Exemplo n.º 1
0
    def extract_junction_points(self, tlight_find_radius=25.):
        carla_topology = self.carla_map.get_topology()
        junctions = carlautil.get_junctions_from_topology_graph(carla_topology)
        tlights = util.filter_to_list(lambda a: 'traffic_light' in a.type_id,
                                      self.carla_world.get_actors())
        tlight_distances = np.zeros((
            len(tlights),
            len(junctions),
        ))
        f = lambda j: carlautil.location_to_ndarray(j.bounding_box.location)
        junction_locations = util.map_to_ndarray(f, junctions)

        g = lambda tl: carlautil.transform_to_location_ndarray(tl.
                                                               get_transform())
        tlight_locations = util.map_to_ndarray(g, tlights)

        for idx, junction in enumerate(junctions):
            tlight_distances[:, idx] = np.linalg.norm(tlight_locations -
                                                      junction_locations[idx],
                                                      axis=1)

        is_controlled_junction = (tlight_distances < tlight_find_radius).any(
            axis=0)
        is_uncontrolled_junction = np.logical_not(is_controlled_junction)
        controlled_junction_locations \
                = junction_locations[is_controlled_junction]
        uncontrolled_junction_locations \
                = junction_locations[is_uncontrolled_junction]
        return util.AttrDict(controlled=controlled_junction_locations,
                             uncontrolled=uncontrolled_junction_locations)
Exemplo n.º 2
0
    def __init__(self, carla_world, carla_map, debug=False):
        """

        Parameters
        ----------
        carla_world : carla.World
        carla_map : carla.Map
        debug : bool

        """
        super().__init__(carla_world, carla_map, debug=debug)
        """Generate slope topology of the map"""
        # get waypoints
        wps = self.carla_map.generate_waypoints(4.0)

        def h(wp):
            l = wp.transform.location
            pitch = wp.transform.rotation.pitch
            return np.array([l.x, l.y, l.z, pitch])

        loc_and_pitch_of_wps = util.map_to_ndarray(h, wps)
        self.wp_locations = loc_and_pitch_of_wps[:, :-1]
        self.wp_pitches = loc_and_pitch_of_wps[:, -1]
        self.wp_pitches = self.wp_pitches % 360.
        self.wp_is_sloped = np.logical_and(
            self.SLOPE_DEGREES < self.wp_pitches,
            self.wp_pitches < self.SLOPE_UDEGREES)
        """Generate intersection topology of the map"""
        # get nodes in graph
        topology = self.carla_map.get_topology()
        G = nx.Graph()
        G.add_edges_from(topology)
        tlights = util.filter_to_list(lambda a: 'traffic_light' in a.type_id,
                                      self.carla_world.get_actors())
        junctions = carlautil.get_junctions_from_topology_graph(G)

        tlight_distances = np.zeros((
            len(tlights),
            len(junctions),
        ))
        f = lambda j: carlautil.location_to_ndarray(j.bounding_box.location)
        junction_locations = util.map_to_ndarray(f, junctions)

        g = lambda tl: carlautil.transform_to_location_ndarray(tl.
                                                               get_transform())
        tlight_locations = util.map_to_ndarray(g, tlights)

        for idx, junction in enumerate(junctions):
            tlight_distances[:, idx] = np.linalg.norm(tlight_locations -
                                                      junction_locations[idx],
                                                      axis=1)

        is_controlled_junction = (tlight_distances <
                                  self.TLIGHT_FIND_RADIUS).any(axis=0)
        is_uncontrolled_junction = np.logical_not(is_controlled_junction)
        self.controlled_junction_locations \
                = junction_locations[is_controlled_junction]
        self.uncontrolled_junction_locations \
                = junction_locations[is_uncontrolled_junction]
Exemplo n.º 3
0
def get_normalized_sensor_data(lidar_measurement):
    """Obtain LIDAR point cloud with rotation oriented to
    world frame of reference, and centered at (0, 0, 0). 

    First convert to world orientation which keeping the origin
    using lidar_measurement.transform, mentioned in:
    https://github.com/carla-simulator/carla/issues/2817

    Rotate points 90 degrees CCW around origin (why?).

    Adjust points 2.5 meters in the z direction to reflect
    how the sensor is placed with resp. to the ego vehicle.

    Lastly reflect the y-axis, mentioned in:
    https://github.com/carla-simulator/carla/issues/2699
    
    Parameters
    ----------
    lidar_measurement : carla.LidarMeasurement or carla.SemanticLidarMeasurement

    Returns
    -------
    np.array
        Has shape (number of points, 3)
        corresponding to (x, y, z) coordinates.
    np.array
        Labels of points. Has shape (number of points,)
    
    TODO: still don't really understand what's going on with
    the transformation matrix produced by transformation.
    Check out open3d_lidar.py for fix
    """
    raw_data = lidar_measurement.raw_data
    if isinstance(lidar_measurement, carla.LidarMeasurement):
        """Array of 32-bits floats (XYZI of each point).
        """
        points = np.frombuffer(raw_data, dtype=np.dtype('f4'))
        points = np.reshape(points, (int(points.shape[0] / 4), 4))
        labels = np.zeros(points.shape[0], dtype=np.uint32)
    elif isinstance(lidar_measurement, carla.SemanticLidarMeasurement):
        """Array containing the point cloud with instance and semantic information.
        For each point, four 32-bits floats are stored.
        XYZ coordinates.
        cosine of the incident angle.
        Unsigned int containing the index of the object hit.
        Unsigned int containing the semantic tag of the object it.
        """
        data = np.frombuffer(raw_data,
                             dtype=np.dtype([('x', np.float32),
                                             ('y', np.float32),
                                             ('z', np.float32),
                                             ('CosAngle', np.float32),
                                             ('ObjIdx', np.uint32),
                                             ('ObjTag', np.uint32)]))
        points = np.array([data['x'], data['y'], data['z'],
                           data['CosAngle']]).T
        labels = data['ObjTag']
    else:
        raise ValueError("Unknown lidar measurement.")

    transform_to_world = carla.Transform(carla.Location(),
                                         lidar_measurement.transform.rotation)
    transform_ccw = carla.Transform(carla.Location(), carla.Rotation(yaw=-90.))
    matrix = np.array(transform_to_world.get_matrix())
    points = np.dot(matrix, points.T).T
    matrix = np.array(transform_ccw.get_matrix())
    points = np.dot(matrix, points.T).T
    points = points[:, :3] \
        + carlautil.location_to_ndarray(carla.Location(z=2.5))
    points[:, 1] *= -1.0
    return points, labels
Exemplo n.º 4
0
    def run_scene(self, scene_idx, spawn_point):
        """Sample frames from a scene where the car is placed on some spawn point.
        """
        vehicle = None
        pedestrian = None
        camera = None
        viewing_center = None
        spectator = self.world.get_spectator()

        # set vehicle blue prints
        veh_blueprints = self.get_vehicle_blueprints()
        blueprint = random.choice(veh_blueprints)
        if blueprint.has_attribute("color"):
            color = random.choice(
                blueprint.get_attribute("color").recommended_values)
            blueprint.set_attribute("color", color)

        try:
            # create a vehicle
            has_vehicle = False
            spawn_point = self.__get_random_spawn_point(scene_idx)
            for _ in range(20):
                _spawn_point = carlautil.carlacopy(spawn_point)
                _spawn_point = self.__perturb_spawn_point(_spawn_point)
                vehicle = self.world.try_spawn_actor(blueprint, _spawn_point)
                if vehicle:
                    has_vehicle = True
                    spawn_point = _spawn_point
                    break

            if not has_vehicle:
                logging.warning("Failed to spawn vehicle with perturb at "
                                f"{carlautil.transform_to_str(spawn_point)}; "
                                "spawning vehicle without perturb.")
                vehicle = self.world.spawn_actor(blueprint, spawn_point)

            if self.show_demos:
                self.world.wait_for_tick()
            else:
                self.world.tick()
            vehicle.apply_control(carla.VehicleControl(hand_brake=True))

            if self.add_pedestrian:
                # reset the center of the scene
                off_x = random.uniform(-self.max_shift, self.max_shift)
                off_y = random.uniform(-self.max_shift, self.max_shift)
                viewing_center = carla.Transform(
                    location=spawn_point.location +
                    carla.Vector3D(x=off_x, y=off_y),
                    rotation=spawn_point.rotation)

                if self.debug:
                    self.__set_random_frame(spectator, viewing_center)

                ped_blueprints = self.get_pedestrian_blueprints()
                extent = vehicle.bounding_box.extent
                r = vehicle.get_transform().rotation
                f = carlautil.location_to_ndarray(
                    r.get_forward_vector()) * extent.x
                r = carlautil.location_to_ndarray(
                    r.get_right_vector()) * extent.y
                l = carlautil.location_to_ndarray(vehicle.get_location())
                vertices = np.array(
                    (l + f + r, l - f + r, l - f - r, l + f - r))
                vertices = util.npu.order_polytope_vertices(vertices[:, :2])
                forbidden = shapely.geometry.Polygon(vertices)

                has_pedestrian = False
                spl = np.array(
                    [viewing_center.location.x, viewing_center.location.y])
                box = np.array([[-1, -1], [1, -1], [1, 1], [-1, 1]
                                ]) * self.ped_halfwidth
                _box = None
                for _ in range(20):
                    for _ in range(50):
                        off_x = random.uniform(-self.max_shift, self.max_shift)
                        off_y = random.uniform(-self.max_shift, self.max_shift)
                        _spl = spl + np.array([off_x, off_y])
                        _box = box + _spl
                        _poly = shapely.geometry.Polygon(_box)
                        intersection = forbidden.intersection(_poly)
                        if intersection.is_empty:
                            z = viewing_center.location.z + 0.8

                            yaw = math.degrees(random.uniform(0, 2 * math.pi))
                            _spawn_point = carla.Transform(
                                location=carla.Location(x=_spl[0],
                                                        y=_spl[1],
                                                        z=z),
                                rotation=carla.Rotation(yaw=yaw))
                            blueprint = random.choice(ped_blueprints)
                            if self.debug:
                                carlautil.debug_polytope_on_xy_plane(
                                    self.world, vertices, z, life_time=5.)
                                carlautil.debug_polytope_on_xy_plane(
                                    self.world, _box, z, life_time=5.)
                            pedestrian = self.world.try_spawn_actor(
                                blueprint, _spawn_point)
                            break

                    if pedestrian:
                        has_pedestrian = True
                        break

                if not has_pedestrian:
                    logging.warning(
                        "Failed to spawn pedestrian at "
                        f"{carlautil.transform_to_str(viewing_center)}; "
                        "skipping pedestrian placement.")

            else:
                viewing_center = spawn_point

            if self.show_demos:
                self.world.wait_for_tick()
                self.demo_azimuthal_rotation(self.world, spectator, vehicle)
                self.demo_polar_rotation(spectator, vehicle)
                self.demo_radius(spectator, vehicle)
                self.demo_shift(spectator, vehicle)
                self.demo_night_lights(spectator, vehicle)
                self.demo_all_lights(spectator, vehicle)
            else:
                self.world.tick()
                self.__set_random_weather()
                # wait for car jiggle to stop
                for _ in range(20):
                    self.world.tick()
                self.__set_random_frame(
                    spectator,
                    viewing_center,
                    xy_shift_camera=not self.add_pedestrian)
                self.world.tick()
                camera, save_snapshot, scene_id = self.attach_camera_to_spectator(
                    scene_idx)
                camera.listen(save_snapshot)
                self.world.tick()
                for jdx in range(self.n_frames):
                    self.__set_random_frame(
                        spectator,
                        viewing_center,
                        xy_shift_camera=not self.add_pedestrian)
                    self.world.tick()
                camera.stop()
                self.world.tick()

        finally:
            if vehicle:
                vehicle.destroy()
            if pedestrian:
                pedestrian.destroy()
            if camera:
                camera.destroy()
            if self.show_demos:
                self.world.set_weather(self.original_weather)
                self.world.wait_for_tick()
            else:
                self.world.tick()
            time.sleep(2)  # wait to avoid crashing