) - curvature) * fade
+
+
 def get_control_lateral_smooth_seconds(brand: str, v_ego: float, vehicle_smooth_seconds: float) -> float:
   if brand == "rivian" or (brand == "subaru" and vehicle_smooth_seconds > 0.0):
     return get_car_lateral_smooth_seconds(brand, v_ego, vehicle_smooth_seconds)
@@ -499,6 +532,10 @@ class Controls:
     # here is positive for RIGHT turns (pauseturn log: left turn at +148 deg steering
     # angle logs desiredCurvature -0.07), so the blinker maps right=+1, left=-1.
     blinker_dir = float(CS.rightBlinker) - float(CS.leftBlinker)
+    # Pull-away twitch guard (see TWITCH_GUARD_*). Requires no turn intent in play, so the
+    # pre-wind ratchet, turn lead and exit opposite-release never see a reduced command.
+    if CC.latActive and blinker_dir == 0.0 and self.turn_hold_curvature == 0.0:
+      new_desired_curvature = limit_curvature_to_plan(model_v2, new_desired_curvature, CS.vEgo)
     # heading swept in the blinker's direction over the whole blinker cycle (any speed):
     # discriminates a turn not yet made from one being exited (see the re-arm below)
     if blinker_dir == 0.0:

selfdrive/controls/lib/longitudinal_planner.py

@@ -9,6 +9,7 @@ from openpilot.common.filter_simple import FirstOrderFilter
 from openpilot.common.realtime import DT_MDL
 from openpilot.selfdrive.modeld.constants import ModelConstants
 from openpilot.starpilot.common.model_versions import is_tinygrad_model_version
+from openpilot.starpilot.controls.lib.starpilot_vcruise import FT_TO_M, OFFSET_FT_MAX, OFFSET_FT_MIN
 from openpilot.selfdrive.controls.lib.longcontrol import LongCtrlState
 from openpilot.selfdrive.controls.lib.longitudinal_mpc_lib.long_mpc import LongitudinalMpc
 from openpilot.selfdrive.controls.lib.longitudinal_mpc_lib.long_mpc import desired_follow_distance
@@ -381,6 +382,11 @@ def get_vehicle_min_accel(CP, v_ego):
 
 # Restored planner constants retained by CEM, stop, and departure paths.
 A_CRUISE_MIN = -1.0
+# The stop distance runs ~9 m long through the mid-approach, which leaves the obstacle slack
+# so it stays silent and deceleration sags. Multiplicative so the trim scales with what is
+# left. Note the car parks where the obstacle sits, so this is also a placement bias — 0.85
+# stopped ~4.6 m short, 0.93 ~1.6 m.
+FORCE_STOP_OBSTACLE_TRIM = 0.93
 STANDSTILL_LEAD_CREEP_RELEASE_MIN_LEAD_SPEED = 0.25
 STANDSTILL_LEAD_CREEP_RELEASE_MIN_LEAD_ACCEL = 0.08
 STANDSTILL_LEAD_CREEP_RELEASE_MIN_GAP_MARGIN = 0.1
@@ -2217,8 +2223,17 @@ class LongitudinalPlanner:
     force_stop_x = None
     force_stop_handoff_m = get_force_stop_handoff_distance(self.CP.carFingerprint)
     if sm['starpilotPlan'].forcingStop and sm['starpilotPlan'].forcingStopLength > force_stop_handoff_m:
+      stop_length = float(sm['starpilotPlan'].forcingStopLength)
+    else:
+      # pre-commit the envelope is only a speed ceiling, which the solver tracks with a lag;
+      # getattr so a stale cereal build degrades to the old behaviour instead of raising
+      stop_length = float(getattr(sm['starpilotPlan'], 'approachStopLength', 0.0))
+    if stop_length > force_stop_handoff_m:
+      # ForceStopDistanceOffset shifts the perceived line for the v_cruise ceiling, so it has
+      # to shift the obstacle too or the slider barely moves anything now that stop_x leads.
+      offset_ft = max(OFFSET_FT_MIN, min(OFFSET_FT_MAX, int(getattr(starpilot_toggles, 'force_stop_distance_offset', 0) or 0)))
       force_stop_x = (
-        float(sm['starpilotPlan'].forcingStopLength) + STOP_DISTANCE +
+        stop_length * FORCE_STOP_OBSTACLE_TRIM + offset_ft * FT_TO_M + STOP_DISTANCE +
         get_force_stop_distance_bias(self.CP.carFingerprint)
       )
 

selfdrive/controls/radard.py

@@ -53,6 +53,8 @@ ADJACENT_STOP_REST_FRAMES = 15
 ADJACENT_STOP_MIN_Y = 1.8         # m — inside this is our own lane
 ADJACENT_STOP_MAX_Y = 7.5         # m — beyond this is roadside, not an adjacent lane
 ADJACENT_STOP_MAX_D = 110.0       # m
+ADJACENT_STOP_QUEUE_GAP_M = 5.0   # m — anything stopped beyond the furthest qualifier means
+                                  # the bar is past it too, so the hint would stop us short
 
 
 class KalmanParams:
@@ -179,6 +181,10 @@ class Track:
     if self.leadTrackID == self.identifier:
       return False
 
+    return self.in_adjacent_lane(model_data)
+
+  def in_adjacent_lane(self, model_data: capnp._DynamicStructReader):
+    """Lane geometry only, no deceleration history — also used to spot a queue ahead."""
     if not (ADJACENT_STOP_MIN_Y < abs(self.yRel) < ADJACENT_STOP_MAX_Y):
       return False
 
@@ -398,9 +404,10 @@ def get_adjacent_lead(tracks: dict[int, Track], standstill: bool, model_data: ca
 def get_adjacent_stopped(tracks: dict[int, Track], model_data: capnp._DynamicStructReader) -> dict[str, Any]:
   """Stop-line hint: a vehicle that decelerated to a stop in a neighbouring lane.
 
-  Takes the FARTHEST qualifying vehicle: in a queue the front car sits at the bar and the
-  rest are closer to us, so the nearest one underestimates the distance. The consumer only
-  shortens with this, so underestimating is the harmful direction.
+  Takes the FARTHEST qualifying vehicle, then drops the hint entirely if a queue reaches
+  past it. Cars already stopped when we acquire them never show the moving -> stopped
+  transition, so the qualifying set is biased toward the back of a line; without this the
+  hint marks a mid-queue bumper and stops us short of the bar.
   """
   if len(model_data.laneLines) < 4:
     return {'status': False}
@@ -410,6 +417,11 @@ def get_adjacent_stopped(tracks: dict[int, Track], model_data: capnp._DynamicStr
     return {'status': False}
 
   furthest = max(candidates, key=lambda c: c.dRel)
+  for c in tracks.values():
+    if (c.dRel > furthest.dRel + ADJACENT_STOP_QUEUE_GAP_M and
+        abs(c.vLead) < ADJACENT_STOP_REST_V and
+        c.in_adjacent_lane(model_data)):
+      return {'status': False}
   return {
     'status': True,
     'dRel': float(furthest.dRel),

selfdrive/controls/tests/test_conditional_experimental_mode.py

@@ -1168,6 +1168,7 @@ def test_starpilot_planner_updates_cem_with_current_frame_state(monkeypatch):
 
   try:
     monkeypatch.setattr(starpilot_planner_module, "calculate_road_curvature", lambda model, v_ego: (0.01, 1.0))
+    monkeypatch.setattr(starpilot_planner_module, "extract_curve_profile", lambda model: ([], []))
     monkeypatch.setattr(planner.starpilot_acceleration, "update", lambda *args, **kwargs: None)
     monkeypatch.setattr(planner.starpilot_events, "update", lambda *args, **kwargs: None)
     monkeypatch.setattr(planner.starpilot_vcruise, "update", lambda *args, **kwargs: 0.0)

selfdrive/controls/tests/test_curve_speed_controller.py

@@ -0,0 +1,506 @@
+import numpy as np
+import pytest
+
+from types import SimpleNamespace
+
+from openpilot.common.realtime import DT_MDL
+from openpilot.starpilot.common.starpilot_variables import DEFAULT_LATERAL_ACCELERATION
+from openpilot.starpilot.controls.lib.curve_speed_controller import (
+  CSC_APPROACH_DECEL,
+  CSC_COMFORT_MARGIN,
+  CSC_COUNT_CAP,
+  CSC_EGO_HEADROOM,
+  CSC_FARFIELD_GAIN,
+  CSC_LAT_ACCEL_MAX,
+  CSC_MIN_SPEED,
+  MAX_CURVATURE,
+  PRIOR_CURVATURE_BP,
+  PRIOR_LAT_ACCEL_V,
+  CSC_NUDGE,
+  CSC_NUDGE_WEIGHT,
+  CSC_OVERRIDE_WATCH_TIME,
+  CSC_TARGET_UP_RATE,
+  CSC_TRAINING_SETTLE_TIME,
+  CurveSpeedController,
+  weighted_isotonic,
+)
+
+
+class FakeParams:
+  def __init__(self, values=None):
+    self.values = dict(values or {})
+
+  def get(self, *args, **kwargs):
+    key = args[0] if args else None
+    return self.values.get(key)
+
+  def put_nonblocking(self, key, value):
+    self.values[key] = value
+
+
+def make_controller(curve_profile=None, curvature_data=None, weather_id=0, reduce_lat=0.0, road_curvature=0.02, driving_in_curve=False):
+  if curve_profile is None:
+    curve_profile = (np.zeros(33), np.linspace(0.0, 300.0, 33))
+
+  planner = SimpleNamespace(
+    params=FakeParams({"CurvatureData": curvature_data} if curvature_data is not None else None),
+    curve_profile=curve_profile,
+    starpilot_weather=SimpleNamespace(weather_id=weather_id, reduce_lateral_acceleration=reduce_lat),
+    road_curvature=road_curvature,
+    driving_in_curve=driving_in_curve,
+    tracking_lead=False,
+    lateral_acceleration=0.0,
+  )
+  controller = CurveSpeedController(SimpleNamespace(starpilot_planner=planner))
+  return planner, controller
+
+
+def make_sm(*, gas=False, brake=False, long_active=True, blinker=False, accel_pressed=False):
+  return {
+    "carControl": SimpleNamespace(longActive=long_active),
+    "carState": SimpleNamespace(gasPressed=gas, brakePressed=brake, leftBlinker=blinker, rightBlinker=False),
+    "starpilotCarState": SimpleNamespace(accelPressed=accel_pressed),
+    "onroadEvents": [],
+  }
+
+
+def single_apex_profile(curvature, distance):
+  distances = np.linspace(0.0, max(distance * 1.5, 1.0), 33)
+  curvatures = np.zeros(33)
+  index = int(np.argmin(np.abs(distances - distance)))
+  distances[index] = distance
+  curvatures[index] = curvature
+  return curvatures, distances
+
+
+def converge(controller, v_ego, v_cruise, frames=600):
+  for _ in range(frames):
+    controller.update_target(v_ego, v_cruise)
+  return controller.target
+
+
+def envelope_speed(controller, curvature, distance):
+  curve_speed = max(float(np.sqrt(controller.lat_accel_for_curvature(curvature) / curvature)), CSC_MIN_SPEED)
+  return float(np.sqrt(curve_speed**2 + 2.0 * CSC_APPROACH_DECEL * distance))
+
+
+def test_straight_road_target_is_cruise_speed():
+  _, controller = make_controller()
+
+  controller.update_target(30.0, 30.0)
+
+  assert controller.target == pytest.approx(30.0)
+
+
+def test_distant_apex_does_not_constrain_until_braking_is_due():
+  # derived from the shipped decel so retuning it doesn't silently invalidate the case
+  _, probe = make_controller()
+  curve_speed = max(float(np.sqrt(probe.lat_accel_for_curvature(0.02) / 0.02)), CSC_MIN_SPEED)
+  beyond_braking = 1.3 * (30.0**2 - curve_speed**2) / (2 * CSC_APPROACH_DECEL)
+
+  _, controller = make_controller(curve_profile=single_apex_profile(0.02, beyond_braking))
+
+  target = converge(controller, 30.0, 30.0)
+
+  assert target == pytest.approx(30.0)
+
+
+def test_apex_in_braking_range_constrains_to_kinematic_envelope():
+  _, controller = make_controller(curve_profile=single_apex_profile(0.02, 150.0))
+
+  target = converge(controller, 30.0, 30.0)
+
+  assert target == pytest.approx(envelope_speed(controller, 0.02, 150.0), abs=0.1)
+  assert target < 30.0
+
+
+def test_exit_recovery_rises_immediately_without_freeze():
+  planner, controller = make_controller(curve_profile=single_apex_profile(0.03, 20.0))
+  low_target = converge(controller, 15.0, 30.0)
+  assert low_target < 20.0
+
+  planner.curve_profile = (np.zeros(33), np.linspace(0.0, 300.0, 33))
+  controller.update_target(15.0, 30.0)
+  assert controller.target > low_target          # rises on the very next frame, no freeze
+  assert controller.target - low_target == pytest.approx(CSC_TARGET_UP_RATE * DT_MDL)
+
+  # and it clears the car by the headroom within the time the up-rate needs
+  frames = int((15.0 + CSC_EGO_HEADROOM - controller.target) / (CSC_TARGET_UP_RATE * DT_MDL)) + 1
+  for _ in range(frames):
+    controller.update_target(15.0, 30.0)
+  assert controller.target >= 15.0 + CSC_EGO_HEADROOM
+
+  recovered = converge(controller, 15.0, 30.0)
+  assert recovered == pytest.approx(30.0)
+
+
+def test_upward_jitter_in_the_envelope_is_rate_limited():
+  # a sweeper the envelope only grazes: raw_target flicks between a mild cap and the
+  # set speed. The target must not chase the jumps, or the glow strobes.
+  planner, controller = make_controller(curve_profile=single_apex_profile(0.002, 40.0))
+  steady = converge(controller, 30.0, 32.0)
+  assert steady < 32.0
+
+  flat = (np.zeros(33), np.linspace(0.0, 300.0, 33))
+  grazing = planner.curve_profile
+  peak = steady
+  for i in range(40):
+    planner.curve_profile = flat if i % 2 else grazing
+    controller.update_target(30.0, 32.0)
+    assert controller.target - peak <= CSC_TARGET_UP_RATE * DT_MDL + 1e-6
+    peak = controller.target
+
+
+def test_firm_distant_curvature_is_corrected_for_the_model_under_read():
+  # the model reads ~0.81x actual at range, so a firm distant bend binds later than it should
+  distance = 90.0
+  _, plain = make_controller(curve_profile=single_apex_profile(0.0045, distance))
+  _, probe = make_controller()
+  corrected = probe._correct_far_field(*single_apex_profile(0.0045, distance))
+
+  assert corrected.max() == pytest.approx(0.0045 * CSC_FARFIELD_GAIN)
+  assert converge(plain, 30.0, 30.0) < envelope_speed(plain, 0.0045, distance) + 1e-6
+
+
+def test_weak_or_near_readings_are_left_alone():
+  _, probe = make_controller()
+
+  # too weak to carry usable magnitude at range
+  weak = probe._correct_far_field(*single_apex_profile(0.002, 90.0))
+  assert weak.max() == pytest.approx(0.002)
+
+  # firm, but close enough that the model is already accurate
+  near = probe._correct_far_field(*single_apex_profile(0.0045, 10.0))
+  assert near.max() == pytest.approx(0.0045)
+
+
+def test_far_field_correction_brings_the_slowdown_forward():
+  profile = single_apex_profile(0.0045, 120.0)
+  _, controller = make_controller(curve_profile=profile)
+
+  corrected = converge(controller, 30.0, 30.0)
+  raw_curvatures, distances = profile
+  uncorrected = float(np.sqrt(
+    max(np.sqrt(controller.lat_accel_for_curvature(0.0045) / 0.0045), CSC_MIN_SPEED) ** 2
+    + 2.0 * CSC_APPROACH_DECEL * 120.0))
+
+  assert corrected < uncorrected      # binds sooner than the model's own reading would
+
+
+def test_fresh_activation_seeds_at_envelope_not_cruise():
+  _, controller = make_controller(curve_profile=(np.full(33, 0.05), np.linspace(0.0, 60.0, 33)))
+
+  controller.update_target(6.0, 30.0)
+
+  assert controller.target < 15.0
+
+
+def test_target_never_trails_accelerating_car_when_unconstrained():
+  planner, controller = make_controller(curve_profile=single_apex_profile(0.03, 20.0))
+  converge(controller, 15.0, 30.0)
+
+  planner.curve_profile = (np.zeros(33), np.linspace(0.0, 300.0, 33))
+  v_ego = 15.0
+  caught_up = None
+  for frame in range(200):
+    v_ego = min(v_ego + 2.0 * DT_MDL, 30.0)
+    controller.update_target(v_ego, 30.0)
+    # the target climbs faster than the car can, so once it is ahead it stays ahead
+    if controller.target >= v_ego:
+      caught_up = caught_up if caught_up is not None else frame
+    assert caught_up is None or controller.target >= min(30.0, v_ego) - 1e-6
+
+  assert caught_up is not None and caught_up * DT_MDL < 2.0
+  assert controller.target == pytest.approx(30.0)
+
+
+def test_target_does_not_ratchet_down_with_ego_speed():
+  _, controller = make_controller(curve_profile=single_apex_profile(0.02, 150.0))
+  target = converge(controller, 30.0, 30.0)
+  assert target > CSC_MIN_SPEED  # a real curve speed, not floored
+
+  controller.update_target(14.0, 30.0)
+
+  assert controller.target == pytest.approx(target, abs=0.2)
+
+
+def test_sharp_curve_target_floors_at_min_speed():
+  _, controller = make_controller(curve_profile=(np.full(33, 0.1), np.linspace(0.0, 100.0, 33)))
+
+  target = converge(controller, 15.0, 30.0)
+
+  assert target == pytest.approx(CSC_MIN_SPEED, abs=0.05)
+
+
+def test_weather_reduces_curve_speed():
+  _, dry = make_controller(curve_profile=single_apex_profile(0.01, 0.0))
+  _, wet = make_controller(curve_profile=single_apex_profile(0.01, 0.0), weather_id=1, reduce_lat=0.2)
+
+  dry_target = converge(dry, 20.0, 30.0)
+  wet_target = converge(wet, 20.0, 30.0)
+
+  assert wet_target < dry_target
+  assert wet_target == pytest.approx(dry_target * np.sqrt(0.8), abs=0.1)
+
+
+def test_prior_gives_higher_lat_accel_for_sharper_curves():
+  _, controller = make_controller()
+
+  assert controller.learned_lat_accel(0.001) == pytest.approx(1.5, abs=0.05)
+  assert controller.learned_lat_accel(MAX_CURVATURE) > controller.learned_lat_accel(0.001)
+  assert controller.learned_lat_accel(MAX_CURVATURE) == pytest.approx(
+    float(np.interp(MAX_CURVATURE, PRIOR_CURVATURE_BP, PRIOR_LAT_ACCEL_V)), abs=0.05)
+  assert controller.lateral_acceleration == pytest.approx(DEFAULT_LATERAL_ACCELERATION)
+
+
+def test_comfort_margin_matches_the_learned_habit():
+  # margin is fixed at 1.0 -- CSC targets exactly the driver's own learned comfort
+  _, controller = make_controller()
+
+  assert CSC_COMFORT_MARGIN == pytest.approx(1.0)
+  assert controller.lat_accel_for_curvature(0.01) == pytest.approx(controller.learned_lat_accel(0.01))
+
+
+def test_binding_distance_reports_the_constraining_point():
+  _, controller = make_controller(curve_profile=single_apex_profile(0.02, 150.0))
+  converge(controller, 30.0, 30.0)
+
+  assert controller.binding_distance == pytest.approx(150.0, abs=1.0)
+
+
+def test_binding_distance_is_zero_when_unconstrained():
+  _, controller = make_controller()
+  converge(controller, 30.0, 30.0)
+
+  assert controller.binding_distance == 0.0
+
+
+def test_heavily_sampled_bucket_dominates_prior():
+  _, controller = make_controller(curvature_data={"0.05": {"average": 3.0, "count": 100000}})
+
+  assert controller.learned_lat_accel(0.05) == pytest.approx(3.0, abs=0.05)
+  assert controller.learned_lat_accel(0.08) >= controller.learned_lat_accel(0.05)
+
+
+def test_learned_curve_stays_monotonic_despite_low_outlier_bucket():
+  _, controller = make_controller(curvature_data={"0.05": {"average": 0.5, "count": 100000}})
+
+  assert controller.learned_lat_accel(0.05) >= controller.learned_lat_accel(0.03)
+
+
+def test_dense_bucket_is_not_overridden_by_sparse_neighbour():
+  # real device data: a running maximum ratcheted the 80-sample bucket up to the 20-sample neighbour
+  _, dense_low = make_controller(curvature_data={
+    "0.003": {"average": 1.95, "count": 20},
+    "0.005": {"average": 1.38, "count": 80},
+  })
+  _, dense_high = make_controller(curvature_data={
+    "0.003": {"average": 1.95, "count": 80},
+    "0.005": {"average": 1.38, "count": 20},
+  })
+
+  assert dense_low.learned_lat_accel(0.005) < 1.95      # not ratcheted to the sparse neighbour
+  assert dense_low.learned_lat_accel(0.005) >= dense_low.learned_lat_accel(0.003)
+  # whichever side is better sampled should pull the fit: swapping the counts must raise it
+  assert dense_high.learned_lat_accel(0.005) > dense_low.learned_lat_accel(0.005)
+
+
+def test_weighted_isotonic_pools_violators_by_weight():
+  fitted = weighted_isotonic(np.array([1.0, 3.0, 1.2]), np.array([1.0, 1.0, 1000.0]))
+
+  assert np.all(np.diff(fitted) >= -1e-9)
+  assert fitted[-1] == pytest.approx(1.2, abs=0.02)
+
+
+def test_weighted_isotonic_leaves_sorted_input_untouched():
+  values = np.array([1.0, 1.5, 2.0, 2.5])
+  fitted = weighted_isotonic(values, np.ones(4))
+
+  assert fitted == pytest.approx(values)
+
+
+def test_legacy_off_grid_curvature_data_merges_into_buckets():
+  _, controller = make_controller(curvature_data={
+    "0.0203": {"average": 2.5, "count": 10},
+    "0.02": {"average": 2.0, "count": 10},
+  })
+
+  assert controller.curvature_data["0.02"]["count"] == 20
+  assert controller.curvature_data["0.02"]["average"] == pytest.approx(2.25)
+
+
+def test_training_update_step_is_capped_by_ema_count():
+  planner, controller = make_controller(curvature_data={"0.02": {"average": 2.0, "count": 10000}}, driving_in_curve=True)
+  planner.lateral_acceleration = 3.0
+  controller.training_timer = CSC_TRAINING_SETTLE_TIME
+
+  controller.log_data(10.0, make_sm(long_active=False))
+
+  data = controller.curvature_data["0.02"]
+  assert data["count"] == 10001
+  assert data["average"] == pytest.approx((2.0 * CSC_COUNT_CAP + 3.0) / (CSC_COUNT_CAP + 1))
+
+
+def test_no_passive_training_right_after_csc_limited_speed():
+  planner, controller = make_controller(curve_profile=single_apex_profile(0.03, 20.0), driving_in_curve=True)
+  planner.lateral_acceleration = 3.0
+  converge(controller, 15.0, 30.0)
+  assert controller.training_quiet_timer > 0.0
+
+  controller.training_timer = CSC_TRAINING_SETTLE_TIME
+  controller.log_data(10.0, make_sm(long_active=False))
+  assert "0.02" not in controller.curvature_data
+  assert not controller.enable_training
+
+  controller.training_quiet_timer = 0.0
+  controller.training_timer = CSC_TRAINING_SETTLE_TIME
+  controller.log_data(10.0, make_sm(long_active=False))
+  assert controller.curvature_data["0.02"]["count"] == 1
+
+
+def test_training_settles_within_a_couple_of_seconds():
+  # a real drive rarely holds every eligibility condition for a whole model horizon,
+  # so the settle time has to be short enough that ordinary curves still teach it
+  planner, controller = make_controller(driving_in_curve=True)
+  planner.lateral_acceleration = 2.4
+  sm = make_sm(long_active=False)
+
+  for _ in range(int(CSC_TRAINING_SETTLE_TIME / DT_MDL) - 2):
+    controller.log_data(10.0, sm)
+  assert "0.02" not in controller.curvature_data
+
+  for _ in range(3):
+    controller.log_data(10.0, sm)
+  assert controller.curvature_data["0.02"]["count"] >= 1
+
+
+def test_brief_ineligibility_does_not_restart_the_settle_timer():
+  planner, controller = make_controller(driving_in_curve=True)
+  planner.lateral_acceleration = 2.4
+  sm = make_sm(long_active=False)
+  for _ in range(int(CSC_TRAINING_SETTLE_TIME / DT_MDL) + 1):
+    controller.log_data(10.0, sm)
+  trained = controller.curvature_data["0.02"]["count"]
+
+  # a lead flickers into the tracker for two frames, then leaves
+  planner.tracking_lead = True
+  controller.log_data(10.0, sm)
+  controller.log_data(10.0, sm)
+  planner.tracking_lead = False
+
+  controller.log_data(10.0, sm)
+  assert controller.curvature_data["0.02"]["count"] == trained + 1
+
+
+def test_sustained_ineligibility_still_drains_the_settle_timer():
+  planner, controller = make_controller(driving_in_curve=True)
+  planner.lateral_acceleration = 2.4
+  engaged = make_sm(long_active=True)
+  manual = make_sm(long_active=False)
+  for _ in range(int(CSC_TRAINING_SETTLE_TIME / DT_MDL) + 1):
+    controller.log_data(10.0, manual)
+
+  for _ in range(int(2 * CSC_TRAINING_SETTLE_TIME / DT_MDL)):
+    controller.log_data(10.0, engaged)
+  assert controller.training_timer == pytest.approx(0.0)
+
+  controller.log_data(10.0, manual)
+  assert not controller.enable_training
+
+
+def settle_override(controller, sm=None, frames=None):
+  """Run the post-override watch out so the pseudo-sample is committed."""
+  sm = sm if sm is not None else make_sm()
+  for _ in range(frames if frames is not None else int(CSC_OVERRIDE_WATCH_TIME / DT_MDL) + 1):
+    controller.handle_override(20.0, False, sm)
+
+
+def test_gas_override_nudges_bucket_up_once_per_episode():
+  _, controller = make_controller()
+  prior = controller.learned_lat_accel(0.02)
+  controller.target = 10.0
+
+  controller.handle_override(20.0, True, make_sm(gas=True))
+  controller.handle_override(20.0, True, make_sm(gas=True))
+  assert "0.02" not in controller.curvature_data  # still watching what the driver holds
+
+  settle_override(controller)
+  assert controller.curvature_data["0.02"]["count"] == CSC_NUDGE_WEIGHT
+  assert controller.curvature_data["0.02"]["average"] > prior
+
+  controller.handle_override(20.0, False, make_sm())
+  controller.target = 10.0
+  controller.handle_override(20.0, True, make_sm(gas=True))
+  settle_override(controller)
+  assert controller.curvature_data["0.02"]["count"] == 2 * CSC_NUDGE_WEIGHT
+
+
+def test_override_learns_the_cornering_the_driver_actually_held():
+  # the whole point: a fixed step needs several rejections to close a real disagreement,
+  # so record what they demonstrated instead
+  planner, observed = make_controller(driving_in_curve=True)
+  observed.target = 10.0
+  observed.handle_override(20.0, True, make_sm(gas=True))
+  planner.lateral_acceleration = 2.9          # they hold the curve much harder than CSC wanted
+  settle_override(observed, make_sm(gas=True))
+
+  _, stepped = make_controller(driving_in_curve=True)
+  stepped._apply_nudge(CSC_NUDGE)             # what the old fixed-step path would have recorded
+
+  assert observed.curvature_data["0.02"]["average"] == pytest.approx(2.9)
+  assert observed.curvature_data["0.02"]["average"] > stepped.curvature_data["0.02"]["average"]
+  assert observed.learned_lat_accel(0.02) > stepped.learned_lat_accel(0.02)
+
+
+def test_override_on_a_straight_still_registers_the_fixed_step():
+  planner, controller = make_controller()
+  prior = controller.learned_lat_accel(0.02)
+  controller.target = 10.0
+
+  controller.handle_override(20.0, True, make_sm(gas=True))
+  planner.lateral_acceleration = 0.0          # never reached a corner
+  settle_override(controller)
+
+  assert controller.curvature_data["0.02"]["average"] == pytest.approx(prior + CSC_NUDGE)
+
+
+def test_res_button_nudges_bucket_up_even_at_target_speed():
+  _, controller = make_controller()
+  prior = controller.learned_lat_accel(0.02)
+  controller.target = 20.0  # car tracking the target, so the gas-press condition would not fire
+
+  controller.handle_override(20.0, True, make_sm(), accel_button=True)
+  settle_override(controller)
+
+  assert controller.curvature_data["0.02"]["count"] == CSC_NUDGE_WEIGHT
+  assert controller.curvature_data["0.02"]["average"] > prior
+
+
+def test_brake_override_nudges_bucket_down():
+  _, controller = make_controller(driving_in_curve=True)
+  prior = controller.learned_lat_accel(0.02)
+
+  controller.handle_override(20.0, True, make_sm(brake=True))
+
+  assert controller.curvature_data["0.02"]["count"] == CSC_NUDGE_WEIGHT
+  assert controller.curvature_data["0.02"]["average"] < prior
+
+
+def test_calibrated_lateral_acceleration_param_is_written_on_flush():
+  planner, controller = make_controller(curvature_data={"0.02": {"average": 2.8, "count": 5000}})
+
+  assert "CalibratedLateralAcceleration" not in planner.params.values
+  controller.flush_data()
+
+  assert planner.params.values["CalibratedLateralAcceleration"] > DEFAULT_LATERAL_ACCELERATION
+  assert controller.lateral_acceleration == planner.params.values["CalibratedLateralAcceleration"]
+
+
+def test_stale_param_from_a_previous_build_is_republished_without_training():
+  # a stale value must not survive a restart just because this drive never trained
+  planner, controller = make_controller(curvature_data={"0.02": {"average": 2.8, "count": 5000}})
+  planner.params.values["CalibratedLateralAcceleration"] = 3.71
+
+  controller.log_data(0.0, make_sm())  # standstill: ineligible -> flush path
+
+  assert planner.params.values["CalibratedLateralAcceleration"] <= CSC_LAT_ACCEL_MAX

selfdrive/controls/tests/test_longitudinal_planner.py

@@ -526,6 +526,7 @@ def make_sm(v_ego: float, desired_accel: float, min_accel: float, *, experimenta
       forcingStop=False,
       redLight=False,
       forcingStopLength=2,
+      approachStopLength=0.0,
     ),
   }
 

selfdrive/controls/tests/test_starpilot_vcruise.py

@@ -5,8 +5,9 @@ import pytest
 from openpilot.common.constants import CV
 from openpilot.common.realtime import DT_MDL
 from openpilot.starpilot.common.starpilot_variables import PLANNER_TIME
-from openpilot.starpilot.controls.lib.curve_speed_controller import CSC_MAX_DECEL_RATE, CurveSpeedController
+from openpilot.starpilot.controls.lib.curve_speed_controller import CSC_GLOW_HOLD_TIME, CSC_GLOW_ON_DELTA
 from openpilot.starpilot.controls.lib.starpilot_vcruise import (
+  FORCE_STOP_CAP_SLACK_M,
   FORCE_STOP_TURN_VETO_STOP_SEEN_HOLD_TIME,
   STANDSTILL_FORCE_STOP_LIGHT_HOLD_TIME,
   StarPilotVCruise,
@@ -53,11 +54,14 @@ def make_vcruise(*, red_light=False, raw_model_stopped=False, forcing_stop=False
     raw_model_stopped=raw_model_stopped,
     road_curvature=road_curvature,
     road_curvature_detected=False,
+    lateral_acceleration=0.0,
   )
   vcruise = StarPilotVCruise(planner)
   vcruise.forcing_stop = forcing_stop
   vcruise.force_stop_timer = 1.0 if forcing_stop else 0.0
   vcruise.tracked_model_length = 0.0 if forcing_stop else planner.model_length
+  # what the not-committed branch would have left behind on the frame before commit
+  vcruise.force_stop_distance_cap = planner.model_length
   return planner, vcruise
 
 
@@ -80,12 +84,12 @@ def make_sm(*, standstill=True, min_steer_speed=0.0, car_fingerprint=""):
   }
 
 
-def update_vcruise(vcruise, sm, toggles, *, now, v_ego=0.0, controls_enabled=True):
+def update_vcruise(vcruise, sm, toggles, *, now, v_ego=0.0, v_cruise=20.0, controls_enabled=True):
   return vcruise.update(
     controls_enabled=controls_enabled,
     now=now,
     time_validated=True,
-    v_cruise=20.0,
+    v_cruise=v_cruise,
     v_ego=v_ego,
     sm=sm,
     starpilot_toggles=toggles,
@@ -143,30 +147,56 @@ def test_santa_fe_force_stop_tune_only_applies_to_that_car():
   assert get_force_stop_low_speed_hold(other) is None
 
 
-def test_curve_speed_controller_holds_target_through_brief_detector_dropout():
+def test_curve_speed_controller_blinker_releases_the_cap_but_keeps_the_plan():
   planner, vcruise = make_vcruise()
   sm = make_sm(standstill=False)
   toggles = make_toggles()
   toggles.curve_speed_controller = True
 
-  def set_curve_target(_v_ego):
-    vcruise.csc.target_set = True
+  calls = []
+
+  def set_curve_target(_v_ego, _v_cruise):
+    calls.append(_v_ego)
     vcruise.csc.target = 14.0
 
   vcruise.csc.update_target = set_curve_target
-  planner.road_curvature_detected = True
   result = update_vcruise(vcruise, sm, toggles, now=10.0, v_ego=20.0)
   assert result == pytest.approx(14.0)
   assert vcruise.csc_controlling_speed
 
-  planner.road_curvature_detected = False
+  # the cap lifts so CSC can't fight the lane change, but the envelope keeps planning
+  # so the curve doesn't have to be re-discovered from the set speed afterwards
+  sm["carState"].leftBlinker = True
   result = update_vcruise(vcruise, sm, toggles, now=10.25, v_ego=20.0)
+  assert result == pytest.approx(20.0)
+  assert not vcruise.csc_controlling_speed
+  assert len(calls) == 2  # still planning, so nothing has to be rediscovered
+
+  # blinker off: the plan is already current, so the cap comes straight back
+  sm["carState"].leftBlinker = False
+  result = update_vcruise(vcruise, sm, toggles, now=10.5, v_ego=20.0)
   assert result == pytest.approx(14.0)
   assert vcruise.csc_controlling_speed
 
-  result = update_vcruise(vcruise, sm, toggles, now=10.8, v_ego=20.0)
-  assert result == pytest.approx(20.0)
+
+def test_curve_speed_controller_reseeds_after_a_real_dropout():
+  planner, vcruise = make_vcruise()
+  sm = make_sm(standstill=False)
+  toggles = make_toggles()
+  toggles.curve_speed_controller = True
+
+  def set_curve_target(_v_ego, _v_cruise):
+    vcruise.csc.target = 14.0
+
+  vcruise.csc.update_target = set_curve_target
+  update_vcruise(vcruise, sm, toggles, now=11.0, v_ego=20.0)
+  assert vcruise.csc_controlling_speed
+
+  # disengaging is a real dropout, not a momentary veto -- that still resets
+  sm["carControl"].longActive = False
+  update_vcruise(vcruise, sm, toggles, now=11.05, v_ego=20.0)
   assert not vcruise.csc_controlling_speed
+  assert vcruise.csc.seed_pending
 
 
 def test_curve_speed_controller_releases_immediately_when_disabled():
@@ -175,16 +205,13 @@ def test_curve_speed_controller_releases_immediately_when_disabled():
   toggles = make_toggles()
   toggles.curve_speed_controller = True
 
-  def set_curve_target(_v_ego):
-    vcruise.csc.target_set = True
+  def set_curve_target(_v_ego, _v_cruise):
     vcruise.csc.target = 14.0
 
   vcruise.csc.update_target = set_curve_target
-  planner.road_curvature_detected = True
   update_vcruise(vcruise, sm, toggles, now=20.0, v_ego=20.0)
   assert vcruise.csc_controlling_speed
 
-  planner.road_curvature_detected = False
   toggles.curve_speed_controller = False
   result = update_vcruise(vcruise, sm, toggles, now=20.1, v_ego=20.0)
   assert result == pytest.approx(20.0)
@@ -198,12 +225,10 @@ def test_curve_speed_controller_can_be_limited_to_driving_without_a_lead():
   toggles.curve_speed_controller = True
   toggles.csc_no_lead = True
 
-  def set_curve_target(_v_ego):
-    vcruise.csc.target_set = True
+  def set_curve_target(_v_ego, _v_cruise):
     vcruise.csc.target = 14.0
 
   vcruise.csc.update_target = set_curve_target
-  planner.road_curvature_detected = True
 
   result = update_vcruise(vcruise, sm, toggles, now=30.0, v_ego=20.0)
   assert result == pytest.approx(14.0)
@@ -221,10 +246,8 @@ def test_curve_speed_controller_stays_enabled_with_a_lead_by_default():
   toggles = make_toggles()
   toggles.curve_speed_controller = True
   planner.starpilot_following.following_lead = True
-  planner.road_curvature_detected = True
 
-  def set_curve_target(_v_ego):
-    vcruise.csc.target_set = True
+  def set_curve_target(_v_ego, _v_cruise):
     vcruise.csc.target = 14.0
 
   vcruise.csc.update_target = set_curve_target
@@ -274,54 +297,276 @@ def test_curve_speed_controller_persists_data_after_leaving_curve():
   assert any(key == "CurvatureData" for key, _ in planner.params.writes)
 
 
-def test_curve_speed_controller_publishes_live_values_to_memory_params():
-  planner, vcruise = make_vcruise(road_curvature=0.02)
+def test_csc_res_press_cancels_for_episode_and_rearms():
+  planner, vcruise = make_vcruise()
   sm = make_sm(standstill=False)
-  sm["carControl"].longActive = False
+  toggles = make_toggles()
+  toggles.curve_speed_controller = True
+
+  curve_target = {"v": 14.0}
+
+  def set_curve_target(_v_ego, _v_cruise):
+    vcruise.csc.target = curve_target["v"]
+
+  vcruise.csc.update_target = set_curve_target
+  result = update_vcruise(vcruise, sm, toggles, now=60.0, v_ego=20.0)
+  assert result == pytest.approx(14.0)
+  assert vcruise.csc_controlling_speed
+
+  sm["starpilotCarState"].accelPressed = True
+  result = update_vcruise(vcruise, sm, toggles, now=60.05, v_ego=20.0)
+  assert result == pytest.approx(20.0)
+  assert not vcruise.csc_controlling_speed
+  assert vcruise.csc_override
+
+  # latches for the rest of the episode, not just while pressed
+  sm["starpilotCarState"].accelPressed = False
+  result = update_vcruise(vcruise, sm, toggles, now=60.1, v_ego=20.0)
+  assert result == pytest.approx(20.0)
+  assert vcruise.csc_override
+
+  # curve ends -> re-arms
+  curve_target["v"] = 20.0
+  update_vcruise(vcruise, sm, toggles, now=60.15, v_ego=20.0)
+  assert not vcruise.csc_override
+
+  curve_target["v"] = 14.0
+  result = update_vcruise(vcruise, sm, toggles, now=60.2, v_ego=20.0)
+  assert result == pytest.approx(14.0)
+  assert vcruise.csc_controlling_speed
+
+
+def test_csc_res_press_does_not_latch_when_csc_was_not_active():
+  planner, vcruise = make_vcruise()
+  sm = make_sm(standstill=False)
+  toggles = make_toggles()
+  toggles.curve_speed_controller = True
+
+  def set_curve_target(_v_ego, _v_cruise):
+    vcruise.csc.target = 14.0
+
+  vcruise.csc.update_target = set_curve_target
+  # press before CSC ever limited: suspends it while held, but must not latch a cancel
+  sm["starpilotCarState"].accelPressed = True
+  result = update_vcruise(vcruise, sm, toggles, now=70.0, v_ego=20.0)
+  assert result == pytest.approx(20.0)
+  assert not vcruise.csc_override
+
+  sm["starpilotCarState"].accelPressed = False
+  result = update_vcruise(vcruise, sm, toggles, now=70.05, v_ego=20.0)
+  assert result == pytest.approx(14.0)
+  assert vcruise.csc_controlling_speed
+
+
+def test_csc_res_press_defers_to_slc_confirmation():
+  planner, vcruise = make_vcruise()
+  sm = make_sm(standstill=False)
+  toggles = make_toggles()
+  toggles.curve_speed_controller = True
+
+  def set_curve_target(_v_ego, _v_cruise):
+    vcruise.csc.target = 14.0
+
+  vcruise.csc.update_target = set_curve_target
+  update_vcruise(vcruise, sm, toggles, now=80.0, v_ego=20.0)
+  assert vcruise.csc_controlling_speed
+
+  # confirming a speed limit must not also cancel the curve slowdown
+  vcruise.slc.speed_limit_changed_timer = 1.0
+  vcruise.slc.unconfirmed_speed_limit = 25.0
+  sm["starpilotCarState"].accelPressed = True
+  update_vcruise(vcruise, sm, toggles, now=80.05, v_ego=20.0)
+  assert not vcruise.csc_override
+
+  sm["starpilotCarState"].accelPressed = False
+  result = update_vcruise(vcruise, sm, toggles, now=80.1, v_ego=20.0)
+  assert result == pytest.approx(14.0)
+  assert vcruise.csc_controlling_speed
+
+
+def test_curve_speed_controller_glow_ignores_a_trivial_graze():
+  planner, vcruise = make_vcruise()
+  sm = make_sm(standstill=False)
+  toggles = make_toggles()
+  toggles.curve_speed_controller = True
+
+  # a long gentle bend where the envelope only shaves a little: the target hovers either
+  # side of the threshold for the whole curve, so a low bar strobes the glow
+  def set_curve_target(_v_ego, _v_cruise):
+    vcruise.csc.target = 20.0 - (CSC_GLOW_ON_DELTA / 2.0)
+
+  vcruise.csc.update_target = set_curve_target
+  result = update_vcruise(vcruise, sm, toggles, now=160.0, v_ego=20.0)
+
+  assert result < 20.0                      # the cap is still applied
+  assert not vcruise.csc_controlling_speed  # it just isn't worth announcing
+
+
+def test_curve_speed_controller_glow_holds_through_a_brief_release():
+  planner, vcruise = make_vcruise()
+  sm = make_sm(standstill=False)
+  toggles = make_toggles()
+  toggles.curve_speed_controller = True
+
+  curve_target = {"v": 14.0}
+
+  def set_curve_target(_v_ego, _v_cruise):
+    vcruise.csc.target = curve_target["v"]
+
+  vcruise.csc.update_target = set_curve_target
+  update_vcruise(vcruise, sm, toggles, now=130.0, v_ego=20.0)
+  assert vcruise.csc_controlling_speed
+
+  # one curve routinely lets go and re-engages; the glow must ride through it
+  curve_target["v"] = 20.0
+  now = 130.0
+  for _ in range(int((CSC_GLOW_HOLD_TIME - 0.2) / DT_MDL)):
+    now += DT_MDL
+    update_vcruise(vcruise, sm, toggles, now=now, v_ego=20.0)
+    assert vcruise.csc_controlling_speed
+
+  curve_target["v"] = 14.0
+  now += DT_MDL
+  update_vcruise(vcruise, sm, toggles, now=now, v_ego=20.0)
+  assert vcruise.csc_controlling_speed
+
+
+def test_curve_speed_controller_glow_clears_once_the_release_sticks():
+  planner, vcruise = make_vcruise()
+  sm = make_sm(standstill=False)
+  toggles = make_toggles()
+  toggles.curve_speed_controller = True
+
+  curve_target = {"v": 14.0}
+
+  def set_curve_target(_v_ego, _v_cruise):
+    vcruise.csc.target = curve_target["v"]
+
+  vcruise.csc.update_target = set_curve_target
+  update_vcruise(vcruise, sm, toggles, now=140.0, v_ego=20.0)
+  assert vcruise.csc_controlling_speed
+
+  curve_target["v"] = 20.0
+  now = 140.0
+  for _ in range(int(CSC_GLOW_HOLD_TIME / DT_MDL) + 1):
+    now += DT_MDL
+    update_vcruise(vcruise, sm, toggles, now=now, v_ego=20.0)
+  assert not vcruise.csc_controlling_speed
+
+
+def test_curve_speed_controller_keeps_the_cap_when_signalling_mid_curve():
+  planner, vcruise = make_vcruise()
+  sm = make_sm(standstill=False)
+  toggles = make_toggles()
+  toggles.curve_speed_controller = True
+
+  def set_curve_target(_v_ego, _v_cruise):
+    vcruise.csc.target = 14.0
+
+  vcruise.csc.update_target = set_curve_target
+  result = update_vcruise(vcruise, sm, toggles, now=150.0, v_ego=20.0)
+  assert result == pytest.approx(14.0)
+
+  # a lane change taken inside a curve must not hand the speed back
   planner.driving_in_curve = True
-  planner.lateral_acceleration = 2.4
-  vcruise.csc.training_timer = PLANNER_TIME
+  sm["carState"].leftBlinker = True
+  result = update_vcruise(vcruise, sm, toggles, now=150.05, v_ego=20.0)
+  assert result == pytest.approx(14.0)
+  assert vcruise.csc_controlling_speed
 
-  vcruise.csc.log_data(20.0, sm)
-
-  assert any(key == "CalibratedLateralAcceleration" for key, _ in planner.params_memory.writes)
-  assert any(key == "CalibrationProgress" for key, _ in planner.params_memory.writes)
-  assert planner.params_memory.values["CalibrationProgress"] > 0.0
+  # on a straight it still yields, so CSC can't fight the manoeuvre
+  planner.driving_in_curve = False
+  result = update_vcruise(vcruise, sm, toggles, now=150.1, v_ego=20.0)
+  assert result == pytest.approx(20.0)
+  assert not vcruise.csc_controlling_speed
 
 
-def test_curve_speed_controller_ramps_toward_curve_speed_at_bounded_rate():
-  planner = SimpleNamespace(
-    params=FakeParams(),
-    road_curvature=0.004,
-    time_to_curve=2.0,
-    starpilot_weather=SimpleNamespace(weather_id=0, reduce_lateral_acceleration=0.0),
-  )
-  controller = CurveSpeedController(SimpleNamespace(starpilot_planner=planner))
-  controller.lateral_acceleration = 2.0
-  controller.target_set = True
-  controller.target = 30.0
+def test_curve_speed_controller_glow_lights_when_the_car_arrives_at_the_cap_from_below():
+  planner, vcruise = make_vcruise()
+  sm = make_sm(standstill=False)
+  toggles = make_toggles()
+  toggles.curve_speed_controller = True
 
-  controller.update_target(30.0)
+  # accelerating out of a slow zone into a curve: the target is never under v_ego, but it
+  # is still the only thing stopping the car from reaching the set speed
+  def set_curve_target(_v_ego, _v_cruise):
+    vcruise.csc.target = 22.0
 
-  assert controller.target == pytest.approx(30.0 - CSC_MAX_DECEL_RATE * DT_MDL)
-  assert controller.target > (controller.lateral_acceleration / planner.road_curvature) ** 0.5
+  vcruise.csc.update_target = set_curve_target
+
+  update_vcruise(vcruise, sm, toggles, now=120.0, v_ego=15.0, v_cruise=32.0)
+  assert not vcruise.csc_controlling_speed          # still climbing, CSC isn't holding it yet
+
+  result = update_vcruise(vcruise, sm, toggles, now=120.05, v_ego=22.0, v_cruise=32.0)
+  assert result == pytest.approx(22.0)
+  assert vcruise.csc_controlling_speed              # arrived at the cap, and it binds
 
 
-def test_curve_speed_controller_does_not_slow_for_curve_speed_above_ego():
-  planner = SimpleNamespace(
-    params=FakeParams(),
-    road_curvature=0.001,
-    time_to_curve=2.0,
-    starpilot_weather=SimpleNamespace(weather_id=0, reduce_lateral_acceleration=0.0),
-  )
-  controller = CurveSpeedController(SimpleNamespace(starpilot_planner=planner))
-  controller.lateral_acceleration = 2.0
-  controller.target_set = True
-  controller.target = 28.0
+def test_curve_speed_controller_glow_stays_off_while_the_target_is_above_v_ego():
+  planner, vcruise = make_vcruise()
+  sm = make_sm(standstill=False)
+  toggles = make_toggles()
+  toggles.curve_speed_controller = True
 
-  controller.update_target(30.0)
+  # a highway sweeper trims the target well under the set speed but never under v_ego,
+  # so the car keeps accelerating and the driver feels nothing
+  def set_curve_target(_v_ego, _v_cruise):
+    vcruise.csc.target = 26.0
+
+  vcruise.csc.update_target = set_curve_target
+  result = update_vcruise(vcruise, sm, toggles, now=90.0, v_ego=20.0, v_cruise=30.0)
+
+  assert result == pytest.approx(26.0)
+  assert not vcruise.csc_controlling_speed
+
+
+def test_curve_speed_controller_glow_holds_through_the_recovery_ramp():
+  planner, vcruise = make_vcruise()
+  sm = make_sm(standstill=False)
+  toggles = make_toggles()
+  toggles.curve_speed_controller = True
+
+  curve_target = {"v": 14.0}
+
+  def set_curve_target(_v_ego, _v_cruise):
+    vcruise.csc.target = curve_target["v"]
+
+  vcruise.csc.update_target = set_curve_target
+  update_vcruise(vcruise, sm, toggles, now=100.0, v_ego=20.0)
+  assert vcruise.csc_controlling_speed
+
+  # past the apex the target climbs back above v_ego while the car is still cornering
+  curve_target["v"] = 18.0
+  update_vcruise(vcruise, sm, toggles, now=100.05, v_ego=15.0)
+  assert vcruise.csc_controlling_speed
+
+  # fully released, but the glow only clears once the release has stuck
+  curve_target["v"] = 20.0
+  now = 100.1
+  update_vcruise(vcruise, sm, toggles, now=now, v_ego=17.0)
+  assert vcruise.csc_controlling_speed
+
+  for _ in range(int(CSC_GLOW_HOLD_TIME / DT_MDL) + 1):
+    now += DT_MDL
+    update_vcruise(vcruise, sm, toggles, now=now, v_ego=17.0)
+  assert not vcruise.csc_controlling_speed
+
+
+def test_curve_speed_controller_hysteresis_keeps_glow_off_for_marginal_targets():
+  planner, vcruise = make_vcruise()
+  sm = make_sm(standstill=False)
+  toggles = make_toggles()
+  toggles.curve_speed_controller = True
+
+  def set_curve_target(_v_ego, _v_cruise):
+    vcruise.csc.target = 19.7
+
+  vcruise.csc.update_target = set_curve_target
+  result = update_vcruise(vcruise, sm, toggles, now=50.0, v_ego=20.0)
+
+  assert result == pytest.approx(19.7)
+  assert not vcruise.csc_controlling_speed
 
-  assert controller.target == pytest.approx(30.0)
 
 
 def test_active_slc_control_target_applies_offset_and_cluster_diff():
@@ -522,14 +767,15 @@ def test_force_stop_stays_committed_while_moving_even_if_scene_opens():
 
 def test_force_stop_reanchors_when_model_reopens_path_without_stop_action():
   planner, vcruise = make_vcruise(red_light=False, raw_model_stopped=False, forcing_stop=True)
-  planner.model_length = 40.0
-  vcruise.tracked_model_length = 10.0
+  planner.model_length = 90.0
+  vcruise.tracked_model_length = 60.0
+  vcruise.force_stop_distance_cap = 90.0
   sm = make_sm(standstill=False)
   sm["modelV2"] = SimpleNamespace(action=SimpleNamespace(shouldStop=False))
 
   result = update_vcruise(vcruise, sm, make_toggles(), now=0.0, v_ego=1.5)
 
-  assert vcruise.tracked_model_length == pytest.approx(40.0)
+  assert vcruise.tracked_model_length == pytest.approx(90.0)
   assert result > 5.0
 
 
@@ -561,6 +807,49 @@ def test_santa_fe_force_stop_holds_through_low_speed_detector_dropout():
   assert result == pytest.approx(0.0)
 
 
+def test_force_stop_does_not_reanchor_inside_reanchor_floor():
+  planner, vcruise = make_vcruise(red_light=False, raw_model_stopped=False, forcing_stop=True)
+  planner.model_length = 90.0
+  vcruise.tracked_model_length = 25.0
+  sm = make_sm(standstill=False)
+  sm["modelV2"] = SimpleNamespace(action=SimpleNamespace(shouldStop=False))
+
+  update_vcruise(vcruise, sm, make_toggles(), now=0.0, v_ego=1.5)
+
+  assert vcruise.tracked_model_length < 25.0
+
+
+def test_force_stop_reanchor_bounded_by_distance_driven():
+  # The line can't recede: a ballooning horizon may not push the stop past where it was at
+  # commit minus the distance driven since.
+  planner, vcruise = make_vcruise(red_light=False, raw_model_stopped=False, forcing_stop=True)
+  planner.model_length = 200.0
+  vcruise.tracked_model_length = 60.0
+  vcruise.force_stop_distance_cap = 70.0
+  sm = make_sm(standstill=False)
+  sm["modelV2"] = SimpleNamespace(action=SimpleNamespace(shouldStop=False))
+
+  update_vcruise(vcruise, sm, make_toggles(), now=0.0, v_ego=10.0)
+
+  assert vcruise.tracked_model_length <= 70.0 + FORCE_STOP_CAP_SLACK_M
+  assert vcruise.tracked_model_length < 100.0  # nowhere near the 200 m the horizon claimed
+
+
+def test_force_stop_cap_slack_tapers_near_the_line():
+  # Slack protects against an under-read at commit; held near the line it would just aim the
+  # solver that far past the stop bar.
+  planner, vcruise = make_vcruise(red_light=False, raw_model_stopped=False, forcing_stop=True)
+  planner.model_length = 200.0
+  vcruise.tracked_model_length = 60.0
+  vcruise.force_stop_distance_cap = 12.0
+  sm = make_sm(standstill=False)
+  sm["modelV2"] = SimpleNamespace(action=SimpleNamespace(shouldStop=False))
+
+  update_vcruise(vcruise, sm, make_toggles(), now=0.0, v_ego=5.0)
+
+  assert vcruise.tracked_model_length < 12.0 + FORCE_STOP_CAP_SLACK_M / 2.0
+
+
 def test_force_stop_does_not_reanchor_committed_model_stop():
   planner, vcruise = make_vcruise(red_light=False, raw_model_stopped=False, forcing_stop=True)
   planner.model_length = 40.0

selfdrive/controls/tests/test_turn_lead.py

@@ -1,13 +1,34 @@
+import math
+import types
+
 from cereal import car
 
 import pytest
 
-from openpilot.selfdrive.controls.controlsd import get_control_lateral_smooth_seconds, turn_lead_allowed
+from openpilot.selfdrive.controls.controlsd import (TWITCH_GUARD_FLOOR, TWITCH_GUARD_MAX_SPEED,
+                                                    get_control_lateral_smooth_seconds,
+                                                    limit_curvature_to_plan, turn_lead_allowed)
 
 
 LateralControlMode = car.CarControl.Actuators.LateralControlMode
 
 
+def _plan(xs, ys):
+  return types.SimpleNamespace(position=types.SimpleNamespace(x=xs, y=ys))
+
+
+def _arc_plan(radius, n=200):
+  # constant-radius arc, ~1 rad of heading — long enough to clear the reach gate
+  return _plan([radius * math.sin(i / n) for i in range(n)],
+               [radius * (1.0 - math.cos(i / n)) for i in range(n)])
+
+
+STRAIGHT_PLAN = _plan([i * 0.5 for i in range(200)], [0.0] * 200)  # 100 m dead straight
+STANDSTILL_STUB_PLAN = _plan([0.0, 0.3], [0.0, 0.0])
+TURN_PLAN = _arc_plan(30.0)          # 0.033 1/m (~81 deg of wheel), 25 m of reach
+GENTLE_BEND_PLAN = _arc_plan(143.0)  # 0.007 1/m, barely bending
+
+
 def test_turn_lead_is_suppressed_only_during_applied_angle_control():
   assert not turn_lead_allowed("rivian", LateralControlMode.angle)
   assert turn_lead_allowed("rivian", LateralControlMode.torque)
@@ -37,3 +58,46 @@ def test_subaru_control_smoothing_uses_vehicle_schedule(v_ego, expected):
 ])
 def test_rivian_control_smoothing_remains_speed_scheduled(v_ego, expected):
   assert get_control_lateral_smooth_seconds("rivian", v_ego, 0.4) == pytest.approx(expected)
+
+
+@pytest.mark.parametrize("curvature", [0.0155, -0.0155])
+def test_twitch_against_a_straight_plan_is_clamped_to_the_floor(curvature):
+  guarded = limit_curvature_to_plan(STRAIGHT_PLAN, curvature, 1.2)
+  assert abs(guarded) == pytest.approx(TWITCH_GUARD_FLOOR)
+  assert math.copysign(1.0, guarded) == math.copysign(1.0, curvature)
+
+
+def test_command_already_below_the_floor_is_untouched():
+  assert limit_curvature_to_plan(STRAIGHT_PLAN, 0.0015, 1.2) == pytest.approx(0.0015)
+
+
+@pytest.mark.parametrize("v_ego", [TWITCH_GUARD_MAX_SPEED, 6.0, 30.0])
+def test_guard_is_inactive_above_its_speed_band(v_ego):
+  assert limit_curvature_to_plan(STRAIGHT_PLAN, 0.0155, v_ego) == pytest.approx(0.0155)
+
+
+def test_guard_fades_out_across_the_speed_band():
+  full = limit_curvature_to_plan(STRAIGHT_PLAN, 0.0155, 1.2)
+  half = limit_curvature_to_plan(STRAIGHT_PLAN, 0.0155, 3.5)
+  assert full < half < 0.0155
+
+
+# turning authority must never be reduced: a real turn's action agrees with its own plan
+@pytest.mark.parametrize("ratio", [0.8, 1.0, 2.0, 3.0])
+def test_real_turns_tracking_their_own_plan_are_untouched(ratio):
+  action = (1.0 / 30.0) * ratio
+  assert limit_curvature_to_plan(TURN_PLAN, action, 1.2) == pytest.approx(action)
+
+
+def test_a_barely_bending_plan_does_not_license_a_large_command():
+  guarded = limit_curvature_to_plan(GENTLE_BEND_PLAN, 0.0155, 1.2)
+  assert TWITCH_GUARD_FLOOR < guarded < 0.008
+
+
+@pytest.mark.parametrize("plan", [STANDSTILL_STUB_PLAN, _plan([], [])])
+def test_guard_stands_down_when_the_plan_is_too_short_to_judge(plan):
+  assert limit_curvature_to_plan(plan, 0.0155, 0.4) == pytest.approx(0.0155)
+
+
+def test_zero_command_stays_zero():
+  assert limit_curvature_to_plan(STRAIGHT_PLAN, 0.0, 1.2) == 0.0

selfdrive/ui/layouts/settings/starpilot/longitudinal.py

@@ -686,12 +686,12 @@ class StarPilotLongitudinalLayout(_SettingsPage):
     self._curve_speed_controller_rows = [
       SettingRow("CalibratedLatAccel", "value", tr_noop("Calibrated Lateral Accel"),
                  subtitle=tr_noop("The learned lateral acceleration from collected driving data. Higher values allow faster cornering."),
-                 get_value=lambda: f"{self._params_memory.get_float('CalibratedLateralAcceleration'):.2f} m/s",
+                 get_value=lambda: f"{self._params.get_float('CalibratedLateralAcceleration'):.2f} m/s",
                  on_click=None,
                  visible=csc_on),
       SettingRow("CalibrationProgress", "value", tr_noop("Calibration Progress"),
                  subtitle=tr_noop("How much curve data has been collected. Normal for the value to stay low."),
-                 get_value=lambda: f"{self._params_memory.get_float('CalibrationProgress'):.2f}%",
+                 get_value=lambda: f"{self._params.get_float('CalibrationProgress'):.2f}%",
                  on_click=None,
                  visible=csc_on),
       SettingRow("ResetCurve", "action", tr_noop("Reset Curve Data"),

selfdrive/ui/onroad/starpilot/starpilot_border.py

@@ -57,7 +57,9 @@ def _csc_state():
 
   plan = sm["starpilotPlan"]
   params = ui_state.ui_params
-  if plan.speedLimitChanged or not params.get_bool("ShowCSCStatus"):
+  # A pending speed limit flashes the speed limit sign, not the border -- it has no reason
+  # to blank this, and doing so hid real curve slowdowns for the whole confirmation window.
+  if not params.get_bool("ShowCSCStatus"):
     return None
 
   car_state = sm["carState"]

starpilot/common/accel_profile.py

@@ -3,8 +3,6 @@ from __future__ import annotations
 
 import math
 
-from openpilot.selfdrive.controls.lib.longitudinal_planner import get_max_accel
-
 ACCELERATION_PROFILES = {
   "STANDARD": 0,
   "ECO": 1,

starpilot/common/assets/device_settings_layout.json

@@ -1043,6 +1043,30 @@
         "parent_key": "CurveSpeedController",
         "settings_tier": "simple"
       },
+      {
+        "key": "CalibratedLateralAcceleration",
+        "label": "Calibrated Lateral Accel",
+        "description": "The learned lateral acceleration from collected driving data. Higher values allow faster cornering.",
+        "picker_description": "Learned cornering comfort from your driving data.",
+        "data_type": "float",
+        "ui_type": "readout",
+        "precision": 2,
+        "unit": " m/s²",
+        "parent_key": "CurveSpeedController",
+        "settings_tier": "simple"
+      },
+      {
+        "key": "CalibrationProgress",
+        "label": "Calibration Progress",
+        "description": "How much curve data has been collected. Normal for the value to stay low.",
+        "picker_description": "How much curve data has been collected.",
+        "data_type": "float",
+        "ui_type": "readout",
+        "precision": 2,
+        "unit": "%",
+        "parent_key": "CurveSpeedController",
+        "settings_tier": "simple"
+      },
       {
         "key": "ResetCurveData",
         "label": "Reset Curve Data",

starpilot/common/starpilot_utilities.py

@@ -138,6 +138,23 @@ def calculate_road_curvature(modelData, v_ego):
   return float(predicted_lateral_acc / max(v_ego, 1)**2), max(time_to_curve, 1)
 
 
+PROFILE_MIN_SPEED = 3.0     # m/s — model points planned near standstill have unusable curvature
+PROFILE_MAX_CURVATURE = 0.1
+
+
+def extract_curve_profile(modelData):
+  orientation_rate = np.abs(np.array(modelData.orientationRate.z))
+  velocity = np.array(modelData.velocity.x)
+  distances = np.array(modelData.position.x)
+
+  # k = psi_dot / v per point, against the model's own planned speed so its
+  # slowdowns don't inflate the curvature
+  curvatures = orientation_rate / np.clip(velocity, PROFILE_MIN_SPEED, None)
+  curvatures = np.where(velocity < PROFILE_MIN_SPEED, 0.0, np.minimum(curvatures, PROFILE_MAX_CURVATURE))
+
+  return curvatures, distances
+
+
 def clean_model_name(name):
   return name.replace("(Default)", "").strip()
 

starpilot/controls/lib/curve_speed_controller.py

@@ -2,18 +2,97 @@
 import numpy as np
 
 from openpilot.common.constants import CV
+from openpilot.common.filter_simple import FirstOrderFilter
 from openpilot.common.realtime import DT_MDL
 
-from openpilot.starpilot.common.starpilot_variables import CITY_SPEED_LIMIT, CRUISING_SPEED, DEFAULT_LATERAL_ACCELERATION, PLANNER_TIME
+from openpilot.starpilot.common.starpilot_variables import (
+  CITY_SPEED_LIMIT,
+  CRUISING_SPEED,
+  DEFAULT_LATERAL_ACCELERATION,
+  PLANNER_TIME,
+)
 
 CALIBRATION_PROGRESS_THRESHOLD = 10 / DT_MDL
 CSC_MIN_SPEED = CITY_SPEED_LIMIT * CV.MPH_TO_MS
-CSC_MAX_DECEL_RATE = 1.5
-MAX_CURVATURE = 0.1
-MIN_CURVATURE = 0.001
-PERCENTILE = 90
-ROUNDING_PRECISION = 5
-STEP = 0.001
+
+# braking distance is (v^2 - v_curve^2) / (2 * this), so lower starts the slowdown
+# sooner and spreads it further.
+CSC_APPROACH_DECEL = 0.3
+CSC_TARGET_UP_RATE = 3.0
+CSC_TARGET_DOWN_RATE = 2.5
+CSC_TARGET_FILTER_RC = 0.4
+CSC_EGO_HEADROOM = 2.0            # target never trails below v_ego, so CSC can't drag re-acceleration
+CSC_RELEASE_DEBOUNCE = 0.25       # s the envelope must stay clear before that floor applies
+CSC_ACTIVE_ON_DELTA = 0.5
+CSC_ACTIVE_OFF_DELTA = 0.25
+CSC_GLOW_ON_DELTA = 1.0           # ~2.2 mph; separate from CSC_ACTIVE_ON_DELTA (training) so a trivial graze doesn't light the glow
+CSC_GLOW_HOLD_TIME = 3.0          # s the cap must stay released before the glow clears, so it doesn't flicker on/off across one curve
+
+CSC_COUNT_CAP = 600               # EMA floor: samples beyond this stop shrinking the update step
+CSC_PRIOR_COUNT = 100             # bucket count at which learned data and the prior have equal weight
+CSC_LAT_ACCEL_MIN = 1.2
+CSC_LAT_ACCEL_MAX = 3.2
+CSC_NUDGE = 0.15
+CSC_NUDGE_WEIGHT = 20             # counts a single override pseudo-sample is worth
+CSC_OVERRIDE_WATCH_TIME = 6.0     # s to keep watching what the driver holds after they reject a cut
+CSC_TRAINING_QUIET_TIME = 5.0     # blocks passive samples after CSC limited speed, so it can't learn its own cap
+CSC_TRAINING_SETTLE_TIME = 2.0    # driver-owned seconds before a sample counts, so it isn't openpilot's leftover speed
+CSC_COMFORT_MARGIN = 1.0          # 1.0 = matches the driver's own learned cornering, no extra cushion
+
+# The model under-reads curvature at range: measured 0.81x actual beyond ~75 m. That holds
+# only where the reading is already firm -- weak distant readings carry no usable magnitude
+# (0.40x median with a 14:1 spread), so scaling those would amplify noise, not signal.
+CSC_FARFIELD_MIN_CURVATURE = 0.004   # ~R 250 m; at this strength range readings were 85%+ reliable
+CSC_FARFIELD_MIN_DISTANCE = 30.0     # inside this the model is already accurate
+CSC_FARFIELD_GAIN = 1.23             # 1 / 0.81
+
+# Buckets are spaced geometrically, not linearly: comfort is a speed and v = sqrt(a/k), so equal
+# steps in k give wildly uneven speed resolution. Regridding is safe -- _normalize_curvature_data
+# re-buckets stored keys on load.
+MIN_CURVATURE = 0.0005            # R 2000 m — gentler than this never constrains anything
+MAX_CURVATURE = 0.02              # R 50 m — already well below the CSC_MIN_SPEED floor
+CURVATURE_BUCKETS = 24            # keeps every bucket under ~7 mph wide without over-thinning the data
+ROUNDING_PRECISION = 6
+CURVATURE_GRID = MIN_CURVATURE * np.power(MAX_CURVATURE / MIN_CURVATURE,
+                                          np.arange(CURVATURE_BUCKETS) / (CURVATURE_BUCKETS - 1))
+LOG_CURVATURE_GRID = np.log(CURVATURE_GRID)
+
+# Drivers accept more lateral acceleration in sharp slow corners than in highway sweepers.
+PRIOR_CURVATURE_BP = [0.001, 0.003, 0.01, 0.03, 0.1]
+PRIOR_LAT_ACCEL_V = [1.5, 1.8, 2.2, 2.6, 2.9]
+
+
+def weighted_isotonic(values, weights):
+  """Weighted non-decreasing fit (pool adjacent violators).
+
+  Keeps comfort from falling as curves tighten, without letting a sparse bucket
+  overrule a well-sampled neighbour the way a running maximum would.
+  """
+  block_values: list[float] = []
+  block_weights: list[float] = []
+  block_sizes: list[int] = []
+
+  for value, weight in zip(values, weights, strict=True):
+    block_values.append(float(value))
+    block_weights.append(float(weight))
+    block_sizes.append(1)
+
+    while len(block_values) > 1 and block_values[-2] > block_values[-1]:
+      merged_weight = block_weights[-2] + block_weights[-1]
+      merged_value = ((block_values[-2] * block_weights[-2]) + (block_values[-1] * block_weights[-1])) / merged_weight
+      block_values.pop()
+      block_weights.pop()
+      merged_size = block_sizes.pop()
+      block_values[-1] = merged_value
+      block_weights[-1] = merged_weight
+      block_sizes[-1] += merged_size
+
+  fitted = np.empty(len(values))
+  index = 0
+  for value, size in zip(block_values, block_sizes, strict=True):
+    fitted[index:index + size] = value
+    index += size
+  return fitted
 
 
 def is_user_overriding_longitudinal(sm):
@@ -42,26 +121,42 @@ class CurveSpeedController:
     self.starpilot_planner = StarPilotVCruise.starpilot_planner
 
     self.enable_training = False
-    self.target_set = False
+    self.nudge_applied = False
+
+    self.override_watch_key = None
+    self.override_watch_peak = 0.0
+    self.override_watch_timer = 0.0
 
     self.training_timer = 0.0
     self.persistence_timer = 0.0
+    self.training_quiet_timer = 0.0
     self.data_dirty = False
 
+    self.target = 0.0
+    self.binding_distance = 0.0
+    self.release_timer = 0.0
+    self.target_filter = FirstOrderFilter(0.0, CSC_TARGET_FILTER_RC, DT_MDL, initialized=False)
+    self.seed_pending = True
+
+    self._long_active_prev = False
+
     curvature_data = self.starpilot_planner.params.get("CurvatureData")
     self.curvature_data = self._normalize_curvature_data(curvature_data)
 
-    self.required_curvatures = [str(round(road_curvature, ROUNDING_PRECISION)) for road_curvature in np.arange(MIN_CURVATURE, MAX_CURVATURE + STEP, STEP)]
+    # built through the bucketer so the keys are byte-identical to what training writes
+    self.required_curvatures = [self._bucket_curvature(curvature) for curvature in CURVATURE_GRID]
 
-    self.update_lateral_acceleration()
-    self._publish_calibration_progress()
+    self.rebuild_lat_accel_curve()
+    # publish on the first flush even if this drive never trains, or the readout
+    # keeps showing whatever a previous build left behind
+    self.data_dirty = True
 
   @staticmethod
   def _bucket_curvature(road_curvature):
-    clipped_curvature = float(np.clip(road_curvature, MIN_CURVATURE, MAX_CURVATURE))
-    bucket_index = round((clipped_curvature - MIN_CURVATURE) / STEP)
-    bucketed_curvature = MIN_CURVATURE + (bucket_index * STEP)
-    return str(round(bucketed_curvature, ROUNDING_PRECISION))
+    clipped_curvature = float(np.clip(abs(road_curvature), MIN_CURVATURE, MAX_CURVATURE))
+    # nearest in log space, so a bucket is a constant speed step rather than a constant radius one
+    bucket_index = int(np.argmin(np.abs(LOG_CURVATURE_GRID - np.log(clipped_curvature))))
+    return str(round(float(CURVATURE_GRID[bucket_index]), ROUNDING_PRECISION))
 
   @classmethod
   def _normalize_curvature_data(cls, curvature_data):
@@ -103,42 +198,36 @@ class CurveSpeedController:
     if not self.data_dirty:
       return
 
-    progress = self._calibration_progress()
-    self.starpilot_planner.params.put_nonblocking("CalibrationProgress", progress)
-    self.starpilot_planner.params.put_nonblocking("CurvatureData", self.curvature_data)
-    self._put_memory_param("CalibrationProgress", progress)
-    self.data_dirty = False
-    self.persistence_timer = 0.0
-
-  def _calibration_progress(self):
     progress = 0.0
     for key in self.required_curvatures:
       if key in self.curvature_data:
         progress += min(self.curvature_data[key]["count"] / CALIBRATION_PROGRESS_THRESHOLD, 1.0)
-    return (progress / len(self.required_curvatures)) * 100
 
-  def _publish_calibration_progress(self):
-    self._put_memory_param("CalibrationProgress", self._calibration_progress())
-
-  def _put_memory_param(self, key, value):
-    params_memory = getattr(self.starpilot_planner, "params_memory", None)
-    if params_memory is not None:
-      params_memory.put_nonblocking(key, value)
+    self.starpilot_planner.params.put_nonblocking("CalibratedLateralAcceleration", self.lateral_acceleration)
+    self.starpilot_planner.params.put_nonblocking("CalibrationProgress", (progress / len(self.required_curvatures)) * 100)
+    self.starpilot_planner.params.put_nonblocking("CurvatureData", self.curvature_data)
+    self.data_dirty = False
+    self.persistence_timer = 0.0
 
   def flush_data(self):
     self._persist_data()
 
   def log_data(self, v_ego, sm):
+    self.training_quiet_timer = max(self.training_quiet_timer - DT_MDL, 0.0)
+
     eligible = (
       v_ego > CRUISING_SPEED and
       not self.starpilot_planner.tracking_lead and
-      is_manual_speed_control(sm)
+      is_manual_speed_control(sm) and
+      self.training_quiet_timer <= 0.0
     )
     self.enable_training = False
 
     if not eligible:
       self.flush_data()
-      self.training_timer = 0.0
+      # decay instead of resetting: a lead flickering in and out of the tracker used to
+      # cost the full re-arm, which left almost nothing to learn from on a real drive
+      self.training_timer = max(self.training_timer - DT_MDL, 0.0)
       self.persistence_timer = 0.0
       return
 
@@ -147,7 +236,7 @@ class CurveSpeedController:
       self.persistence_timer += DT_MDL
 
     in_curve = (
-      self.training_timer >= PLANNER_TIME and
+      self.training_timer >= CSC_TRAINING_SETTLE_TIME and
       self.starpilot_planner.driving_in_curve and
       not (sm["carState"].leftBlinker or sm["carState"].rightBlinker)
     )
@@ -157,11 +246,11 @@ class CurveSpeedController:
 
       if road_curvature in self.curvature_data:
         data = self.curvature_data[road_curvature]
-        average = data["average"]
-        count = data["count"]
+        # capped so an established bucket still tracks a change in driving style
+        effective_count = min(data["count"], CSC_COUNT_CAP)
         self.curvature_data[road_curvature] = {
-          "average": ((average * count) + lateral_acceleration) / (count + 1),
-          "count": count + 1
+          "average": ((data["average"] * effective_count) + lateral_acceleration) / (effective_count + 1),
+          "count": data["count"] + 1
         }
       else:
         self.curvature_data[road_curvature] = {
@@ -170,8 +259,7 @@ class CurveSpeedController:
         }
 
       self.data_dirty = True
-      self.update_lateral_acceleration()
-      self._publish_calibration_progress()
+      self.rebuild_lat_accel_curve()
       self.enable_training = True
 
       if self.persistence_timer >= PLANNER_TIME:
@@ -179,30 +267,166 @@ class CurveSpeedController:
     elif self.data_dirty:
       self.flush_data()
 
-  def update_lateral_acceleration(self):
-    if self.curvature_data:
-      all_samples = [data["average"] for data in self.curvature_data.values()]
-      self.lateral_acceleration = float(np.percentile(all_samples, PERCENTILE))
+  def handle_override(self, v_ego, was_controlling, sm, accel_button=False):
+    long_active = bool(sm["carControl"].longActive)
+    long_dropped = self._long_active_prev and not long_active
+    self._long_active_prev = long_active
+
+    self._update_override_watch(sm)
+
+    if not was_controlling:
+      self.nudge_applied = False
+      return
+
+    if self.nudge_applied:
+      return
+
+    if accel_button or (sm["carState"].gasPressed and self.target < v_ego - 0.5):
+      # Watch what the driver actually holds instead of stepping by a fixed amount -- CSC is
+      # suspended while overridden, so their cornering now measures their real comfort.
+      self.override_watch_key = self._bucket_curvature(abs(self.starpilot_planner.road_curvature))
+      self.override_watch_peak = abs(self.starpilot_planner.lateral_acceleration)
+      self.override_watch_timer = CSC_OVERRIDE_WATCH_TIME
+      self.nudge_applied = True
+    elif (getattr(sm["carState"], "brakePressed", False) or long_dropped) and self.starpilot_planner.driving_in_curve:
+      self._apply_nudge(-CSC_NUDGE)
+
+  def _update_override_watch(self, sm):
+    if self.override_watch_key is None:
+      return
+
+    lateral_acceleration = abs(self.starpilot_planner.lateral_acceleration)
+    if lateral_acceleration > self.override_watch_peak:
+      # credit the bucket the peak actually happened in, not the one at the button press
+      self.override_watch_peak = lateral_acceleration
+      self.override_watch_key = self._bucket_curvature(abs(self.starpilot_planner.road_curvature))
+
+    self.override_watch_timer -= DT_MDL
+    if self.override_watch_timer > 0.0 and (is_user_overriding_longitudinal(sm) or
+                                            self.starpilot_planner.driving_in_curve):
+      return
+
+    key = self.override_watch_key
+    self.override_watch_key = None
+    # floored at the old fixed step, so a rejection that never reaches a corner still counts
+    # and this path can only ever raise the bucket
+    self._record_pseudo_sample(key, max(self.override_watch_peak,
+                                        self.learned_lat_accel(float(key)) + CSC_NUDGE))
+
+  def _apply_nudge(self, offset):
+    key = self._bucket_curvature(abs(self.starpilot_planner.road_curvature))
+    # relative to the learned value, not the margined one, or repeated overrides walk the bucket down
+    self._record_pseudo_sample(key, self.learned_lat_accel(float(key)) + offset)
+    self.nudge_applied = True
+
+  def _record_pseudo_sample(self, key, sample):
+    sample = float(np.clip(sample, CSC_LAT_ACCEL_MIN, CSC_LAT_ACCEL_MAX))
+
+    data = self.curvature_data.get(key, {"average": sample, "count": 0})
+    effective_count = min(data["count"], CSC_COUNT_CAP)
+    total = effective_count + CSC_NUDGE_WEIGHT
+    self.curvature_data[key] = {
+      "average": ((data["average"] * effective_count) + (sample * CSC_NUDGE_WEIGHT)) / total,
+      "count": data["count"] + CSC_NUDGE_WEIGHT,
+    }
+
+    self.rebuild_lat_accel_curve()
+    self.data_dirty = True
+    self.flush_data()
+
+  def rebuild_lat_accel_curve(self):
+    grid_k = np.array([float(key) for key in self.required_curvatures])
+    prior = np.interp(grid_k, PRIOR_CURVATURE_BP, PRIOR_LAT_ACCEL_V)
+
+    blended = prior.copy()
+    counts = np.zeros(len(grid_k))
+    for i, key in enumerate(self.required_curvatures):
+      data = self.curvature_data.get(key)
+      if data:
+        confidence = data["count"] / (data["count"] + CSC_PRIOR_COUNT)
+        blended[i] = confidence * data["average"] + (1.0 - confidence) * prior[i]
+        counts[i] = data["count"]
+
+    blended = np.clip(blended, CSC_LAT_ACCEL_MIN, CSC_LAT_ACCEL_MAX)
+    blended = weighted_isotonic(blended, counts + CSC_PRIOR_COUNT)
+
+    self._curve_k = grid_k
+    self._curve_a = blended
+
+    if counts.sum() > 0:
+      self.lateral_acceleration = float(np.average(blended, weights=counts))
     else:
       self.lateral_acceleration = DEFAULT_LATERAL_ACCELERATION
 
-    self.starpilot_planner.params.put_nonblocking("CalibratedLateralAcceleration", self.lateral_acceleration)
-    self._put_memory_param("CalibratedLateralAcceleration", self.lateral_acceleration)
+  def learned_lat_accel(self, curvature):
+    """Comfort level learned for this curvature, before any control margin."""
+    return float(np.interp(abs(curvature), self._curve_k, self._curve_a))
 
-  def update_target(self, v_ego):
-    lateral_acceleration = self.lateral_acceleration
-    if self.starpilot_planner.starpilot_weather.weather_id != 0:
-      lateral_acceleration -= self.lateral_acceleration * self.starpilot_planner.starpilot_weather.reduce_lateral_acceleration
+  def lat_accel_for_curvature(self, curvature):
+    lat_accel = np.interp(np.abs(curvature), self._curve_k, self._curve_a) * CSC_COMFORT_MARGIN
 
-    if self.target_set:
-      csc_speed = (lateral_acceleration / abs(self.starpilot_planner.road_curvature))**0.5
-      csc_speed = max(float(csc_speed), CSC_MIN_SPEED)
-      if csc_speed >= v_ego:
-        self.target = v_ego
-      else:
-        time_to_curve = max(float(self.starpilot_planner.time_to_curve), DT_MDL)
-        decel_rate = float(np.clip((v_ego - csc_speed) / time_to_curve, 0.0, CSC_MAX_DECEL_RATE))
-        self.target = float(np.clip(self.target - decel_rate * DT_MDL, csc_speed, v_ego))
+    weather = self.starpilot_planner.starpilot_weather
+    if weather.weather_id != 0:
+      lat_accel = lat_accel * (1.0 - weather.reduce_lateral_acceleration)
+
+    return lat_accel
+
+  @staticmethod
+  def _correct_far_field(curvatures, distances):
+    """Undo the model's known under-read of distant curvature, where the reading is firm."""
+    firm = (curvatures >= CSC_FARFIELD_MIN_CURVATURE) & (distances >= CSC_FARFIELD_MIN_DISTANCE)
+    return np.minimum(np.where(firm, curvatures * CSC_FARFIELD_GAIN, curvatures), MAX_CURVATURE)
+
+  def reset(self, v_cruise):
+    self.target = float(v_cruise)
+    self.release_timer = 0.0
+    self.target_filter.x = float(v_cruise)
+    self.target_filter.initialized = True
+    self.seed_pending = True
+
+  def update_target(self, v_ego, v_cruise):
+    if not self.target_filter.initialized:
+      self.reset(v_cruise)
+
+    curvatures, distances = self.starpilot_planner.curve_profile
+    if len(curvatures) == 0:
+      raw_target = float(v_cruise)
+      self.binding_distance = 0.0
     else:
-      self.target_set = True
-      self.target = v_ego
+      curvatures = self._correct_far_field(curvatures, distances)
+      lat_accel = self.lat_accel_for_curvature(curvatures)
+      point_speeds = np.sqrt(lat_accel / np.maximum(curvatures, 1e-4))
+      point_speeds = np.maximum(point_speeds, CSC_MIN_SPEED)
+      allowed_speeds = np.sqrt(point_speeds**2 + 2.0 * CSC_APPROACH_DECEL * np.maximum(distances, 0.0))
+      binding_index = int(np.argmin(allowed_speeds))
+      raw_target = min(float(allowed_speeds[binding_index]), float(v_cruise))
+      self.binding_distance = float(distances[binding_index]) if raw_target < v_cruise else 0.0
+
+    # a fresh activation starts at the envelope, or it spends seconds ramping down
+    # toward a curve it already sees (engaging or launching into a turn)
+    if self.seed_pending:
+      seed = min(float(v_cruise), max(raw_target, v_ego + CSC_EGO_HEADROOM))
+      self.target = seed
+      self.target_filter.x = seed
+      self.seed_pending = False
+
+    if raw_target >= v_ego:
+      self.release_timer += DT_MDL
+    else:
+      self.release_timer = 0.0
+
+    # The headroom aim goes through the rate limiter with everything else; applying it
+    # after the clamp let every upward jitter in raw_target reach the target unsmoothed.
+    filtered = self.target_filter.update(raw_target)
+    self.target = float(np.clip(max(filtered, min(raw_target, v_ego + CSC_EGO_HEADROOM)),
+                                self.target - CSC_TARGET_DOWN_RATE * DT_MDL,
+                                self.target + CSC_TARGET_UP_RATE * DT_MDL))
+
+    # Once the envelope really has released, the target must not sit under the car or it
+    # drags re-acceleration. Debounced, because a single jittery frame doing this yanks a
+    # legitimate cut back up to v_ego and strobes the glow on sweepers.
+    if self.release_timer >= CSC_RELEASE_DEBOUNCE:
+      self.target = max(self.target, min(raw_target, v_ego))
+
+    if self.target < v_cruise - CSC_ACTIVE_ON_DELTA:
+      self.training_quiet_timer = CSC_TRAINING_QUIET_TIME

starpilot/controls/lib/starpilot_vcruise.py

@@ -6,7 +6,13 @@ from openpilot.common.constants import CV
 from openpilot.common.realtime import DT_MDL
 
 from openpilot.starpilot.common.starpilot_variables import CITY_SPEED_LIMIT, CRUISING_SPEED
-from openpilot.starpilot.controls.lib.curve_speed_controller import CurveSpeedController, is_manual_speed_control
+from openpilot.starpilot.controls.lib.curve_speed_controller import (
+  CSC_ACTIVE_OFF_DELTA,
+  CSC_GLOW_HOLD_TIME,
+  CSC_GLOW_ON_DELTA,
+  CurveSpeedController,
+  is_manual_speed_control,
+)
 from openpilot.starpilot.controls.lib.speed_limit_controller import SpeedLimitController
 from openpilot.selfdrive.controls.lib.longitudinal_vehicle_tunes import (
   get_force_stop_distance_bias,
@@ -16,7 +22,6 @@ from openpilot.selfdrive.controls.lib.longitudinal_vehicle_tunes import (
 )
 
 CSC_MIN_SPEED = CITY_SPEED_LIMIT * CV.MPH_TO_MS
-CSC_CURVE_RELEASE_HOLD_TIME = 0.75
 OVERRIDE_FORCE_STOP_TIMER = 10
 STANDSTILL_FORCE_STOP_CLEAR_TIME = 0.75
 # Open-loop — green is undetectable at standstill, so this only needs to cover the
@@ -59,6 +64,8 @@ LEAD_VETO_M_OVERRIDES = {
 }
 FORCE_STOP_APPROACH_DECEL = 0.65  # m/s^2 — speed ceiling before commit. LOWER = more early
                           # braking; don't go under FORCE_STOP_MODEL_APPROACH_DECEL
+# approachStopLength is published RAW: model_length converges from above, so rate-limiting
+# it inward freezes it far out and the constraint never binds. Tried, measured, don't re-add.
 ADAS_MAX_MS = 17.88       # 40 mph — cross-street ADAS guard
 DASH_SEED_M = 27.0        # ~88 ft — typical ADAS detection distance, used to snap
                           # tracked length closer when dashboard confirms a sign
@@ -76,6 +83,14 @@ FORCE_STOP_TURN_VETO_STEERING_ANGLE = 25.0
 FORCE_STOP_CURVE_VETO_MAX_ROAD_CURVATURE = 0.003
 FORCE_STOP_TURN_VETO_STOP_SEEN_HOLD_TIME = 4.0
 FORCE_STOP_DISTANCE_REANCHOR_MIN_GAP = 3.0  # m — ignore small model-horizon noise
+FORCE_STOP_REANCHOR_MIN_M = 40.0  # m — inside this only ratchet down; shouldStop doesn't
+                          # assert until ~10 m, so horizon jitter would release the stop
+FORCE_STOP_CAP_SLACK_M = 15.0  # m — the line can't move away, so tracked can never exceed
+                          # what it was at commit minus distance driven. Slack covers an
+                          # under-read at commit; without it that would stop us short.
+FORCE_STOP_CAP_TAPER_M = 60.0  # m — slack fades to 0 as the cap closes. The solver aims at
+                          # tracked, so slack held near the line is braking for a stop bar
+                          # that far past the real one.
 
 # Knob bounds (mirror of UI slider; defense in depth)
 OFFSET_FT_MIN = -20
@@ -179,9 +194,11 @@ class StarPilotVCruise:
     self.force_stop_from_light = False
     self.force_stop_light_clear_since = None
     self.controls_enabled_previously = False
+    self.approach_stop_length = 0.0  # published as starpilotPlan.approachStopLength
     # Kinematic distance estimator. Same attribute also published as
     # starpilotPlan.forcingStopLength, so the existing reader keeps working.
     self.tracked_model_length = 0.0
+    self.force_stop_distance_cap = 0.0  # odometry ceiling, re-seeded until commit
 
     self.stop_sign_confirmed = False
     self.stop_seen_on_approach_at = None
@@ -190,8 +207,9 @@ class StarPilotVCruise:
     self._nav_instruction_state = {}
     self._applied_slc_control_target = 0.0
     self.csc_controlling_speed = False
+    self.csc_glow_release_timer = 0.0
+    self.csc_override = False
     self.csc_target = 0.0
-    self.csc_curve_last_seen_at = None
 
   def _update_nav_instruction_state(self):
     raw = self.starpilot_planner.params_memory.get("NavInstructionState") or {}
@@ -557,28 +575,62 @@ class StarPilotVCruise:
       starpilot_toggles.curve_speed_controller and
       (not getattr(starpilot_toggles, "csc_no_lead", False) or not following_lead)
     )
-    csc_curve_detected = csc_available and self.starpilot_planner.road_curvature_detected
-    if csc_curve_detected:
-      self.csc.update_target(v_ego)
+    # The blinker veto is for lane changes/turns, not for an already-real curve -- releasing it
+    # there let the car accelerate into the bend, then claw the speed back once the blinker cleared.
+    csc_blinker_on = ((sm["carState"].leftBlinker or sm["carState"].rightBlinker) and
+                      not self.starpilot_planner.driving_in_curve)
+    csc_was_controlling = self.csc_controlling_speed
+    # a pending SLC confirmation owns the accel button
+    slc_confirmation_pending = self.slc.speed_limit_changed_timer > DT_MDL and self.slc.unconfirmed_speed_limit >= 1
+    csc_accel_button = bool(sm["starpilotCarState"].accelPressed) and not slc_confirmation_pending
 
-      self.csc_controlling_speed = True
-      self.csc_target = self.csc.target
-      self.csc_curve_last_seen_at = now
-    else:
-      csc_release_hold = bool(
-        csc_available and
-        self.csc_controlling_speed and
-        self.csc_curve_last_seen_at is not None and
-        self._elapsed_seconds(now, self.csc_curve_last_seen_at) < CSC_CURVE_RELEASE_HOLD_TIME
-      )
-      if not csc_release_hold:
-        self.csc.log_data(v_ego, sm)
+    # Latched outside the availability branch: the press itself suspends CSC this frame, so
+    # latching inside it would never see the press, and the slowdown would return on release.
+    if csc_was_controlling and csc_accel_button:
+      self.csc_override = True
+    if not (long_control_active and starpilot_toggles.curve_speed_controller):
+      self.csc_override = False
 
+    if csc_available and not csc_blinker_on:
+      self.csc.update_target(v_ego, v_cruise)
+
+      if self.csc_override and self.csc.target > v_cruise - CSC_ACTIVE_OFF_DELTA:
+        self.csc_override = False
+
+      if self.csc_override:
         self.csc_controlling_speed = False
-        self.csc.target_set = False
-        self.csc_curve_last_seen_at = None
-
+        self.csc_glow_release_timer = 0.0
         self.csc_target = v_cruise
+      else:
+        self.csc_target = self.csc.target
+        # A low target alone means nothing until the car has actually reached it (slowed down
+        # to it, or accelerated up into it). Release still waits for the set speed, so the glow
+        # spans the hold and the recovery, not just the braking.
+        if self.csc_target < v_cruise - CSC_GLOW_ON_DELTA and v_ego >= self.csc_target - CSC_ACTIVE_OFF_DELTA:
+          self.csc_controlling_speed = True
+          self.csc_glow_release_timer = 0.0
+        elif self.csc_target > v_cruise - CSC_ACTIVE_OFF_DELTA:
+          # hold through a brief release: one curve routinely lets go and re-engages
+          self.csc_glow_release_timer += DT_MDL
+          if self.csc_glow_release_timer >= CSC_GLOW_HOLD_TIME:
+            self.csc_controlling_speed = False
+        else:
+          self.csc_glow_release_timer = 0.0
+    elif csc_available:
+      # Release the cap so CSC can't fight the lane change, but keep planning -- resetting here
+      # threw the braking plan away and re-planned from the set speed with the curve closer.
+      self.csc.update_target(v_ego, v_cruise)
+      self.csc_controlling_speed = False
+      self.csc_glow_release_timer = 0.0
+      self.csc_target = v_cruise
+    else:
+      self.csc.reset(v_cruise)
+      self.csc_controlling_speed = False
+      self.csc_glow_release_timer = 0.0
+      self.csc_target = v_cruise
+
+    self.csc.handle_override(v_ego, csc_was_controlling, sm, accel_button=csc_accel_button)
+    self.csc.log_data(v_ego, sm)
 
     # Pfeiferj's Speed Limit Controller
     self.slc.starpilot_toggles = starpilot_toggles
@@ -605,6 +657,9 @@ class StarPilotVCruise:
     offset_ft = max(OFFSET_FT_MIN, min(OFFSET_FT_MAX, offset_ft_raw))
     offset_m = offset_ft * FT_TO_M
 
+    # cleared on every path; only the far-approach envelope below republishes it
+    self.approach_stop_length = 0.0
+
     if force_standstill_enabled and not self.override_force_standstill:
       self.forcing_stop = True
       self.tracked_model_length = 0.0
@@ -632,7 +687,7 @@ class StarPilotVCruise:
           model_wants_stop = False
         if (
           not dash_active and
-          self.tracked_model_length > force_stop_handoff_m and
+          self.tracked_model_length > max(force_stop_handoff_m, FORCE_STOP_REANCHOR_MIN_M) and
           not model_wants_stop and
           model_length > self.tracked_model_length + FORCE_STOP_DISTANCE_REANCHOR_MIN_GAP and
           (
@@ -644,6 +699,12 @@ class StarPilotVCruise:
           self.tracked_model_length = model_length
         else:
           self.tracked_model_length = min(self.tracked_model_length, model_length)
+        # Odometry ceiling: the line can't recede, so a re-anchor may never exceed what we
+        # had at commit minus what we've driven. Bounds a ballooning horizon (seen +95 m)
+        # that the REANCHOR_MIN floor can't catch, since that floor trusts the estimate.
+        self.force_stop_distance_cap = max(self.force_stop_distance_cap - (v_ego * DT_MDL), 0.0)
+        cap_slack = FORCE_STOP_CAP_SLACK_M * min(self.force_stop_distance_cap / FORCE_STOP_CAP_TAPER_M, 1.0)
+        self.tracked_model_length = min(self.tracked_model_length, self.force_stop_distance_cap + cap_slack)
         if dash_active:
           if model_length < DASH_MODEL_AGREE_M:
             self.tracked_model_length = min(self.tracked_model_length, DASH_SEED_M)
@@ -677,6 +738,7 @@ class StarPilotVCruise:
       self.stop_sign_confirmed = False
 
       self.tracked_model_length = self.starpilot_planner.model_length
+      self.force_stop_distance_cap = self.tracked_model_length
 
       targets = [v_cruise]
       if self.csc_target >= CSC_MIN_SPEED:
@@ -722,6 +784,8 @@ class StarPilotVCruise:
         adjacent_stop_d = self._get_adjacent_stop_distance(sm)
         if adjacent_stop_d is not None:
           approach_d = min(approach_d, adjacent_stop_d)
+        # pre-offset, so it hands off to forcingStopLength at commit without a step
+        self.approach_stop_length = max(approach_d, 0.0)
         approach_d += offset_m + force_stop_distance_bias_m
         if approach_d > force_stop_handoff_m:
           targets.append(math.sqrt(2.0 * FORCE_STOP_APPROACH_DECEL * (approach_d - force_stop_handoff_m)))

starpilot/controls/starpilot_planner.py

@@ -20,7 +20,7 @@ from openpilot.selfdrive.controls.lib.lead_behavior import (
 from openpilot.selfdrive.controls.lib.longitudinal_mpc_lib.long_mpc import A_CHANGE_COST, DANGER_ZONE_COST, J_EGO_COST, STOP_DISTANCE
 from openpilot.selfdrive.controls.lib.longitudinal_vehicle_tunes import get_lead_follow_jerk_scale
 
-from openpilot.starpilot.common.starpilot_utilities import calculate_lane_width, calculate_road_curvature
+from openpilot.starpilot.common.starpilot_utilities import calculate_lane_width, calculate_road_curvature, extract_curve_profile
 from openpilot.starpilot.common.starpilot_variables import CRUISING_SPEED, MINIMUM_LATERAL_ACCELERATION, PLANNER_TIME, THRESHOLD
 from openpilot.starpilot.controls.lib.conditional_chill_mode import ConditionalChillMode
 from openpilot.starpilot.controls.lib.conditional_experimental_mode import ConditionalExperimentalMode
@@ -31,7 +31,10 @@ from openpilot.starpilot.controls.lib.starpilot_vcruise import StarPilotVCruise
 from openpilot.starpilot.controls.lib.weather_checker import WeatherChecker
 
 RADARLESS_TRACK_HOLD_TIME = 0.45
-FORCE_STOP_JERK_SCALE = 0.32  # accel-change cost multiplier while forcing_stop (125 -> ~40)
+FORCE_STOP_JERK_SCALE = 0.20  # accel-change cost multiplier for the whole stop approach,
+                              # envelope included (125 -> 25). Lower = reaches the braking
+                              # target sooner; it does not make the target deeper. Response
+                              # is super-linear here, so raise it if onset feels like a step.
 FORCE_STOP_JERK_SCALE_OVERRIDES = {
   # The Elantra's current force-stop ramp is smooth, but it waits too long
   # before building decel and then arrives at the initial brake too abruptly.
@@ -211,6 +214,7 @@ class StarPilotPlanner:
     self.model_stopped = self.raw_model_stopped or self.starpilot_vcruise.forcing_stop
 
     self.road_curvature, self.time_to_curve = calculate_road_curvature(sm["modelV2"], v_ego)
+    self.curve_profile = extract_curve_profile(sm["modelV2"])
 
     self.road_curvature_detected = (1 / abs(self.road_curvature))**0.5 < v_ego > CRUISING_SPEED and not (sm["carState"].leftBlinker or sm["carState"].rightBlinker)
 
@@ -308,7 +312,9 @@ class StarPilotPlanner:
     except (KeyError, IndexError, TypeError, AttributeError):
       car_params = None
 
-    if self.starpilot_vcruise.forcing_stop:
+    # Also while the far-approach envelope is running: at onset the ramp reaches only
+    # ~-0.5 m/s^2 after a second, so the first seconds of a detected red are mostly lost.
+    if self.starpilot_vcruise.forcing_stop or self.starpilot_vcruise.approach_stop_length > 0.0:
       jerk_scale = get_force_stop_jerk_scale(car_params)
     elif self.tracking_lead:
       # Elantra vision leads can hand off from cruise to lead0 while closing
@@ -327,6 +333,9 @@ class StarPilotPlanner:
     starpilotPlan.cscControllingSpeed = self.starpilot_vcruise.csc_controlling_speed
     starpilotPlan.cscSpeed = float(self.starpilot_vcruise.csc_target)
     starpilotPlan.cscTraining = self.starpilot_vcruise.csc.enable_training
+    starpilotPlan.cscOverridden = self.starpilot_vcruise.csc_override
+    starpilotPlan.cscLearnedLatAccel = float(self.starpilot_vcruise.csc.learned_lat_accel(self.road_curvature))
+    starpilotPlan.cscBindingDistance = float(self.starpilot_vcruise.csc.binding_distance)
 
     starpilotPlan.desiredFollowDistance = int(self.starpilot_following.desired_follow_distance)
     starpilotPlan.disableThrottle = (
@@ -346,6 +355,7 @@ class StarPilotPlanner:
 
     starpilotPlan.forcingStop = self.starpilot_vcruise.forcing_stop
     starpilotPlan.forcingStopLength = self.starpilot_vcruise.tracked_model_length
+    starpilotPlan.approachStopLength = float(self.starpilot_vcruise.approach_stop_length)
     starpilotPlan.stopSignConfirmed = self.starpilot_vcruise.stop_sign_confirmed
 
     starpilotPlan.starpilotEvents = self.starpilot_events.events.to_msg()

starpilot/system/the_galaxy/assets/components/tools/device_settings.css

@@ -384,6 +384,14 @@
   padding: 0.2rem 0.6rem;
 }
 
+/* read-only: no border, since there is nothing here to click or edit */
+.ds-row-readout {
+  background-color: transparent;
+  border: none;
+  color: var(--text-muted);
+  font-style: italic;
+}
+
 .ds-stepper-container {
   width: 100%;
 }

starpilot/system/the_galaxy/assets/components/tools/device_settings.js

@@ -478,6 +478,16 @@ function formatSliderValue(val, stepStr, precisionInt, key) {
   return Number(v.toFixed(dec)).toString()
 }
 
+function formatReadoutValue(p) {
+  const raw = state.values[p.key]
+  const v = parseFloat(raw)
+  if (raw === undefined || raw === null || Number.isNaN(v)) return "--"
+
+  const precision = p.precision !== undefined && p.precision !== null ? Number(p.precision) : 2
+  const formatted = Number(v.toFixed(Math.max(0, precision))).toString()
+  return p.unit ? `${formatted}${p.unit}` : formatted
+}
+
 function formatNumericForInput(value, precision) {
   const n = Number(value)
   if (!Number.isFinite(n)) return ""
@@ -1492,6 +1502,7 @@ function renderSettingRow(p) {
   const isText = p.ui_type === "text"
   const isColor = p.ui_type === "color"
   const isAction = p.ui_type === "action"
+  const isReadout = p.ui_type === "readout"
   const isGroup = isGroupParam(p)
   const isChild = p.parent_key ? "ds-child-modifier" : ""
   const lockReason = () => getSettingLockReason(p)
@@ -1642,7 +1653,7 @@ function renderSettingRow(p) {
           @click="${() => resetColorParam(p)}">Stock</button>
       </div>
     `
-  } else if (!isGroup) {
+  } else if (!isGroup && !isReadout) {
     if (p.key === "IsRHD") {
       rowControl = html`
         <div style="display:flex; align-items:center; gap:0.75rem;">
@@ -1712,8 +1723,9 @@ function renderSettingRow(p) {
             </div>
           ` : ""}
         </div>
-        ${(isNumeric || isColor) ? html`<span class="ds-row-value" id="ds-display-${p.key}">${() => {
+        ${(isNumeric || isColor || isReadout) ? html`<span class="ds-row-value ${isReadout ? "ds-row-readout" : ""}" id="ds-display-${p.key}">${() => {
             if (isColor) return formatColorDisplayValue(p)
+            if (isReadout) return formatReadoutValue(p)
             const currentValue = state.sliderPreviewValues[p.key] ?? state.values[p.key]
             const bounds = numericBounds(p)
             return currentValue !== undefined ? formatSliderValue(currentValue, String(bounds.step), p.precision, p.key) : ".."

starpilot/system/the_galaxy/the_galaxy.py

@@ -5483,6 +5483,16 @@ def setup(app):
     result["VehicleParked"] = _get_vehicle_parked()
     result["AlphaLongitudinalAvailable"] = _get_alpha_longitudinal_available()
     result["HasRivianAngleHarness"] = _get_has_rivian_angle_harness()
+    # read-only: excluded from allowed_keys (and so from the write paths) but still
+    # worth surfacing as a display-only readout
+    try:
+      result["CalibratedLateralAcceleration"] = _get_current_param_value("CalibratedLateralAcceleration", float, defaults_lookup)
+    except Exception:
+      result["CalibratedLateralAcceleration"] = None
+    try:
+      result["CalibrationProgress"] = _get_current_param_value("CalibrationProgress", float, defaults_lookup)
+    except Exception:
+      result["CalibrationProgress"] = None
 
     return jsonify(_sanitize_json_value(result)), 200
 

tools/longitudinal/analyze_csc.py

@@ -0,0 +1,295 @@
+#!/usr/bin/env python3
+"""Curve Speed Controller field report: does it cut the lateral-accel tail, how
+often does it engage, and how often do drivers reject it.
+
+Usage:
+  ./analyze_csc.py <route-or-segment>            # e.g. a1b2c3d4e5f6g7h8|2026-08-14--10-30-00
+  ./analyze_csc.py <rlog-path> [<rlog-path> ...]
+  ./analyze_csc.py <route> --json report.json
+"""
+from __future__ import annotations
+
+import argparse
+import json
+import math
+from dataclasses import dataclass, field
+from pathlib import Path
+
+import numpy as np
+
+DT = 0.05                      # modelV2/starpilotPlan cadence
+MS_TO_MPH = 2.23694
+M_TO_MILES = 1.0 / 1609.34
+HIGHWAY_SPEED = 60.0 / MS_TO_MPH   # above this, engagement is the over-slowing regression risk
+CURVE_LAT_ACCEL = 1.3              # MINIMUM_LATERAL_ACCELERATION
+EPISODE_GAP_S = 1.0
+V_CRUISE_UNSET = 255
+
+
+@dataclass
+class Frame:
+  t: float = 0.0
+  v_ego: float = 0.0
+  a_ego: float = 0.0
+  curvature: float = 0.0
+  gas: bool = False
+  brake: bool = False
+  accel_pressed: bool = False
+  long_active: bool = False
+  blinker: bool = False     # CSC gating input: a blinker suspends it entirely
+  csc_active: bool = False
+  csc_overridden: bool = False
+  csc_training: bool = False
+  csc_speed: float = 0.0
+  v_cruise: float = 0.0     # applied cruise speed, already reduced by CSC
+  set_speed: float = 0.0    # what the driver dialled in, so cuts are measurable
+  learned_lat_accel: float = 0.0
+  binding_distance: float = 0.0
+
+  @property
+  def lat_accel(self) -> float:
+    return self.v_ego ** 2 * abs(self.curvature)
+
+
+@dataclass
+class Episode:
+  start: float
+  end: float
+  peak_cut: float = 0.0
+  peak_lat_accel: float = 0.0
+  min_a_ego: float = 0.0
+  entry_speed: float = 0.0
+  binding_distance: float = 0.0
+  cancelled: bool = False
+  gas: bool = False
+  brake: bool = False
+
+  @property
+  def duration(self) -> float:
+    return self.end - self.start
+
+
+def read_events(identifier: str):
+  """A downloaded rlog reads directly; anything else goes through LogReader."""
+  path = Path(identifier)
+  if path.is_file():
+    from cereal import log as capnp_log
+
+    data = path.read_bytes()
+    if data[:4] == b"\x28\xb5\x2f\xfd":
+      import zstandard
+      data = zstandard.ZstdDecompressor().decompress(data, max_output_size=2 << 30)
+    return capnp_log.Event.read_multiple_bytes(data)
+
+  from openpilot.tools.lib.logreader import LogReader, ReadMode  # needs the device stack
+
+  return LogReader(identifier, default_mode=ReadMode.AUTO, sort_by_time=True)
+
+
+def read_frames(identifier: str) -> list[Frame]:
+  """Join carState/controlsState/starpilotPlan onto the plan's cadence."""
+  frames: list[Frame] = []
+  latest = Frame()
+  t0 = None
+  have_plan = False
+
+  for msg in read_events(identifier):
+    which = msg.which()
+    if which == "carState":
+      cs = msg.carState
+      latest.v_ego = float(cs.vEgo)
+      latest.a_ego = float(cs.aEgo)
+      latest.gas = bool(cs.gasPressed)
+      latest.brake = bool(cs.brakePressed)
+      latest.blinker = bool(cs.leftBlinker or cs.rightBlinker)
+      set_kph = float(cs.vCruise)
+      latest.set_speed = set_kph / 3.6 if 0 < set_kph < V_CRUISE_UNSET else 0.0
+    elif which == "carControl":
+      latest.long_active = bool(msg.carControl.longActive)
+    elif which == "controlsState":
+      latest.curvature = float(msg.controlsState.curvature)
+    elif which == "starpilotCarState":
+      latest.accel_pressed = bool(getattr(msg.starpilotCarState, "accelPressed", False))
+    elif which == "starpilotPlan":
+      plan = msg.starpilotPlan
+      have_plan = True
+      if t0 is None:
+        t0 = msg.logMonoTime / 1e9
+      latest.t = msg.logMonoTime / 1e9 - t0
+      latest.csc_active = bool(plan.cscControllingSpeed)
+      latest.csc_training = bool(plan.cscTraining)
+      latest.csc_speed = float(plan.cscSpeed)
+      latest.v_cruise = float(plan.vCruise)
+      # absent in older logs
+      latest.csc_overridden = bool(getattr(plan, "cscOverridden", False))
+      latest.learned_lat_accel = float(getattr(plan, "cscLearnedLatAccel", 0.0))
+      latest.binding_distance = float(getattr(plan, "cscBindingDistance", 0.0))
+      frames.append(Frame(**vars(latest)))
+
+  if not have_plan:
+    raise SystemExit(f"no starpilotPlan messages in {identifier} — is this a StarPilot route?")
+  return frames
+
+
+def build_episodes(frames: list[Frame]) -> list[Episode]:
+  episodes: list[Episode] = []
+  current: Episode | None = None
+  last_active_t = -math.inf
+
+  for f in frames:
+    if f.csc_active:
+      if current is None or (f.t - last_active_t) > EPISODE_GAP_S:
+        current = Episode(start=f.t, end=f.t, entry_speed=f.v_ego,
+                          binding_distance=f.binding_distance, min_a_ego=f.a_ego)
+        episodes.append(current)
+      current.end = f.t
+      if f.set_speed > 0:
+        current.peak_cut = max(current.peak_cut, f.set_speed - f.csc_speed)
+      current.peak_lat_accel = max(current.peak_lat_accel, f.lat_accel)
+      current.min_a_ego = min(current.min_a_ego, f.a_ego)
+      current.gas |= f.gas
+      current.brake |= f.brake
+      last_active_t = f.t
+    elif current is not None and (f.t - last_active_t) <= EPISODE_GAP_S:
+      # an override releases CSC on the same frame it registers, so the rejection
+      # always lands just past the end of the episode it rejected
+      current.cancelled |= f.csc_overridden or f.accel_pressed
+      current.gas |= f.gas
+      current.brake |= f.brake
+
+  return episodes
+
+
+def curve_lat_accel_peaks(frames: list[Frame]) -> list[float]:
+  """Peak lateral acceleration of each distinct curve, engaged driving only."""
+  peaks: list[float] = []
+  peak = 0.0
+  in_curve = False
+  for f in frames:
+    if not f.long_active:
+      continue
+    if f.lat_accel >= CURVE_LAT_ACCEL:
+      in_curve = True
+      peak = max(peak, f.lat_accel)
+    elif in_curve:
+      peaks.append(peak)
+      peak = 0.0
+      in_curve = False
+  if in_curve:
+    peaks.append(peak)
+  return peaks
+
+
+def summarize(frames: list[Frame], episodes: list[Episode]) -> dict:
+  driving = [f for f in frames if f.v_ego > 5.0]
+  engaged = [f for f in driving if f.long_active]
+  active = [f for f in engaged if f.csc_active]
+  distance_mi = sum(f.v_ego * DT for f in driving) * M_TO_MILES
+  peaks = curve_lat_accel_peaks(frames)
+  highway = [e for e in episodes if e.entry_speed >= HIGHWAY_SPEED]
+
+  def pct(n, d):
+    return 100.0 * n / d if d else 0.0
+
+  return {
+    "route": {
+      "duration_min": len(frames) * DT / 60.0,
+      "distance_mi": distance_mi,
+      "engaged_pct": pct(len(engaged), len(driving)),
+      "mean_speed_mph": float(np.mean([f.v_ego for f in driving]) * MS_TO_MPH) if driving else 0.0,
+    },
+    "engagement": {
+      "active_pct_of_engaged": pct(len(active), len(engaged)),
+      "episodes": len(episodes),
+      "episodes_per_mile": len(episodes) / distance_mi if distance_mi > 0.1 else 0.0,
+      "median_duration_s": float(np.median([e.duration for e in episodes])) if episodes else 0.0,
+      "max_duration_s": max((e.duration for e in episodes), default=0.0),
+      "median_cut_mph": float(np.median([e.peak_cut for e in episodes]) * MS_TO_MPH) if episodes else 0.0,
+      "max_cut_mph": max((e.peak_cut for e in episodes), default=0.0) * MS_TO_MPH,
+      "median_anticipation_m": float(np.median([e.binding_distance for e in episodes])) if episodes else 0.0,
+    },
+    "outcome_lat_accel": {
+      "curves_seen": len(peaks),
+      "median": float(np.median(peaks)) if peaks else 0.0,
+      "p90": float(np.percentile(peaks, 90)) if peaks else 0.0,
+      "p99": float(np.percentile(peaks, 99)) if peaks else 0.0,
+      "max": max(peaks, default=0.0),
+      "over_3_0_pct": pct(sum(1 for p in peaks if p > 3.0), len(peaks)),
+    },
+    "acceptance": {
+      "cancelled_episodes": sum(1 for e in episodes if e.cancelled),
+      "cancel_rate_pct": pct(sum(1 for e in episodes if e.cancelled), len(episodes)),
+      "gas_during_episode_pct": pct(sum(1 for e in episodes if e.gas), len(episodes)),
+      "brake_during_episode_pct": pct(sum(1 for e in episodes if e.brake), len(episodes)),
+    },
+    "comfort": {
+      "median_min_a_ego": float(np.median([e.min_a_ego for e in episodes])) if episodes else 0.0,
+      "hardest_decel": min((e.min_a_ego for e in episodes), default=0.0),
+    },
+    "highway_watch": {
+      "episodes_above_60mph": len(highway),
+      "max_cut_mph": max((e.peak_cut for e in highway), default=0.0) * MS_TO_MPH,
+    },
+    "learning": {
+      "training_pct_of_driving": pct(sum(1 for f in driving if f.csc_training), len(driving)),
+      "learned_lat_accel_min": min((f.learned_lat_accel for f in active), default=0.0),
+      "learned_lat_accel_max": max((f.learned_lat_accel for f in active), default=0.0),
+    },
+  }
+
+
+def print_report(name: str, s: dict) -> None:
+  r, e, o, a, c, h, l = (s["route"], s["engagement"], s["outcome_lat_accel"],
+                         s["acceptance"], s["comfort"], s["highway_watch"], s["learning"])
+  print(f"\n=== {name}")
+  print(f"  {r['duration_min']:.1f} min, {r['distance_mi']:.1f} mi, "
+        f"{r['mean_speed_mph']:.0f} mph avg, engaged {r['engaged_pct']:.0f}% of driving")
+
+  print("\n  DOES IT WORK -- peak lateral accel per curve (engaged)")
+  print(f"    {o['curves_seen']} curves   median {o['median']:.2f}   p90 {o['p90']:.2f}   "
+        f"p99 {o['p99']:.2f}   max {o['max']:.2f} m/s^2")
+  print(f"    curves over 3.0 m/s^2: {o['over_3_0_pct']:.1f}%   <-- this tail should shrink vs a CSC-off route")
+
+  print("\n  DO USERS ACCEPT IT")
+  print(f"    cancel rate (RES+) {a['cancel_rate_pct']:.0f}%   gas {a['gas_during_episode_pct']:.0f}%   "
+        f"brake {a['brake_during_episode_pct']:.0f}%   of {e['episodes']} episodes")
+  print("    cancels/gas high => too slow;  brake high => too fast")
+
+  print("\n  ENGAGEMENT")
+  print(f"    {e['active_pct_of_engaged']:.1f}% of engaged time, {e['episodes_per_mile']:.2f} episodes/mi, "
+        f"median {e['median_duration_s']:.1f}s (max {e['max_duration_s']:.1f}s)")
+  print(f"    speed cut median {e['median_cut_mph']:.1f} mph, max {e['max_cut_mph']:.1f} mph")
+  print(f"    braking begins {e['median_anticipation_m']:.0f} m ahead (median)")
+
+  print("\n  COMFORT / REGRESSION WATCH")
+  print(f"    decel median {c['median_min_a_ego']:.2f}, hardest {c['hardest_decel']:.2f} m/s^2")
+  print(f"    highway (>60 mph) episodes: {h['episodes_above_60mph']}, max cut {h['max_cut_mph']:.1f} mph"
+        f"   <-- over-slowing complaints start here")
+
+  print("\n  LEARNING")
+  print(f"    training {l['training_pct_of_driving']:.1f}% of driving; "
+        f"learned comfort in use {l['learned_lat_accel_min']:.2f}-{l['learned_lat_accel_max']:.2f} m/s^2")
+
+
+def main() -> None:
+  parser = argparse.ArgumentParser(description="Curve Speed Controller field report.")
+  parser.add_argument("routes", nargs="+", help="route/segment identifier(s) or rlog path(s)")
+  parser.add_argument("--json", type=Path, help="also write the raw numbers here")
+  args = parser.parse_args()
+
+  reports = {}
+  for identifier in args.routes:
+    name = Path(identifier).name if Path(identifier).exists() else identifier
+    frames = read_frames(identifier)
+    episodes = build_episodes(frames)
+    summary = summarize(frames, episodes)
+    reports[name] = summary
+    print_report(name, summary)
+
+  if args.json:
+    args.json.write_text(json.dumps(reports, indent=2))
+    print(f"\nwrote {args.json}")
+
+
+if __name__ == "__main__":
+  main()