perf(voxel_grid_downsample_filter): performance tuning

[atsushi421]

Description

This PR makes voxel_grid_downsample_filter faster without changing the logical output.
The tail latency of the voxel_grid_downsample_filter node gets about x2.3 faster with this PR merged.

Measurement Condition

• Ubuntu22.04 + ROS2 Humble + Autoware Universe rosbag simulation
• Core Isolated
• Core Frequency Fixed (2.3GHz)
• L3 Cache: 24MB

Related links

• Details and performance analysis of previous implementations.

Tests performed

I have comfirmed the following between point clouds with the same timestamp before and after this PR:

• the number of point clouds is the same
• the distance between the nearest point is less than 0.001
• the 3D plot results are almost identical

To reproduce: https://drive.google.com/drive/folders/1mON3GN-2v8uUPjw1oejasflOP_O-V0kc?usp=share_link

Notes for reviewers

Pre-review checklist for the PR author

The PR author must check the checkboxes below when creating the PR.

• I've confirmed the contribution guidelines.
• The PR follows the pull request guidelines.

In-review checklist for the PR reviewers

The PR reviewers must check the checkboxes below before approval.

• The PR follows the pull request guidelines.
• The PR has been properly tested.
• The PR has been reviewed by the code owners.

Post-review checklist for the PR author

The PR author must check the checkboxes below before merging.

• There are no open discussions or they are tracked via tickets.
• The PR is ready for merge.

After all checkboxes are checked, anyone who has write access can merge the PR.

[yukkysaito]

@atsushi421 Thank you for your PR 👍 I just commented

[atsushi421]

@yukkysaito
Thank you for your review!
I have carved out the filtering process into a separate class to improve readability.
It is also possible to switch to the PCL voxel grid filter by calling filter() instead of faster_filter().

[yukkysaito]

Suggested change

	// Copyright 2020 Tier IV, Inc.
	// Copyright 2023 TIER IV, Inc.

[atsushi421]

Fixed in 6451075.

[yukkysaito]

@atsushi421 Looks like no change 🙏

[yukkysaito]

@atsushi421 Thank you for your modification 👍 it is looks like almost good!

I compared the publish messages before and after this PR and confirmed that the data sizes for the same timestamps matched.

Are you able to confirm that the point clouds are the same, not just the data size? 🙏

[atsushi421]

@atsushi421 Thank you for your modification +1 it is looks like almost good!

I compared the publish messages before and after this PR and confirmed that the data sizes for the same timestamps matched.

Are you able to confirm that the point clouds are the same, not just the data size? pray

@yukkysaito
Thank you for your considerate comments.
Upon inspection, it was observed that the point clouds did not match exactly, most likely due to the floating point calculation process.
As an alternative approach, a threshold was defined for the point-to-point distance, and it was verified that the change in coordinates between the points before and after this pull request remained below the threshold.
Additionally, point clouds with the same timestamp were plotted on shared 3D coordinates to visually confirm minimal discrepancies.
For further details, please refer to the "Tests Performed" section of this pull request.

[yukkysaito]

LGTM

[ralwing]

There is a typo in filename: dowNsample

[ralwing]

typo

[atsushi421]

@ralwing
Thank you for pointing this out.
I have corrected the typo (e5d40f0).

[codecov]

Codecov Report

Patch coverage has no change and project coverage change: -0.02 ⚠️

Comparison is base (6c6973b) 14.38% compared to head (8742484) 14.37%.

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #3679      +/-   ##
==========================================
- Coverage   14.38%   14.37%   -0.02%     
==========================================
  Files        1563     1566       +3     
  Lines      107631   107725      +94     
  Branches    31175    31175              
==========================================
  Hits        15481    15481              
- Misses      75325    75419      +94     
  Partials    16825    16825              

Flag	Coverage Δ		*Carryforward flag
differential	7.52% <0.00%> (?)
total	14.38% <ø> (+<0.01%)	⬆️	Carriedforward from 6c6973b

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

Impacted Files	Coverage Δ
...cessor/crop_box_filter/crop_box_filter_nodelet.hpp	0.00% <ø> (ø)
...ple_filter/faster_voxel_grid_downsample_filter.hpp	0.00% <0.00%> (ø)
...ocessor/include/pointcloud_preprocessor/filter.hpp	0.00% <ø> (ø)
...sor/outlier_filter/ring_outlier_filter_nodelet.hpp	0.00% <ø> (ø)
...include/pointcloud_preprocessor/transform_info.hpp	0.00% <0.00%> (ø)
...ple_filter/faster_voxel_grid_downsample_filter.cpp	0.00% <0.00%> (ø)
...le_filter/voxel_grid_downsample_filter_nodelet.cpp	0.00% <0.00%> (ø)
sensing/pointcloud_preprocessor/src/filter.cpp	7.84% <0.00%> (ø)

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.