James
112 lines
4.0 KiB
Python
112 lines
4.0 KiB
Python
#!/usr/bin/env python3
|
|
import numpy as np
|
|
|
|
from openpilot.common.realtime import DT_MDL
|
|
|
|
from openpilot.frogpilot.common.frogpilot_variables import CRUISING_SPEED, DEFAULT_LATERAL_ACCELERATION, PLANNER_TIME
|
|
|
|
CALIBRATION_PROGRESS_THRESHOLD = 10 / DT_MDL
|
|
MAX_CURVATURE = 0.1
|
|
MIN_CURVATURE = 0.01
|
|
PERCENTILE = 90
|
|
ROUNDING_PRECISION = 3
|
|
STEP = 0.001
|
|
|
|
class CurveSpeedController:
|
|
def __init__(self, FrogPilotVCruise):
|
|
self.frogpilot_planner = FrogPilotVCruise.frogpilot_planner
|
|
|
|
self.enable_training = False
|
|
self.target_set = False
|
|
|
|
self.training_timer = 0
|
|
|
|
self.curvature_data = self.frogpilot_planner.params.get("CurvatureData")
|
|
|
|
self.calculate_weights()
|
|
self.update_lateral_acceleration()
|
|
|
|
def calculate_weights(self):
|
|
curvatures = np.arange(MIN_CURVATURE, MAX_CURVATURE + STEP, STEP)
|
|
mid_point = (MIN_CURVATURE + MAX_CURVATURE) / 2
|
|
|
|
self.curvature_weights = {}
|
|
for curvature in curvatures:
|
|
distance = abs(curvature - mid_point) / (MAX_CURVATURE - MIN_CURVATURE)
|
|
weight = 1.0 + (4.0 * (1 - distance))
|
|
self.curvature_weights[str(round(curvature, ROUNDING_PRECISION))] = weight
|
|
|
|
def log_data(self, long_control_active, v_ego, sm):
|
|
self.enable_training = v_ego > CRUISING_SPEED
|
|
self.enable_training &= not self.frogpilot_planner.tracking_lead
|
|
self.enable_training &= not long_control_active
|
|
|
|
if self.enable_training:
|
|
self.training_timer += DT_MDL
|
|
|
|
if self.training_timer >= PLANNER_TIME and self.frogpilot_planner.driving_in_curve and not (sm["carState"].leftBlinker or sm["carState"].rightBlinker):
|
|
lateral_acceleration = abs(self.frogpilot_planner.lateral_acceleration)
|
|
road_curvature = abs(round(self.frogpilot_planner.road_curvature, ROUNDING_PRECISION))
|
|
|
|
key = str(road_curvature)
|
|
if key in self.curvature_data:
|
|
data = self.curvature_data[key]
|
|
|
|
average = data["average"]
|
|
count = data["count"]
|
|
|
|
self.curvature_data[key] = {
|
|
"average": ((average * count) + lateral_acceleration) / (count + 1),
|
|
"count": count + 1
|
|
}
|
|
else:
|
|
self.curvature_data[key] = {
|
|
"average": lateral_acceleration,
|
|
"count": 1
|
|
}
|
|
else:
|
|
self.enable_training = False
|
|
|
|
elif self.training_timer >= PLANNER_TIME:
|
|
progress = 0.0
|
|
total_weight = 0.0
|
|
|
|
for key in list(self.curvature_weights.keys()):
|
|
if key in self.curvature_data:
|
|
progress += min(self.curvature_data[key]["count"] / CALIBRATION_PROGRESS_THRESHOLD, 1.0) * self.curvature_weights[key]
|
|
|
|
total_weight += self.curvature_weights[key]
|
|
|
|
self.frogpilot_planner.params.put_nonblocking("CalibrationProgress", float(min((progress / total_weight) * 100, 100.0)))
|
|
self.frogpilot_planner.params.put_nonblocking("CurvatureData", self.curvature_data)
|
|
self.update_lateral_acceleration()
|
|
|
|
self.training_timer = 0
|
|
|
|
else:
|
|
self.training_timer = 0
|
|
|
|
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))
|
|
else:
|
|
self.lateral_acceleration = DEFAULT_LATERAL_ACCELERATION
|
|
|
|
self.frogpilot_planner.params.put_nonblocking("CalibratedLateralAcceleration", self.lateral_acceleration)
|
|
|
|
def update_target(self, v_ego):
|
|
lateral_acceleration = self.lateral_acceleration
|
|
if self.frogpilot_planner.frogpilot_weather.weather_id != 0:
|
|
lateral_acceleration -= self.lateral_acceleration * self.frogpilot_planner.frogpilot_weather.reduce_lateral_acceleration
|
|
|
|
if self.target_set:
|
|
csc_speed = (lateral_acceleration / abs(self.frogpilot_planner.road_curvature))**0.5
|
|
decel_rate = (v_ego - csc_speed) / self.frogpilot_planner.time_to_curve
|
|
|
|
self.target -= decel_rate * DT_MDL
|
|
self.target = np.clip(self.target, CRUISING_SPEED, csc_speed)
|
|
else:
|
|
self.target_set = True
|
|
self.target = v_ego
|