Box falling through belt
Since the upgrade to Gazebo 8.4, I've been seeing issues with the box spontaneously falling through the conveyor belt. It's not common, maybe one in fifteen runs, but it has happened multiple times. Is it safe to assume that if this happens during the qualification trial, we'll get a second chance?
Comments (14)
-
-
reporter I haven't actually seen it live because of how infrequently it happens, but it is consistent with behavior reported in my text logs. Not long after this, I start the belt running at full power, and I can tell from other sensors that the box is not moving with it.
-
Ok, thanks. There are some artefacts that get introduced with log playback (boxes falling slowly) but this sounds more like a simulation bug than a playback artefact. I'll take a closer looks and get back to you with an answer to your question tomorrow.
-
reporter Here's a clip of the box falling straight through without even having any parts in it. You can tell that the belt is moving and the box is not because it continues to dispense the rest of the boxes. https://www.youtube.com/watch?v=kO3Xjxm_gIc
-
Thank you for the additional clip; if there's anything in particular that you find makes it more likely to happen that will help us look into it faster.
In the context of qual1 I can confirm that if this occurs during an automated trial run of a participant's system it will warrant a rerun.
-
- changed status to resolved
@joeygannon I think this was caused by "bad luck" on the physics side of the simulation. The shipping box model has been altered in https://bitbucket.org/osrf/ariac/pull-requests/94/ to hopefully make this less likely to occur. I haven't been able to reproduce the issue even with the original shipping box model, so it is difficult to be certain that this has been fixed, but I am going to close this issue optimistically and we can reopen it if you/others still encounter it.
-
reporter - changed status to open
This is still happening on 2.1.4
-
OK, thanks for re-opening.
-
- marked as minor
-
FYI 2.1.7 included another hopeful fix for this issue.
-
reporter Any idea when the new version will go into the dockerhub image? Looks like it's still stuck on 2.1.6.
-
thanks for pointing this out. We'll push it up within the next hour.
-
reporter - changed status to resolved
I've run enough tests now to feel confident that this is fixed. Thank you for your diligence on this issue!
-
Good to hear! Thanks for reporting back
- Log in to comment
Thanks for the report. Have you ever seen this happen while the simulation is running, or only ever in log playback?