fix(autoware_image_projection_based_fusion): resolve issue with segmentation pointcloud fusion node failing with multiple mask inputs

[vividf]

Description

This PR addresses issue #8460. The current design of the segmentation pointcloud fusion node is incompatible with the fusion base class because the fuseOnSingleImage function always uses the original input pointcloud in each iteration (camera), instead of the pointcloud that has filtered out points that fall outside the mask before processing. In the end, the current algorithm will lead to the problem that the output pointcloud will only filter out the points with the last camera's mask, instead of all cameras' masks

A simple fix would be to copy the output pointcloud and use it as the input for the next iteration. However, this approach increases processing time by approximately 0.2 ms per iteration (camera). Moreover, it impacts four other nodes in the autoware_image_projection_based_fusion package.

To avoid these drawbacks, I opted to modify the algorithm directly within the segmentation pointcloud fusion node. The new solution involves storing the offsets of points that are filtered out in a set. During post-processing, the points in the set are ignored, and the remaining points are copied to the output pointcloud.

Related links

Parent Issue:

• Link: segmentation_pointcloud_fusion can't work proper with multiple mask #8460

How was this PR tested?

I test the PR with xx1 bag with using three camera (0, 1, 2)

launch yolox segmentation (modify multiple_yolox.launch.xml by adding output/mask)

ros2 launch autoware_tensorrt_yolox multiple_yolox.launch.xml

launch fusion node

ros2 launch autoware_image_projection_based_fusion segmentation_pointcloud_fusion.launch.xml

launch the xx1 tf

ros2 launch static_tf_publisher static_transforms_launch.py

Before (only filter out the camera 2 information (left) )

After

Notes for reviewers

None.

Interface changes

None.

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.

[codecov]

Codecov Report

Attention: Patch coverage is 0% with 13 lines in your changes missing coverage. Please review.

Project coverage is 25.02%. Comparing base (4cc1f75) to head (8820ac6).

Files with missing lines	Patch %	Lines
...fusion/src/segmentation_pointcloud_fusion/node.cpp	0.00%	13 Missing ⚠️

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #8769      +/-   ##
==========================================
- Coverage   25.02%   25.02%   -0.01%     
==========================================
  Files        1322     1322              
  Lines       97979    97990      +11     
  Branches    37811    37815       +4     
==========================================
+ Hits        24521    24522       +1     
- Misses      70992    71001       +9     
- Partials     2466     2467       +1     

Flag	Coverage Δ		*Carryforward flag
differential	4.33% <0.00%> (?)
total	25.03% <ø> (+<0.01%)	⬆️	Carriedforward from 4cc1f75

*This pull request uses carry forward flags. Click here to find out more.

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[badai-nguyen]

@vividf I really appreciate your PR.
LGTM