Port nav2 systems test to new gazebo

stevedanomodolor commented 1 week ago

Bug report

Required Info:

Operating System:
ROS2 Version:
Version or commit hash:
DDS implementation:

Steps to reproduce issue

Expected behavior

Actual behavior

Additional information

Feature request

Feature description

This is to keep track of the nav2 system tests that has been ported. Will be closed as soon as all the tests has been ported.
To be ported
[x] localization
[x] system
[x] system_failure
[x] updown
[x] waypoint_follower
[ ] gps_navigation
[x] behaviors/spin
[x] behaviors/wait
[x] behaviors/backup
[x] behaviors/drive_on_heading
[x] behaviors/assisted_teleop
[x] costmap_filters
[x] cleanup old files
Implementation considerations

stevedanomodolor commented 6 days ago

@SteveMacenski, so I am not going crazy, does the wait, backup, drive on heading behavior tests fail once in a while. It does not seem to be stable.

SteveMacenski commented 6 days ago

@SteveMacenski, so I am not going crazy, does the wait, backup, drive on heading behavior tests fail once in a while. It does not seem to be stable.

You're not crazy, they're poorly written and it drives me bananas. I want to rewrite them at some point (#4349) to not try to fake everything out in some crazy way. I'm not really sure why this is the way that it is, I suspect it may have been written before we had any GZ simulations setup in the systems tests to work from. It is my plan to have a new contributor rewrite them as a project, but certainly if anyone else was interested in just deleting them and restarting now with the new GZ and the Simple Commander API, I think it could be done much, much more simply.

But, they should generally speaking work - it shouldn't be like 50%, it would be like ~80% successful, except for the Spin test that I disabled because it was failing ~80% of the time. Wait basically never failed. Backup/Drive on Heading was like 9/10 successful.

I see you have a couple in draft (#4471 #4472), is that why?

stevedanomodolor commented 6 days ago

@SteveMacenski, so I am not going crazy, does the wait, backup, drive on heading behavior tests fail once in a while. It does not seem to be stable.

You're not crazy, they're poorly written and it drives me bananas. I want to rewrite them at some point (#4349) to not try to fake everything out in some crazy way. I'm not really sure why this is the way that it is, I suspect it may have been written before we had any GZ simulations setup in the systems tests to work from. It is my plan to have a new contributor rewrite them as a project, but certainly if anyone else was interested in just deleting them and restarting now with the new GZ and the Simple Commander API, I think it could be done much, much more simply.

But, they should generally speaking work - it shouldn't be like 50%, it would be like ~80% successful, except for the Spin test that I disabled because it was failing ~80% of the time. Wait basically never failed. Backup/Drive on Heading was like 9/10 successful.

I see you have a couple in draft (#4471 #4472), is that why?

Yes thats why they are on draft. I couldn't figure out why they fail.

SteveMacenski commented 3 days ago

Working on the costmap filters right now and the file cleanup

ros-navigation / navigation2