feat(autoware_planning_validator): add collision checking for predicted objects

[kyoichi-sugahara]

Description

launcher PR should be merged first.

Execute collision detection at each time step based on predicted object movements by checking the planning module output.

Currently, this feature is disabled due to frequent collision detections. We will enable this functionality once we confirm the detection mechanism is sufficiently reliable and beneficial.

Even in the current state, error information is included in the diagnostic topics, allowing for planning evaluation using this data.

How was this PR tested?

run scenario_simulator_v2 and confirmed that detecting collision.

diag_error_trajectory_validation_trajectory_collision_buffered-.2024-11-01-16-52-13.mp4

Looking at the time cross-section, locations where collisions are predicted are visualized with red markers.

Notes for reviewers

None.

Interface changes

add subscriber for PredictedObjects

ROS Parameter Changes

Additions

Version	Parameter Name	Type	Default Value	Description
Added	trajectory_to_object_distance_threshold	double	30.0	A threshold for filtering objects to improve computational efficiency of collision checking. Only objects within this distance [m] from their nearest point on the trajectory are considered for collision checking.
Added	ego_to_object_distance_threshold	double	50.0	A threshold for filtering objects to improve computational efficiency of collision checking. Only objects within this distance [m] from the current ego position are considered for collision checking.
Added	time_tolerance_threshold	double	0.1	Time tolerance threshold [s] for collision detection between ego and object trajectories. Collision check is performed only when the time difference between ego and object predicted positions is within this threshold to detect temporal-spatial intersections.

Effects on system behavior

None.

[github-actions]

Thank you for contributing to the Autoware project!

🚧 If your pull request is in progress, switch it to draft mode.

Please ensure:

• You've checked our contribution guidelines.
• Your PR follows our pull request guidelines.
• All required CI checks pass before marking the PR ready for review.

[maxime-clem]

If you want to efficiently check for collisions then you should use an rtree to store and do queries on the ego trajectory footprints.

[kyoichi-sugahara]

@satoshi-ota
After visualizing the footprint of predicted path of ego vehicle and the polygon of obstacles where collision is detected, I will merge them.
Thank you so much for the corrections.