def compute_all_paths_for_actor(
vehicle: carla.Actor,
dist_between_waypoints: float,
path_length: int = 10,
lane_type: carla.LaneType = carla.LaneType.Driving,
verbose: bool = False):
world = vehicle.get_world()
location = vehicle.get_location()
paths = compute_all_paths_from_location(location, world,
dist_between_waypoints,
path_length, lane_type, verbose)
return paths
def __init__(self, ego_veh: carla.Actor, gt_config: dict, debug=False):
"""
Constructor method.
Input:
ego_veh: Carla.Actor obj of the ego vehicle. Carla.Actor provides a get_world() method to get the
Carla.World it belongs to; thus, ego vehicle actor is sufficient to retrieve other info.
gt_config (dict): Configurations.
debug (bool): True to turn on debug features.
"""
self.debug = debug
# Retrieve carla.World object from ego vehicle carla.Actor object
carla_world = ego_veh.get_world()
# Dict as an buffer to store all ground truth data of interest
# Using dict helps automate data selection during recording since data can be queried by keys
# This buffer is automatically updated when the child ground truth extractors update their buffer
self.all_gt = {'static': {}, 'seq': {}}
# Traffic signs
self.all_gt['static']['traffic_sign'] = self.get_traffic_signs(
carla_world, self.debug)
# Front bumper's transform in Carla's coordinate system
# It's for the convenience of querying waypoints for lane using carla's APIs
# TODO: Try put this frame's origin at the intersection of camera FOV and ground surface?
self._fbumper_carla_tform = None
# Pose ground truth extractor
self.pose_gt = PoseGTExtractor(ego_veh, gt_config['pose'])
# Lane ground truth extractor
self.lane_gt = LaneGTExtractor(carla_world, gt_config['lane'],
self.debug)
# Set up buffer for sequential data
# Refer to pose ground truth extractor's gt buffer
self.all_gt['seq']['pose'] = self.pose_gt.gt
# Refer to lane ground truth extractor's gt buffer
self.all_gt['seq']['lane'] = self.lane_gt.gt
