Quickstart

To quickly start using the project follow these initial steps:

1. CARLA and the repository have been setup by following the Installation instructions.

2. Begin by launching the CARLA simulator environment. Ensure it is properly configured and running on your system.

3. Next, navigate to the project directory where you have cloned the repository. Use the provided scripts and configurations to initiate the driving AI within CARLA. For a quickstart run this minimal setup script:

   python examples/minimal.py
   

   Alternatively you can construct a minimal example in the following way:

   from agents.lunatic_agent import LunaticAgent
   from classes.worldmodel import GameFramework
   
   # Sets up Carla and Pygame, and Hydra in a limited way
   game_framework = GameFramework.quickstart()
   
   ego = GameFramework.request_new_actor("car", rolename="hero", random_location=True)
   
   # Create a lunatic agent
   agent = LunaticAgent(game_framework.agent_config, vehicle=ego)
   
   try:
       while game_framework.continue_loop:
           with game_framework(agent):
               agent.run_step()
               # Use the mouse to override the agents controls; use ESC to quit
               if game_framework.controller.parse_events(agent.get_control()):
                   break
               agent.apply_control()
   except:
       # Destroy spawned actors
       game_framework.cleanup()
       raise
   

   Attention

   Note that this setup omits some steps and does not provide access too all features. Look into the AgentGameLoop.py module for a comprehensive setup.

4. Alternatively you can use the provided run_leaderboard_agent.sh script to test the LunaticChallenger.

   Tip

   The default route (Town10HD) is loaded faster, however its speed is sometimes worse than the big Town12 that takes longer to load. Consider turning off ENABLE_DATA_MATRIX or ENABLE_RSS to gain performance inside the leaderboard_agent.py or conf/agent/leaderboard.yaml.

Additionally, for controlling the driving AI during simulation, you can utilize the hotkeys implemented using Pygame. In-game, you can press the ‘H’ key to view the controls or refer to the full list of keyboard controls the Keyboard Controls.

External Setup: Using the ScenarioRunner and other Initializations

During the agents creation the script waits until an external actor is found, by default named 'hero'. This allows the LunaticAgent to be used with other scripts that set up the scenario and spawn this special actor. To initialize a route the agent should follow call:

# CARLA's BasicAgent interface
# plan: list of [carla.Waypoint, RoadOption] representing the route to be followed
lunatic_agent.set_global_plan(plan, clean_queue=True):

# leaderboard AutonomousAgent interface
# global_plan_gps: "tuple[Dict[str, float], RoadOption]"
# global_plan_world_coord: "tuple[carla.Transform, RoadOption]"
lunatic_challenger.set_global_plan(global_plan_gps, global_plan_world_coord)

Important

For performance and interface reasons the LunaticAgent uses the CarlaDataProvider from the scenario runner as a backbone to access global information.

To detect actors, all actors must be spawned over the CarlaDataProvider, further the LunaticAgent must have access to the same CarlaDataProvider object.
Use from launch_tools import CarlaDataProvider to access the same instance.

Keyboard Controls

Depending on the setup certain keys, e.g. WASD or user input, will not work, also user input can be disabled when calling agent.parse_keyboard_input(False).

Use ARROWS or WASD keys for control.

        W            : throttle
        S            : brake
        AD           : steer
        Q            : toggle reverse
        Space        : hand-brake
        P            : toggle autopilot (depends on setup)

        TAB          : change view
        Backspace    : change vehicle

        R            : toggle recording images to disk

        F2           : toggle RSS visualization mode
        F3           : increase log level
        F4           : decrease log level
        F5           : increase map log level
        F6           : decrease map log level
        B            : toggle RSS Road Boundaries Mode
        G            : RSS check drop current route (experimental)
        T            : toggle RSS (NotImplemented)
        N            : pause simulation (only slow down)

        F1           : toggle HUD
        H/?          : toggle help
        ESC          : quit