search

ros-industrial / motoman

ROS-Industrial Motoman support (http://wiki.ros.org/motoman)
143 stars 194 forks source link

Fix for "within range" error #536

Closed qweewqrty closed 1 year ago

qweewqrty commented 1 year ago

While using Motoman GP12 with ROS I regularly got problem: "Trajectory start position doesn't match current robot position". I found, that, after complete moveit trajectory, robot position have small fluctuation (within 0.005 rad) for couple of seconds. If moveit generate plan while position changing - error occured. I checked motoman code, that generated error message, and found that it belong to FS100 controller. But my GP12 had YRC1000. I disabled that check and my robot worked fine for hours. Seems like this check really needed for FS100, not YRC1000. I wrote fix for that [roblem in this pull request.

gavanderhoorn commented 1 year ago

Thanks for the PR. That's always appreciated.

However:

I checked motoman code, that generated error message, and found that it belong to FS100 controller

could you clarify what made you conclude it "belong[s] to FS100 controller"?

ted-miller commented 1 year ago

I checked motoman code, that generated error message, and found that it belong to FS100 controller.

The comment in the code is misleading. This check should be performed for all controller models.

I recommend monitoring the in_motion flag of the RobotStatus message. This should indicate whether it is safe for MoveIt to plan the next trajectory

qweewqrty commented 1 year ago

could you clarify what made you conclude it "belong[s] to FS100 controller"?

Comment in the code said, that this check needed by FS100 controller.

qweewqrty commented 1 year ago

I recommend monitoring the in_motion flag of the RobotStatus message. This should indicate whether it is safe for MoveIt to plan the next trajectory

Thanks, I'll check this way ASAP, and, if problem will persist, return here =)

gavanderhoorn commented 1 year ago

could you clarify what made you conclude it "belong[s] to FS100 controller"?

Comment in the code said, that this check needed by FS100 controller.

Please always link to code or text when you make references to it. That makes our lives much easier. Otherwise we have to search.

EricMarcil commented 1 year ago

Just to clarify why that code is there. The MotoRos driver moves the robot by increments, so if you start moving the robot at the wrong place, it will end-up at the wrong place. Without that start check, if you grabbed the wrong robot position (for example while the robot is still moving), plan a trajectory, and send it to the controller, then the robot could potentially move to an unexpected position. We've had a few cases where we increase a bit the tolerence on the start motion, but do NOT completely remove it.

qweewqrty commented 1 year ago

could you clarify what made you conclude it "belong[s] to FS100 controller"?

Comment in the code said, that this check needed by FS100 controller.

Please always link to code or text when you make references to it. That makes our lives much easier. Otherwise we have to search.

Will do!

qweewqrty commented 1 year ago

It felt like changing tolerance didn't do anything. I set it up to 10 rad and still got an "within range" error. And when I thought, that check didn't important for my controller - that had some kind of sense. Now I think, I'll do some more debugging, to understand, what's going on.

gavanderhoorn commented 1 year ago

It felt like changing tolerance didn't do anything. I set it up to 10 rad and still got an "within range" error. And when I thought, that check didn't important for my controller - that had some kind of sense. Now I think, I'll do some more debugging, to understand, what's going on.

are you saying you always get the error "Trajectory start position doesn't match current robot position"?

If that's the case, that would be something to diagnose.

But the changes proposed in your PR are not a correct solution.

If you feel you are getting that error while "everything should just work", please open an issue instead, so we can try and find out what's going on with your system.

gavanderhoorn commented 1 year ago

With that said, I'm going to close the PR for now.

To diagnose any problems with the traj-start-pos checking logic, please open an issue.

If it turns out we do need to fix something, we could re-open this PR.

@qweewqrty: in any case, thank you for submitting the PR. That is always appreciated.