[Other issue]: RMF sending multiple identical goal with different orientation

[DoppiaEffe94]

Before proceeding, is there an existing issue or discussion for this?

• I have done a search for similar issues and discussions.

Description

Hi All!

First of all I would like to congratulate you for the huge and amazing job you have done so far and what are you still doing!

I'm trying to solve an issue with RMF for my robotics project with multiple robots.
In this scenario I have a couple of huskies which are managed by a free fleet system and each one of them has a full control fleet adapter integrated.

Now, everything works correctly until, in the middle of simulation, the robot seems to receive multiple goals with the same x and y, but different yaw. The same behavior happens, not only in a waypoint, but in the middle of a lane.
See below.

It seems quite strange here since in the moment these goals are sent there's no conflict to solve and the robot could simply follow the other goals.

I tried to look for similar issues around the discussions, but, a part from the #429 (which didn't solve my problem because ResponsiveWait is set to False), there are no discussions at all.

At this point a few questions comes out:

• Who can be the responsible of this behavior?
• How can i solve this issue?

Required information:

• Operating system and version: Ubuntu 22.04.
• ROS distribution and version: Humble.
• OpenRMF version branch: humble.
• Fleet management system: Free fleet.
• Fleet adapter: Full control fleet adapter.

Expected behavior:

The robot should remain still in whatever situation (conflicts or not).

Thank you for a possible answer!

[naveenrobo]

@mxgrey we are also getting same issue. is there a way to fix this?

[mxgrey]

I haven't seen this kind of spinning happen mid-travel.

Are you able to provide a minimum reproducible simulation that I can run on my end? Otherwise it's going to be very tricky to identify the cause since there are so many layers where a problem could be happening.

[naveenrobo]

It's not happening during mid-travel. It happens when responsive wait is enabled and robot reaches a goal then we get same goal again with different orientation. We will try to create a sim that can reproduce the issue.

[mxgrey]

It happens when responsive wait is enabled

Do you mean disabled?

If responsive wait is enabled then that's a known behavior, and different from what the original poster in this thread is talking about. With responsive wait turned on, Open-RMF will periodically verify that the robot is on the last waypoint that it was sent to. If you report the robot's state in a way that suggests it's not entirely on that point, then you'll periodically receive a plan that asks the robot to go to that point. You have three options:

1. Make sure you're reporting that the robot is completely on the point
2. Ignore the commands if you believe the robot is already on the point
3. Turn off the responsive wait behavior

In general I recommend using (3) plus giving your fleet a "finishing task" to have the robots go park or charge when they're done with their tasks.