feat(autoware_image_based_projection_fusion): redesign image based projection fusion node

[vividf]

Description

This PR refactors the fusion node, introducing several new features:

• Parsing diagnostics from the concatenate node to adjust the offset between 3D messages and ROIs dynamically.
• Implementing a stricter matching strategy.
• Allowing the collector to control when the fusion process is executed.
• Publishing diagnostics for the fusion status

More information is provided in the readme

Related links

Parent Issue:
TIER IV INTERNAL link

PRs that need to be merged together with this PR

• autoware_launch vehicle 1
• autoware_launch vehicle 2
• autoware_launch general

How was this PR tested?

Evaluator

Diagnostics

Comparison

Rosbag: TIER4_ROSBAG

Frame comparison

New algorithm

Old algorithm

Video comparison

New algorithm vs Old algorithm

Processing time

Almost the same (1-2 ms difference)
new:

old:

• Note that this cannot be tested directly with the current (old) algorithm because the trigger for processing time is incorrect, making it impossible to accurately measure the time from preprocessing through fusion to postprocessing.
  

Notes for reviewers

How to run with awf-latest in pilot.auto

• Follow the TIER4_INTERNAL_LINK to build the project from main branch.
• Build the project
• Replace the universe
• Build the autoware_pointcloud_preprocessor and autoware_image_projection_based_fusion

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.

[github-actions]

Documentation URL: https://autowarefoundation.github.io/autoware.universe/pr-10016/
Modified URLs:

• https://autowarefoundation.github.io/autoware.universe/pr-10016/perception/autoware_image_projection_based_fusion/

[codecov]

Codecov Report

Attention: Patch coverage is 0.88810% with 558 lines in your changes missing coverage. Please review.

Project coverage is 26.35%. Comparing base (b7acf99) to head (3f01fcf).
Report is 1 commits behind head on main.

Files with missing lines	Patch %	Lines
..._image_projection_based_fusion/src/fusion_node.cpp	0.00%	256 Missing ⚠️
...tion_based_fusion/src/fusion_matching_strategy.cpp	0.00%	146 Missing ⚠️
...e_projection_based_fusion/src/fusion_collector.cpp	0.00%	117 Missing ⚠️
...sed_fusion/src/roi_detected_object_fusion/node.cpp	0.00%	11 Missing ⚠️
...on_based_fusion/src/roi_pointcloud_fusion/node.cpp	0.00%	8 Missing ⚠️
...ion_based_fusion/src/pointpainting_fusion/node.cpp	0.00%	6 Missing ⚠️
...ction_based_fusion/src/roi_cluster_fusion/node.cpp	0.00%	5 Missing ⚠️
...image_projection_based_fusion/fusion_collector.hpp	0.00%	3 Missing ⚠️
...fusion/src/segmentation_pointcloud_fusion/node.cpp	0.00%	3 Missing ⚠️
...ojection_based_fusion/fusion_matching_strategy.hpp	0.00%	2 Missing ⚠️
... and 1 more

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #10016      +/-   ##
==========================================
- Coverage   26.48%   26.35%   -0.14%     
==========================================
  Files        1386     1391       +5     
  Lines      107338   107495     +157     
  Branches    41270    41332      +62     
==========================================
- Hits        28430    28330     -100     
- Misses      76074    76304     +230     
- Partials     2834     2861      +27     

Flag	Coverage Δ		*Carryforward flag
differential	20.62% <0.71%> (?)
differential-cuda	16.96% <0.71%> (?)
total	26.39% <100.00%> (-0.10%)	⬇️	Carriedforward from 3ba2da8

*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.

[technolojin]

I updated the branch but still build test issue persist.
#10198

@vividf Can you check it if the branch is based on the updated main branch???

[technolojin]

LGTM

Dear code owners, please review this PR.
This PR improves message synchronization so that overall latency decreases.

[vividf]

@technolojin
I updated the parameters in 842eabc , 0f4c027, and 01b076b to match the sample example explained in the README.
Initially, I planned to set the parameters for xx1 gen1, but the offset settings differ significantly from our other project. I felt that using parameters from an older project in our sample example might not be ideal for understanding.

So, I created a fake sample and set corresponding fake parameters instead.

I will set the parameters for each project in their respective launchers.

[vividf]

Regarding how to test this PR with the launcher.

• Follow the TIER4_INTERNAL_LINK to build the project from main branch.
• Build the project
• Replace the universe
• Build the autoware_pointcloud_preprocessor and autoware_image_projection_based_fusion

[YoshiRi]

Thanks for your tough works! I have few comments. Please check it.

[YoshiRi]

Since fractional_timestamp_set_ is std::set and elements in set are const, this part won't work actually.

Try erase and re-insert, or replace data structure like vector.

  for (auto it = fractional_set.begin(); it != fractional_set.end(); ++it) {
    if (std::abs(*it - fractional_part) < tolerance) {
      // If it belongs to an existing group, average the timestamp
      double updated_value = (*it + fractional_part) / 2.0;
      fractional_set.erase(it);
      fractional_set.insert(updated_value);
      return;
    }
  }

[vividf]

You are right! Thanks a lot!
Fixed in 9213beb

[YoshiRi]

I do not understand why did you want to do this for now. I do not agree this for several reasons.
We can discuss this in another PR.

[vividf]

Thanks!
I changed this for testing, but I forgot to change it back!
Fixed in ac52b2f

[technolojin]

[YoshiRi]

LGTM