Skip to content

Conversation

@WaelDLZ
Copy link

@WaelDLZ WaelDLZ commented Jan 22, 2026

This PR makes a few changes so the functions we use in the WOSAC evaluation code can have access to the original scenario_id and object_ids from the WOMD data.

This enables to evaluate any Sim Agents method using our code, by matching the scenario_ids and object_ids between the simulated rollouts and our ground truth trajectories.

It also makes the WOSAC code more readable by explicitly adding the is_track_to_predict flag.

This mostly changes functions only used in WOSAC so it shouldn't break anything.

EXCEPT for the fact that it asks to regenerate binaries, also if the CARLA maps do not have a scenario_id weird things might happen. I didn't think about it yet (it should be easy to put a default value for maps that dont have a scenario_id though)

…the Wosac bindings

- Requires to recompile the maps though
- Didn't think about CARLA compatibility yet

Next steps:

- Do the same work with track_ids (should be easy)
- Adapt the SMART evaluation code (I hope easy as well)
…s_track_to_predict as a bool.

Next step: update the import trajectories evaluations script.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants