feat(perception_rviz_plugin): add 2d dummy bus with height

[taikitanaka3]

Description

current 2d car is moving by default but for some functions in avoidance , no stopping are , intersection occlusion spot needs stopping obstacle so I add rviz plugins for stopping bus.

• add 2d dummy stopping bus
  

Related links

Tests performed

by psim

Notes for reviewers

run psim and add panel from + button

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.

[codecov]

Codecov Report

Merging #903 (8f26b4b) into main (3bcb248) will decrease coverage by 0.01%.
The diff coverage is 0.00%.

@@           Coverage Diff            @@
##            main    #903      +/-   ##
========================================
- Coverage   9.54%   9.53%   -0.02%     
========================================
  Files        933     933              
  Lines      57727   57815      +88     
  Branches   10408   10408              
========================================
  Hits        5510    5510              
- Misses     47687   47775      +88     
  Partials    4530    4530              

Flag	Coverage Δ		*Carryforward flag
differential	0.00% <0.00%> (?)
total	9.54% <0.00%> (ø)		Carriedforward from fd7d7e1

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

Impacted Files	Coverage Δ
...ier4_perception_rviz_plugin/src/tools/car_pose.cpp	0.00% <0.00%> (ø)
...ption_rviz_plugin/src/tools/interactive_object.hpp	0.00% <ø> (ø)
..._perception_rviz_plugin/src/tools/unknown_pose.cpp	0.00% <0.00%> (ø)
...rception_rviz_plugin/src/tools/pedestrian_pose.cpp	0.00% <0.00%> (ø)
...ption_rviz_plugin/src/tools/delete_all_objects.cpp	0.00% <0.00%> (ø)
...ption_rviz_plugin/src/tools/interactive_object.cpp	0.00% <0.00%> (ø)

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update 3bcb248...8f26b4b. Read the comment docs.

[taikitanaka3]

@HiroIshida
[Q] if I put bus in rviz it's a bit heavy processing point cloud.
is there anyways to make computational cost lighter?

[HiroIshida]

@taikitanaka3

my observation

I think visualization rather than algorithm may cause the problem. Probably, the number of pointcloud to visualize causes the problem.

experiment

I tried you new rviz plugin with the setting below:

Considering the possibility that newly implemented algorithm in
#527 make the simulation slower, I embeddedd the time measurement function to the dummy perception publisher's timer callback, and measure the difference between processing time of ego-centric (new) and object centric point cloud generation methods. The result below reveals that point cloud generation algorithm does affect much here.

Note that typical processing time of the algorithms should be
object-centric without raytracing < ego-centric << object-centric with raytracing .
I omitted object-centric with raytracing in this experiment.

newer algorithm, Ego-centric (with raytracing in principle)

[dummy_perception_publisher_node-35] [INFO] [1652690826.930240479] [simulation.dummy_perception_publisher]: callback elapsed time: 32.6487
[dummy_perception_publisher_node-35] [INFO] [1652690827.027392244] [simulation.dummy_perception_publisher]: callback elapsed time: 29.8299
[dummy_perception_publisher_node-35] [INFO] [1652690827.124755941] [simulation.dummy_perception_publisher]: callback elapsed time: 27.1835
[dummy_perception_publisher_node-35] [INFO] [1652690827.223872107] [simulation.dummy_perception_publisher]: callback elapsed time: 26.2822

old algorithm, Object-centric (disabled raytracing)

[dummy_perception_publisher_node-35] [INFO] [1652691405.282255247] [simulation.dummy_perception_publisher]: callback elapsed time: 23.8785
[dummy_perception_publisher_node-35] [INFO] [1652691405.383950450] [simulation.dummy_perception_publisher]: callback elapsed time: 25.5756
[dummy_perception_publisher_node-35] [INFO] [1652691405.482130112] [simulation.dummy_perception_publisher]: callback elapsed time: 23.7911
[dummy_perception_publisher_node-35] [INFO] [1652691405.582512018] [simulation.dummy_perception_publisher]: callback elapsed time: 24.1669

I then though maybe visualization (or entailing communication) take much computational effort a lot, so I compare htop result turning on and off the rviz. The result below shows significant computational load decline after turning off rviz. I'm not certainly sure, but from the observation so far, visualization rather than algorithm should be improved in some way.

.png)

[HiroIshida]

@taikitanaka3
Let me give bit more time to probe this issue (my next part-time shift is 5/18). Now I realized that just comparing processing time between the algorithm is not enough.

[HiroIshida]

@taikitanaka3
I think you are right. Newer (ego-centric) method take up much computation time, although computation time dose not varies between the two.

when running with object centric mode: <param name="object_centric_pointcloud" value="true"/>

$ ps -aux |grep dummy_perception_publisher_node|head -1| awk '{print $2}' | xargs -i pidstat -p {} 5 100
Linux 5.14.0-1036-oem (stonet4) 	05/18/2022 	_x86_64_	(12 CPU)

05:41:54 AM   UID       PID    %usr %system  %guest   %wait    %CPU   CPU  Command
05:41:59 AM  1000    188562    3.80    0.00    0.00    0.40    3.80     6  dummy_perceptio
05:42:04 AM  1000    188562    4.00    0.00    0.00    0.00    4.00     5  dummy_perceptio
05:42:09 AM  1000    188562    4.00    0.00    0.00    0.20    4.00     3  dummy_perceptio
05:42:14 AM  1000    188562    4.00    0.00    0.00    0.20    4.00     6  dummy_perceptio
05:42:19 AM  1000    188562    3.80    0.00    0.00    0.40    3.80     7  dummy_perceptio
^C
Average:     1000    188562    3.92    0.00    0.00    0.24    3.92     -  dummy_perceptio

when running with ego-centric mode: <param name="object_centric_pointcloud" value="false"/>

$ ps -aux |grep dummy_perception_publisher_node|head -1| awk '{print $2}' | xargs -i pidstat -p {} 5 100
Linux 5.14.0-1036-oem (stonet4) 	05/18/2022 	_x86_64_	(12 CPU)

05:46:31 AM   UID       PID    %usr %system  %guest   %wait    %CPU   CPU  Command
05:46:36 AM  1000    199150   13.60    0.40    0.00    0.80   14.00     5  dummy_perceptio
05:46:41 AM  1000    199150   13.60    1.00    0.00    0.60   14.60     7  dummy_perceptio
05:46:46 AM  1000    199150   13.20    0.80    0.00    0.60   14.00     3  dummy_perceptio
05:46:51 AM  1000    199150   13.20    1.00    0.00    0.80   14.20     4  dummy_perceptio
05:46:56 AM  1000    199150   13.40    1.20    0.00    0.60   14.60     9  dummy_perceptio
^C
Average:     1000    199150   13.40    0.88    0.00    0.68   14.28     -  dummy_perceptio

As a workaround, you can set <param name="object_centric_pointcloud" value="true"/> to reduce the computational load. But by doing this, the generated pointcloud may be poor as previously mentioned in #527

I have some idea to reduce the computational load when ego-centric mode. Let me some time to try the idea and make a new PR.

[HiroIshida]

@taikitanaka3
I made a PR that reduce the aboud 50% computational load. (in the measurement, I used this PR's bus objects)
#926

[taikitanaka3]

@HiroIshida
Thank you I will take a look and test this at today evening.

[shmpwk]

@taikitanaka3
Could you add info about this nice feature at README?

[shmpwk]

I added icon for 2d Dummy Bus rviz tool by 8f26b4b