Every frame the server sends a package with the measurements and images gathered to the client. This document describes the details of these measurements.
Since CARLA can be run at fixed-frame rate, we keep track of two different time-stamps.
|platform_timestamp||uint32||Time-stamp of the current frame, in milliseconds as given by the OS.|
|game_timestamp||uint32||In-game time-stamp, milliseconds elapsed since the beginning of the current level.|
In real-time mode, the elapsed time between two time steps should be similar both platform and game time-stamps. When run in fixed-time step, the game time-stamp increments in constant time steps (delta=1/FPS) while the platform time-stamp keeps the actual time elapsed.
|transform||Transform||World transform of the player.|
|acceleration||Vector3D||Current acceleration of the player.|
|forward_speed||float||Forward speed in km/h.|
|collision_vehicles||float||Collision intensity with other vehicles.|
|collision_pedestrians||float||Collision intensity with pedestrians.|
|collision_other||float||General collision intensity (everything else but pedestrians and vehicles).|
|intersection_otherlane||float||Percentage of the car invading other lanes.|
|intersection_offroad||float||Percentage of the car off-road.|
|autopilot_control||Control||Vehicle's autopilot control that would apply this frame.|
The transform contains the location and rotation of the player.
|orientation [deprecated]||Vector3D||Orientation in Cartesian coordinates.|
|rotation||Rotation3D||Pitch, roll, and yaw.|
Collision variables keep an accumulation of all the collisions occurred during this episode. Every collision contributes proportionally to the intensity of the collision (norm of the normal impulse between the two colliding objects).
Three different counts are kept (pedestrians, vehicles, and other). Colliding objects are classified based on their tag (same as for semantic segmentation).
Collisions are not annotated if the vehicle is not moving (<1km/h) to avoid annotating undesired collision due to mistakes in the AI of non-player agents.
The lane intersection measures the percentage of the vehicle invading the opposite lane. The off-road intersection measures the percentage of the vehicle outside the road.
These values are computed intersecting the bounding box of the vehicle (as a 2D rectangle) against the map image of the city. These images are generated in the editor and serialized for runtime use. You can find them too in the release package under the folder "RoadMaps".
autopilot_control measurement contains the control values that the in-game
autopilot system would apply as if it were controlling the vehicle.
This is the same structure used to send the vehicle control to the server.
|steer||float||Steering angle between [-1.0, 1.0] (*)|
|throttle||float||Throttle input between [ 0.0, 1.0]|
|brake||float||Brake input between [ 0.0, 1.0]|
|hand_brake||bool||Whether the hand-brake is engaged|
|reverse||bool||Whether the vehicle is in reverse gear|
To activate the autopilot from the client, send this
to the server. Note that you can modify it before sending it back.
measurements, sensor_data = carla_client.read_data() control = measurements.player_measurements.autopilot_control # modify here control if wanted. carla_client.send_control(control)
(*) The actual steering angle depends on the vehicle used. The default Mustang has a maximum steering angle of 70 degrees (this can be checked in the vehicle's front wheel blueprint).
Non-player agents info
To receive info of every non-player agent in the scene every frame you need to activate this option in the settings file sent by the client at the beginning of the episode.
If enabled, the server attaches a list of agents to the measurements package every frame. Each of these agents has an unique id that identifies it, and belongs to one of the following classes
- Vehicle Contains its transform, box-extent, and forward speed.
- Pedestrian Contains its transform, box-extent, and forward speed. (*)
- Traffic light Contains its transform and state (green, yellow, red).
- Speed-limit sign Contains its transform and speed-limit.
(*) At this point every pedestrian is assumed to have the same bounding-box size.
Transform and bounding box
The transform defines the location and orientation of the agent. The bounding box is assumed to be centered at the agent's location. The box extent gives the radii dimensions of the bounding box of the agent.
As seen in the picture, the Z coordinate of the box is not fitted to vehicle's height.