Unstable Lane Change depending on distance with ScenarioRunner

[Fornerio]

CARLA version: 9.13 Platform/OS: Ubuntu Python 3.8.13

Hello everyone, I am collecting different scenarios for a synthetic scenarios classification system. I am currently working on lane change on a curved road, and I am using LaneChange class from atomic_behaviors.py.

The vehicle performs a correct maneuver only using _distance_lanechange values from a certain range, otherwise the vehicle skids since it steers too much.

Is there a way to smooth the trajectory (e.g. by adjusting the steering angle) in order to obtain a correct lane change for every possible value of _distance_lanechange parameter? Here is the creation of the behaviour for the scenario:

def _create_behavior(self):

        behaviour = py_trees.composites.Sequence("CarOn_{}_Lane" .format(self._direction))
        leading = self.other_actors[len(self.other_actors)-1]
        INIT, just_drive, just_drive2 = SpawnEgoLeadingVehicles( self.ego_vehicles[0], leading, self._velocity_ego, self._velocity, self.starting_point_ego, self._transform_visible.transform)

        just_drive, just_drive2 = TrueTrafficCutOutFront(self.other_actors, self._velocity_ego, just_drive, just_drive2, self._velocity, self.ego_vehicles[0], self.th)

        behaviour.add_child(INIT)
        behaviour.add_child(just_drive)
        behaviour.add_child(just_drive2)

        lc = py_trees.composites.Parallel(
            "LC", policy=py_trees.common.ParallelPolicy.SUCCESS_ON_ONE)

        car_driving_ego_2 = WaypointFollower(self.ego_vehicles[0], self._velocity_ego)

        lc.add_child(car_driving_ego_2)

        if(self._direction=='right'):
            lane_change = LaneChange(actor=leading, speed=self._velocity, direction = 'left', distance_same_lane=25, distance_other_lane=50,
                 distance_lane_change=120, lane_changes=1, name='LaneChange')
        else:
            print('Changing lane R')
        lane_change = LaneChange(actor=leading, speed=self._velocity, direction = 'right', distance_same_lane=25, distance_other_lane=50,
                 distance_lane_change=120, lane_changes=1, name='LaneChange')
        lc.add_child(lane_change)

        behaviour.add_child(lc)

        root = py_trees.composites.Sequence("Behavior", policy=py_trees.common.ParallelPolicy.SUCCESS_ON_ONE)
        root.add_child(behaviour)

        return root

Speed needs to be about 30/36 m/s since the classification system involves recognizing highway scenarios. Thanks to everyone in advance,

Luca Forneris University of Genoa

[stale[bot]]

This issue has been automatically marked as stale because it has not had recent activity. It will be closed if no further activity occurs. Thank you for your contributions.