wip

selfdrive/frogpilot/controls/frogpilot_planner.py

@@ -0,0 +1,325 @@
+import numpy as np
+
+import cereal.messaging as messaging
+
+from cereal import log
+
+from openpilot.common.conversions import Conversions as CV
+from openpilot.common.numpy_fast import interp
+from openpilot.common.params import Params
+
+from openpilot.selfdrive.car.interfaces import ACCEL_MIN, ACCEL_MAX
+from openpilot.selfdrive.controls.lib.desire_helper import LANE_CHANGE_SPEED_MIN
+from openpilot.selfdrive.controls.lib.drive_helpers import V_CRUISE_MAX
+from openpilot.selfdrive.controls.lib.longitudinal_mpc_lib.long_mpc import A_CHANGE_COST, J_EGO_COST, COMFORT_BRAKE, STOP_DISTANCE, get_jerk_factor, \
+                                                                           get_safe_obstacle_distance, get_stopped_equivalence_factor, get_T_FOLLOW
+from openpilot.selfdrive.controls.lib.longitudinal_planner import A_CRUISE_MIN, Lead, get_max_accel
+
+from openpilot.system.version import get_short_branch
+
+from openpilot.selfdrive.frogpilot.controls.lib.conditional_experimental_mode import ConditionalExperimentalMode
+from openpilot.selfdrive.frogpilot.controls.lib.frogpilot_functions import CITY_SPEED_LIMIT, CRUISING_SPEED, calculate_lane_width, calculate_road_curvature
+from openpilot.selfdrive.frogpilot.controls.lib.map_turn_speed_controller import MapTurnSpeedController
+from openpilot.selfdrive.frogpilot.controls.lib.model_manager import RADARLESS_MODELS
+from openpilot.selfdrive.frogpilot.controls.lib.speed_limit_controller import SpeedLimitController
+
+# Acceleration profiles - Credit goes to the DragonPilot team!
+                 # MPH = [0., 18,  36,  63,  94]
+A_CRUISE_MIN_BP_CUSTOM = [0., 8., 16., 28., 42.]
+                 # MPH = [0., 6.71, 13.4, 17.9, 24.6, 33.6, 44.7, 55.9, 67.1, 123]
+A_CRUISE_MAX_BP_CUSTOM = [0.,    3,   6.,   8.,  11.,  15.,  20.,  25.,  30., 55.]
+
+A_CRUISE_MIN_VALS_ECO = [-0.001, -0.010, -0.28, -0.56, -0.56]
+A_CRUISE_MAX_VALS_ECO = [3.5, 3.2, 2.3, 2.0, 1.15, .80, .58, .36, .30, .091]
+
+A_CRUISE_MIN_VALS_SPORT = [-0.50, -0.52, -0.55, -0.57, -0.60]
+A_CRUISE_MAX_VALS_SPORT = [3.5, 3.5, 3.3, 2.8, 1.5, 1.0, .75, .6, .38, .2]
+
+TRAFFIC_MODE_BP = [0., CITY_SPEED_LIMIT]
+
+TARGET_LAT_A = 1.9  # m/s^2
+
+def get_min_accel_eco(v_ego):
+  return interp(v_ego, A_CRUISE_MIN_BP_CUSTOM, A_CRUISE_MIN_VALS_ECO)
+
+def get_max_accel_eco(v_ego):
+  return interp(v_ego, A_CRUISE_MAX_BP_CUSTOM, A_CRUISE_MAX_VALS_ECO)
+
+def get_min_accel_sport(v_ego):
+  return interp(v_ego, A_CRUISE_MIN_BP_CUSTOM, A_CRUISE_MIN_VALS_SPORT)
+
+def get_max_accel_sport(v_ego):
+  return interp(v_ego, A_CRUISE_MAX_BP_CUSTOM, A_CRUISE_MAX_VALS_SPORT)
+
+class FrogPilotPlanner:
+  def __init__(self, CP):
+    self.CP = CP
+
+    self.params = Params()
+    self.params_memory = Params("/dev/shm/params")
+
+    self.cem = ConditionalExperimentalMode()
+    self.lead_one = Lead()
+    self.mtsc = MapTurnSpeedController()
+
+    self.release = get_short_branch() == "FrogPilot"
+
+    self.radarless_model = self.params.get("Model", encoding='utf-8') in RADARLESS_MODELS
+
+    self.override_slc = False
+
+    self.jerk = 0
+    self.overridden_speed = 0
+    self.mtsc_target = 0
+    self.slc_target = 0
+    self.t_follow = 0
+    self.vtsc_target = 0
+
+  def update(self, carState, controlsState, frogpilotCarControl, frogpilotNavigation, liveLocationKalman, modelData, radarState):
+    v_cruise_kph = min(controlsState.vCruise, V_CRUISE_MAX)
+    v_cruise = v_cruise_kph * CV.KPH_TO_MS
+    v_ego = max(carState.vEgo, 0)
+    v_lead = self.lead_one.vLead
+
+    if self.acceleration_profile == 1:
+      self.max_accel = get_max_accel_eco(v_ego)
+    elif self.acceleration_profile in (2, 3):
+      self.max_accel = get_max_accel_sport(v_ego)
+    elif not controlsState.experimentalMode:
+      self.max_accel = get_max_accel(v_ego)
+    else:
+      self.max_accel = ACCEL_MAX
+
+    v_cruise_changed = (self.mtsc_target or self.vtsc_target) < v_cruise
+
+    if self.deceleration_profile == 1 and not v_cruise_changed:
+      self.min_accel = get_min_accel_eco(v_ego)
+    elif self.deceleration_profile == 2 and not v_cruise_changed:
+      self.min_accel = get_min_accel_sport(v_ego)
+    elif not controlsState.experimentalMode:
+      self.min_accel = A_CRUISE_MIN
+    else:
+      self.min_accel = ACCEL_MIN
+
+    check_lane_width = self.adjacent_lanes or self.blind_spot_path or self.lane_detection
+    if check_lane_width and v_ego >= LANE_CHANGE_SPEED_MIN:
+      self.lane_width_left = float(calculate_lane_width(modelData.laneLines[0], modelData.laneLines[1], modelData.roadEdges[0]))
+      self.lane_width_right = float(calculate_lane_width(modelData.laneLines[3], modelData.laneLines[2], modelData.roadEdges[1]))
+    else:
+      self.lane_width_left = 0
+      self.lane_width_right = 0
+
+    road_curvature = calculate_road_curvature(modelData, v_ego)
+
+    if self.lead_one.status and self.CP.openpilotLongitudinalControl:
+      base_jerk = get_jerk_factor(self.custom_personalities, self.aggressive_jerk, self.standard_jerk, self.relaxed_jerk, controlsState.personality)
+      base_t_follow = get_T_FOLLOW(self.custom_personalities, self.aggressive_follow, self.standard_follow, self.relaxed_follow, controlsState.personality)
+      self.safe_obstacle_distance = int(np.mean(get_safe_obstacle_distance(v_ego, self.t_follow)))
+      self.safe_obstacle_distance_stock = int(np.mean(get_safe_obstacle_distance(v_ego, base_t_follow)))
+      self.stopped_equivalence_factor = int(np.mean(get_stopped_equivalence_factor(v_lead)))
+      self.jerk, self.t_follow = self.update_follow_values(base_jerk, self.lead_one, base_t_follow, frogpilotCarControl.trafficModeActive, v_ego, v_lead)
+    else:
+      self.safe_obstacle_distance = 0
+      self.safe_obstacle_distance_stock = 0
+      self.stopped_equivalence_factor = 0
+      self.t_follow = 1.45
+
+    self.v_cruise = self.update_v_cruise(carState, controlsState, controlsState.enabled, liveLocationKalman, modelData, road_curvature, v_cruise, v_ego)
+
+    if self.conditional_experimental_mode and self.CP.openpilotLongitudinalControl or self.green_light_alert:
+      self.cem.update(carState, controlsState.enabled, frogpilotNavigation, self.lead_one, modelData, road_curvature, self.t_follow, v_ego)
+
+    if self.radarless_model:
+      model_leads = list(modelData.leadsV3)
+      if len(model_leads) > 0:
+        model_lead = model_leads[0]
+        self.lead_one.update(model_lead.x[0], model_lead.y[0], model_lead.v[0], model_lead.a[0], model_lead.prob)
+      else:
+        self.lead_one.reset()
+    else:
+      self.lead_one = radarState.leadOne
+
+  def update_follow_values(self, jerk, lead_one, t_follow, trafficModeActive, v_ego, v_lead):
+    if trafficModeActive:
+      jerk = interp(v_ego, TRAFFIC_MODE_BP, self.traffic_mode_jerk)
+      t_follow = interp(v_ego, TRAFFIC_MODE_BP, self.traffic_mode_t_follow)
+
+    stopping_distance = STOP_DISTANCE + max(self.increased_stopping_distance - v_ego if not trafficModeActive else 0, 0)
+    lead_distance = self.lead_one.dRel + stopping_distance
+
+    # Offset by FrogAi for FrogPilot for a more natural takeoff with a lead
+    if self.aggressive_acceleration:
+      distance_factor = np.maximum(1, lead_distance - (v_ego * t_follow))
+      standstill_offset = max(stopping_distance - v_ego, 0)
+      acceleration_offset = np.clip((v_lead - v_ego) + standstill_offset - COMFORT_BRAKE, 1, distance_factor)
+      jerk /= acceleration_offset
+      t_follow /= acceleration_offset
+
+    # Offset by FrogAi for FrogPilot for a more natural approach to a slower lead
+    if self.smoother_braking:
+      distance_factor = np.maximum(1, lead_distance - (v_lead * t_follow))
+      far_lead_offset = max(lead_distance - CITY_SPEED_LIMIT, 0) if self.smoother_braking_far_lead else 0
+      braking_offset = np.clip((v_ego - v_lead) + far_lead_offset - COMFORT_BRAKE, 1, distance_factor)
+      if self.smoother_braking_jerk:
+        jerk *= max(braking_offset**(1 / COMFORT_BRAKE), 1)
+      t_follow /= braking_offset
+
+    return jerk, t_follow
+
+  def update_v_cruise(self, carState, controlsState, enabled, liveLocationKalman, modelData, road_curvature, v_cruise, v_ego):
+    gps_check = (liveLocationKalman.status == log.LiveLocationKalman.Status.valid) and liveLocationKalman.positionGeodetic.valid and liveLocationKalman.gpsOK
+
+    v_cruise_cluster = max(controlsState.vCruiseCluster, controlsState.vCruise) * CV.KPH_TO_MS
+    v_cruise_diff = v_cruise_cluster - v_cruise
+
+    v_ego_cluster = max(carState.vEgoCluster, v_ego)
+    v_ego_diff = v_ego_cluster - v_ego
+
+    # Pfeiferj's Map Turn Speed Controller
+    if self.map_turn_speed_controller and v_ego > CRUISING_SPEED and enabled and gps_check:
+      mtsc_active = self.mtsc_target < v_cruise
+      self.mtsc_target = np.clip(self.mtsc.target_speed(v_ego, carState.aEgo), CRUISING_SPEED, v_cruise)
+
+      if self.mtsc_curvature_check and road_curvature < 1.0 and not mtsc_active:
+        self.mtsc_target = v_cruise
+      if self.mtsc_target == CRUISING_SPEED:
+        self.mtsc_target = v_cruise
+    else:
+      self.mtsc_target = v_cruise
+
+    # Pfeiferj's Speed Limit Controller
+    if self.speed_limit_controller:
+      SpeedLimitController.update(v_ego)
+      unconfirmed_slc_target = SpeedLimitController.desired_speed_limit
+
+      # Check if the new speed limit has been confirmed by the user
+      if self.speed_limit_confirmation:
+        if self.params_memory.get_bool("SLCConfirmed") or self.slc_target == 0:
+          self.slc_target = unconfirmed_slc_target
+          self.params_memory.put_bool("SLCConfirmed", False)
+      else:
+        self.slc_target = unconfirmed_slc_target
+
+      # Override SLC upon gas pedal press and reset upon brake/cancel button
+      self.override_slc &= self.overridden_speed > self.slc_target
+      self.override_slc |= carState.gasPressed and v_ego > self.slc_target
+      self.override_slc &= enabled
+
+      # Use the override speed if SLC is being overridden
+      if self.override_slc:
+        if self.speed_limit_controller_override == 1:
+          # Set the speed limit to the manual set speed
+          if carState.gasPressed:
+            self.overridden_speed = v_ego + v_ego_diff
+          self.overridden_speed = np.clip(self.overridden_speed, self.slc_target, v_cruise + v_cruise_diff)
+        elif self.speed_limit_controller_override == 2:
+          # Set the speed limit to the max set speed
+          self.overridden_speed = v_cruise + v_cruise_diff
+      else:
+        self.overridden_speed = 0
+    else:
+      self.slc_target = v_cruise
+
+    # Pfeiferj's Vision Turn Controller
+    if self.vision_turn_controller and v_ego > CRUISING_SPEED and enabled:
+      orientation_rate = np.array(np.abs(modelData.orientationRate.z)) * self.curve_sensitivity
+      velocity = np.array(modelData.velocity.x)
+
+      max_pred_lat_acc = np.amax(orientation_rate * velocity)
+      max_curve = max_pred_lat_acc / (v_ego**2)
+      adjusted_target_lat_a = TARGET_LAT_A * self.turn_aggressiveness
+
+      self.vtsc_target = (adjusted_target_lat_a / max_curve)**0.5
+      self.vtsc_target = np.clip(self.vtsc_target, CRUISING_SPEED, v_cruise)
+    else:
+      self.vtsc_target = v_cruise
+
+    targets = [self.mtsc_target, max(self.overridden_speed, self.slc_target) - v_ego_diff, self.vtsc_target]
+    filtered_targets = [target if target > CRUISING_SPEED else v_cruise for target in targets]
+
+    return min(filtered_targets)
+
+  def publish(self, sm, pm):
+    frogpilot_plan_send = messaging.new_message('frogpilotPlan')
+    frogpilot_plan_send.valid = sm.all_checks(service_list=['carState', 'controlsState'])
+    frogpilotPlan = frogpilot_plan_send.frogpilotPlan
+
+    frogpilotPlan.accelerationJerk = A_CHANGE_COST * (float(self.jerk) if self.lead_one.status else 1)
+    frogpilotPlan.accelerationJerkStock = A_CHANGE_COST
+    frogpilotPlan.adjustedCruise = float(min(self.mtsc_target, self.vtsc_target) * (CV.MS_TO_KPH if self.is_metric else CV.MS_TO_MPH))
+    frogpilotPlan.conditionalExperimental = self.cem.experimental_mode
+    frogpilotPlan.desiredFollowDistance = self.safe_obstacle_distance - self.stopped_equivalence_factor
+    frogpilotPlan.egoJerk = J_EGO_COST * (float(self.jerk) if self.lead_one.status else 1)
+    frogpilotPlan.egoJerkStock = J_EGO_COST
+    frogpilotPlan.jerk = float(self.jerk)
+    frogpilotPlan.safeObstacleDistance = self.safe_obstacle_distance
+    frogpilotPlan.safeObstacleDistanceStock = self.safe_obstacle_distance_stock
+    frogpilotPlan.stoppedEquivalenceFactor = self.stopped_equivalence_factor
+    frogpilotPlan.laneWidthLeft = self.lane_width_left
+    frogpilotPlan.laneWidthRight = self.lane_width_right
+    frogpilotPlan.minAcceleration = self.min_accel
+    frogpilotPlan.maxAcceleration = self.max_accel
+    frogpilotPlan.tFollow = float(self.t_follow)
+    frogpilotPlan.vCruise = float(self.v_cruise)
+
+    frogpilotPlan.redLight = self.cem.red_light_detected
+
+    frogpilotPlan.slcOverridden = bool(self.override_slc)
+    frogpilotPlan.slcOverriddenSpeed = float(self.overridden_speed)
+    frogpilotPlan.slcSpeedLimit = self.slc_target
+    frogpilotPlan.slcSpeedLimitOffset = SpeedLimitController.offset
+    frogpilotPlan.unconfirmedSlcSpeedLimit = SpeedLimitController.desired_speed_limit
+
+    frogpilotPlan.vtscControllingCurve = bool(self.mtsc_target > self.vtsc_target)
+
+    pm.send('frogpilotPlan', frogpilot_plan_send)
+
+  def update_frogpilot_params(self):
+    self.is_metric = self.params.get_bool("IsMetric")
+
+    self.conditional_experimental_mode = self.CP.openpilotLongitudinalControl and self.params.get_bool("ConditionalExperimental")
+    if self.conditional_experimental_mode:
+      self.cem.update_frogpilot_params()
+
+    custom_alerts = self.params.get_bool("CustomAlerts")
+    self.green_light_alert = custom_alerts and self.params.get_bool("GreenLightAlert")
+
+    self.custom_personalities = self.params.get_bool("CustomPersonalities")
+    self.aggressive_jerk = self.params.get_float("AggressiveJerk")
+    self.aggressive_follow = self.params.get_float("AggressiveFollow")
+    self.standard_jerk = self.params.get_float("StandardJerk")
+    self.standard_follow = self.params.get_float("StandardFollow")
+    self.relaxed_jerk = self.params.get_float("RelaxedJerk")
+    self.relaxed_follow = self.params.get_float("RelaxedFollow")
+    self.traffic_jerk = self.params.get_float("TrafficJerk")
+    self.traffic_follow = self.params.get_float("TrafficFollow")
+    self.traffic_mode_jerk = [self.traffic_jerk, self.aggressive_jerk] if self.custom_personalities and not self.release else [1.0, 0.5]
+    self.traffic_mode_t_follow = [self.traffic_follow, self.aggressive_follow] if self.custom_personalities and not self.release else [0.5, 1.0]
+
+    custom_ui = self.params.get_bool("CustomUI")
+    self.adjacent_lanes = custom_ui and self.params.get_bool("AdjacentPath")
+    self.blind_spot_path = custom_ui and self.params.get_bool("BlindSpotPath")
+
+    nudgeless_lane_change = self.params.get_bool("NudgelessLaneChange")
+    self.lane_detection = nudgeless_lane_change and self.params.get_int("LaneDetectionWidth") != 0
+
+    longitudinal_tune = self.CP.openpilotLongitudinalControl and self.params.get_bool("LongitudinalTune")
+    self.acceleration_profile = self.params.get_int("AccelerationProfile") if longitudinal_tune else 0
+    self.deceleration_profile = self.params.get_int("DecelerationProfile") if longitudinal_tune else 0
+    self.aggressive_acceleration = longitudinal_tune and self.params.get_bool("AggressiveAcceleration")
+    self.increased_stopping_distance = self.params.get_int("StoppingDistance") * (1 if self.is_metric else CV.FOOT_TO_METER) if longitudinal_tune else 0
+    self.smoother_braking = longitudinal_tune and self.params.get_bool("SmoothBraking")
+    self.smoother_braking_far_lead = self.smoother_braking and self.params.get_bool("SmoothBrakingFarLead") and not self.release
+    self.smoother_braking_jerk = self.smoother_braking and self.params.get_bool("SmoothBrakingJerk") and not self.release
+
+    self.map_turn_speed_controller = self.CP.openpilotLongitudinalControl and self.params.get_bool("MTSCEnabled")
+    self.mtsc_curvature_check = self.map_turn_speed_controller and self.params.get_bool("MTSCCurvatureCheck")
+    self.params_memory.put_float("MapTargetLatA", 2 * (self.params.get_int("MTSCAggressiveness") / 100))
+
+    self.speed_limit_controller = self.CP.openpilotLongitudinalControl and self.params.get_bool("SpeedLimitController")
+    self.speed_limit_confirmation = self.speed_limit_controller and self.params.get_bool("SLCConfirmation")
+    self.speed_limit_controller_override = self.speed_limit_controller and self.params.get_int("SLCOverride")
+
+    self.vision_turn_controller = self.CP.openpilotLongitudinalControl and self.params.get_bool("VisionTurnControl")
+    self.curve_sensitivity = self.params.get_int("CurveSensitivity") / 100 if self.vision_turn_controller else 1
+    self.turn_aggressiveness = self.params.get_int("TurnAggressiveness") / 100 if self.vision_turn_controller else 1