Configuration

Scenario Configuration Overview

The general scenario config files, such as demo.yaml, define various parameters for the simulation. Below is a detailed explanation of each item:

1. common_params

   Purpose: Defines common parameters shared across all modules.

   Modules:

   • common_param The basic version of parameters.

   • common_params_edge_on Parameters for edge units being activated.

2. vehicle_perception

   Purpose: Configures how the vehicle perceives its environment.

   Modules:

   • perception_false Object positions are retrieved directly from the server.

   • perception_true Object positions are retrieved through perception.

3. vehicle_localization

   Purpose: Manages vehicle localization for determining its precise position.

   Modules:

   • localization_false Ego position is retrieved from the server.

   • localization_true Ego position is retrieved via localization algorithms.

4. game_map

   Purpose: Controls the traffic map state.

   Modules:

   • map_activated Activates the traffic map.

   • map_deactivated Deactivates the traffic map.

5. behavior

   Purpose: Configures vehicle behavior, including planning and maximum speed.

   Modules:

   • behavior_default Enables automated planning for vehicles.

   • behavior_demo_manual Uses manually provided trajectories from files.

   • behavior_scenario Configures the behavior for OpenScenario.

6. controller

   Purpose: Handles vehicle control based on the planned trajectory.

   Modules:

   • pid_controller Uses a PID controller for vehicle movement.

7. traffic_manager

   Purpose: Manages interactions with traffic, including lane changes and traffic light rules.

   Modules:

   • lane_change_false Disables automatic lane changes.

   • lane_change_true Enables automatic lane changes.

8. scenario

   Purpose: Defines testing scenarios with pre-defined destinations and perception info.

   Modules:

   The name should be consistent with the name of the scenario.

Behavior Configuration

The behavior config files, such as behavior_coop_perception.yaml, specify detailed configurations for vehicle behavior and planning.

• is_manually Specifies if the behavior is manual (False for automated).

• max_speed Maximum vehicle speed in km/h.

• tailgate_speed Speed for approaching another vehicle ASAP.

• max_speed_margin Margin for calculating the target speed (max_speed - margin).

• speed_decrease Speed reduction during car-following mode for distance keeping.

• safety_time Time-to-collision (TTC) threshold for reducing speed.

• emergency_param Threshold for initiating an emergency stop.

• ignore_traffic_light Allows ignoring traffic lights.

• overtake_allowed Indicates if overtaking is allowed.

• collision_time_ahead Time for collision checking.

• overtake_counter_recover Cooldown period for overtaking.

• sample_resolution Distance between two waypoints.

• buffer_size Number of waypoints in the planning buffer.

• trajectory_update_freq Frequency of trajectory point updates.

• trajectory_sample_horizon Time horizon for trajectory sampling.

• global_route_update_freq Frequency for updating the global route.

• trajectory_dt Time step for sampling trajectory points.

• min_dist Minimum distance to remove close waypoints.

• debug Enables visual debugging of waypoints.

• debug_trajectory Enables visual debugging of trajectory points and paths.

Controller Configuration

The controller config details the parameters for PID controller.

• controller.type Specifies the controller type. Supported: PID.

• controller.args.lat.k_p Proportional gain for lateral control.

• controller.args.lat.k_d Derivative gain for lateral control.

• controller.args.lat.k_i Integral gain for lateral control.

• controller.args.lon.k_p Proportional gain for longitudinal control.

• controller.args.lon.k_d Derivative gain for longitudinal control.

• controller.args.lon.k_i Integral gain for longitudinal control.

• controller.args.dt Time step for PID computation. Should match fixed_delta_seconds.

• controller.args.max_brake Maximum braking force (0 to 1.0).

• controller.args.max_throttle Maximum throttle force (0 to 1.0).

• controller.args.max_steering Maximum steering angle in radians.

• controller.args.max_steering_rate Maximum rate of steering change.

• controller.args.time_interval Time interval for applying control signals.

Localization Configuration

The localization configuration file contains settings for positioning and motion tracking of the ego vehicle, including GNSS-based localization and Kalman filter parameters.

• localization.activate Activates localization. If False, ego position is retrieved from the server.

• localization.dt Time step for Kalman filter. Matches fixed_delta_seconds.

• localization.gnss.noise_alt_stddev Standard deviation of altitude noise in meters.

• localization.gnss.noise_lat_stddev Standard deviation of latitude noise in radians.

• localization.gnss.noise_lon_stddev Standard deviation of longitude noise in radians.

• localization.gnss.heading_direction_stddev Standard deviation of heading noise in degrees.

• localization.gnss.speed_stddev Standard deviation of speed noise in m/s.

• localization.debug_helper.show_animation Enables real-time trajectory visualization.

• localization.debug_helper.x_scale Multiplier for x-coordinate visualization.

• localization.debug_helper.y_scale Multiplier for y-coordinate visualization.

Map Configuration

The map config defines the settings for the road map visualization.

• game_map.pixels_per_meter Number of pixels per meter in the map raster.

• game_map.raster_size Raster dimensions in pixels [width, height].

• game_map.lane_sample_resolution Distance between sampled waypoints along a lane.

• game_map.visualize Enables visualization of the map.

• game_map.activate Activates the map for planning and visualization.

Perception Configuration

The perception configuration file sets parameters for sensing, including cameras, LiDAR, and the YOLO model for object detection.

• perception.activate Enables perception module. If False, uses oracle data.

• perception.model Perception model type (e.g., yolo).

• perception.camera.visualize Number of camera images to visualize (0 for none).

• perception.camera.num Number of cameras mounted (up to 3: frontal, left, and right).

• perception.lidar.visualize Enables LIDAR visualization using Open3D.

• perception.lidar.channels Number of LIDAR channels.

• perception.lidar.range Maximum range of LIDAR in meters.

• perception.lidar.points_per_second Number of points emitted per second.

• perception.lidar.rotation_frequency LIDAR rotation frequency in Hz.

• perception.lidar.upper_fov Upper field-of-view limit in degrees.

• perception.lidar.lower_fov Lower field-of-view limit in degrees.

• perception.lidar.dropoff_general_rate General drop-off rate for point intensity.

• perception.lidar.dropoff_intensity_limit Minimum intensity before dropping points.

• perception.lidar.dropoff_zero_intensity Intensity value treated as zero.

• perception.lidar.noise_stddev Standard deviation of noise in LIDAR measurements.

Traffic Manager Configuration

The traffic manager configuration file controls the background vehicles, including speed adjustments, range of traffic flow, and lane-changing policies.

• carla_traffic_manager.sync_mode Synchronizes with the world setting.

• carla_traffic_manager.deterministic_mode Enables deterministic behavior for repeatability.

• carla_traffic_manager.deterministic_seed Seed value for deterministic mode.

• carla_traffic_manager.global_distance Minimum distance in meters to maintain between vehicles.

• carla_traffic_manager.global_speed_perc Speed adjustment relative to the default speed.

• carla_traffic_manager.set_osm_mode Enables OpenStreetMap (OSM) mode for navigation.

• carla_traffic_manager.auto_lane_change Allows automatic lane changes. Set to False in this configuration.

• carla_traffic_manager.random Randomizes vehicle colors and models.

• carla_traffic_manager.ignore_lights_percentage Percentage of vehicles ignoring traffic lights.

• carla_traffic_manager.vehicle_list Defines vehicles for simulation (can be a list or number).

• carla_traffic_manager.range Vehicle spawn range: [x_min, x_max, y_min, y_max, x_step, y_step].

World Settings

The world configuration file defines the environmental and simulation settings, including rendering, weather, and randomness.

• world.sync_mode Enables synchronization mode for the world.

• world.client_host Hostname or IP address of the CARLA client (default: localhost).

• world.client_port Port number for the CARLA client connection (default: 2000).

• world.no_rendering_mode Disables rendering to improve simulation performance.

• world.fixed_delta_seconds Fixed time step for the simulation in seconds.

• world.seed Random seed for reproducibility.

• world.weather.sun_altitude_angle Sun altitude angle in degrees. 90 is midday, and -90 is midnight.

• world.weather.cloudiness Cloud cover percentage. 0 represents a clear sky, and 100 represents thick clouds.

• world.weather.precipitation Rain intensity percentage. 0 means no rain, and 100 means the heaviest rain.

• world.weather.precipitation_deposits Amount of water puddles created on the road. 0 represents no puddles, and 100 represents roads completely covered with water.

• world.weather.wind_intensity Wind intensity percentage, influencing the behavior of rain. 0 means no wind.

• world.weather.fog_density Thickness of the fog. 100 is the thickest fog.

• world.weather.fog_distance Distance in meters at which fog begins. 0 means fog starts immediately.

• world.weather.fog_falloff Density of the fog. Higher values make the fog denser and heavier, limiting its height.

• world.weather.wetness Wetness level of the environment. 0 means completely dry conditions.