-
Support replaying multiple cooperative perception dataset in CARLA
-
Help you add new sensors to the agents to have additional sensor data, including but not limited to:
- Depth sensor
- Event camera
- Semantic LiDAR
- Radar
-
Dump new groundtruth that the origin dataset does not have, including but not limited to:
- HDMap (CoBEVT (CoRL2022) used this groundtruth )
- Trajectory for each vehicle
-
Change the number of CAVs! The original OPV2V data has fixed CAVs, but with this tool, you can assign any vehicle to a CAV equipped with sensors.
Check the CARLA official website: https://carla.readthedocs.io/en/latest/start_quickstart/
Please change the parameters in logreplay/hypes_yaml/replay.yaml
to configure your replay.
We have provided very detailed explanation for each line in the yaml.
We have by default provide Semantic Lidar and BEV semantic camera sensors in the yaml file and codebase.
If you want to add some new sensors, please add your sensor under logreplay/sensors/
and refer to the semantic_lidar
or bev_semantic_camera.py
to write your own sensor class.
Start your CARLA simulator first. Right now we only support version 0.9.12
Execute the following command:
cd logreplay
python scenario/scenarios_manager.py