CARLA生成全局路径规划的代码在

carla/PythonAPI/carla/agents/navigation

在自己的carla客户端py文件中

from agents.navigation.basic_agent import BasicAgent  # pylint: disable=import-error

如果是pycharm开发，要在pycharm的Settings - Project Structure, Add Content Root中加入agents的上级文件夹。

获得所有的生成点

spawn_points = world.get_map().get_spawn_points()

在其中选择一个起点和一个终点

start_location = spawn_points[0]
destination = spawn_points[38]

 调用BasicAgent的set_destination函数，生成一个全局导航路径。

agent = BasicAgent(vehicle, 30)
global_route_trace = agent.set_destination(destination.location, start_location=start_location)

上面代码的vehicle是一个汽车的类实例（如何生成一个汽车，参见博主另一篇文章：在CARLA中手动开车，添加双目相机stereo camera，激光雷达Lidar-CSDN博客）

设置观察视角

spectator = world.get_spectator()
transform = carla.Transform()
bv_transform = carla.Transform(transform.location + carla.Location(z=250,x=0), carla.Rotation(yaw=0, pitch=-90))
spectator.set_transform(bv_transform)

将路径可视化，画出上图的效果

world.debug.draw_string(destination.location, str("destination"), life_time=100)
world.debug.draw_arrow(destination.location, destination.location + destination.get_forward_vector(), life_time=100)

for tup_i in global_route_trace:
    waypoint_i = tup_i[0]
    location = waypoint_i.transform.location
    world.debug.draw_arrow(location, location + waypoint_i.transform.get_forward_vector(), life_time=100)

以上！