Have a UR10E with robotiq Hand-E and force/torque sensor. I am using a stationary vertical shaft router with a 1/2" diameter sanding drum and teaching the path with a finished product and a printed reference wheel to simulate the sanding drum with a .010" undersize as the “teach tool”. In active drive mode the robot bounces, pulls away, when it contacts the wheel and when I resist that movement it rebounds into the wheel and starts an oscillation. If I let go of it it drifts off an inch or so and stops. I zero the active drive before starting to record the path.
Any idea of what is the problem?
Can you please tell me which version of copilot and polyscope are you running?
Be sure you define the payload and center of mass correctly prior to using the active drive.
There are 2 different methods of movement with the active drive. The rabbit and snail.
The rabbit is intended for smooth movement while the robot isn’t touching anything. If you need to teach a path with a tool and need contact between them, then the snail is suggested. It will remove the bounce effect that occurs when the tool comes in contact with a surface.
I assume you are running an older version of the URcap as with the new version, it automatically switches to the snail/contact mode as soon as a contact is detected with a surface to avoid the bounce.
Finishing Copilot 3.0, not sure of the polyscope. The robot is new with an incept date of March 2020. I had noticed the bounce is worse in rabbit but snail still has an oscillation I can feel when programming. It also gently drifts away as soon as you select active drive. This makes one person programming difficult as the total path is broken into several shorter paths to make “editing” easier as the entire path doesn’t need to be redone. The drift makes unnecessary movement in the final complete path.
I have another post regarding force required to move the robot in teach mode. Can it be reduced? The part is small and flexible and the force to move the robot exceeds the force to flex the part and reduces the tactile response in following the part edge along the tool. Then with a live tool the part gets gouged due to the path being taught by a flexed part.
We are at version 1.14.5 of Copilot
I would start by updating polyscope to the latest version as well as the copilot.
The bounce effect has been improved.
Also regarding the drift, this happens as the force sensor feels a force is applied to the tool. You can press that button to reset the forces to 0 while nothing is touching the tool.
It will immediately stop drifting.
Unfortunately, it is not possible to reduce the needed force to move the robot. To help with the teaching process, we usually recommend having stiff contact with the part. Since your part is flexible, this makes it more difficult to teach. Would you have the possibility to 3D print the part so you could teach it on a stiff surface?
I was looking at the wrong thing.
UR Caps v 126.96.36.19976
Polyscope v 188.8.131.5253
I usually zero then record the 7-10 paths around the part. Do I need to zero each time before I start recording a path?
We can probably print the part with some gusseting to stiffen it up. I’ll look into that.
Thank you for the help.