09-05-2025, 04:37 AM
(This post was last modified: 09-05-2025, 04:39 AM by Apex.
Edit Reason: Additional Information
)
Hi All.
I am encountering errors with a UR5/CB3.1 where both MoveC and MoveL will overshoot their targets by a considerable margin. In the case of MoveC, moving between the following three sets of joint angles, which form a 30 degree arc, results in an overshoot of three or four degrees.
-104.342079, -105.270730, -113.423561, -135.392874, -29.432259, 136.049416
-110.914384, -100.749880, -119.571584, -135.593098, -50.957413, 138.275169
-120.712291, -99.309278, -121.514807, -135.727312, -75.725615, 139.589409
It is only when performing a subsequent MoveJ back to the last target that the robot ends up where it should be. Note that rounding is set to -1 (ie off) at the start of the Python program. Further, I am encountering similar issues with MoveL where the robot overshoots the end point by several millimetres - a clearly obvious amount.
I initially thought it might be caused by the way my Python program adds frames and targets to the RoboDK station tree before referring to the target items by name when performing movements. Saving the RoboDK file to HDD prior to running online appeared to solve the issues, but now it occurs despite doing this. Further, this overshooting issue is intermittent, despite running the exact same code with no modification. Some times the robot stops right on the target, and other times it overshoots.
At this point I am not sure where the issue lies - with RoboDK, PolyScope, or my code. Is this an issue anyone else has encountered with a UR5?
EDIT
The only thing that may be causing it is that all of my targets are Cartesian, and I usually force the robot into a specific pose using MoveJ before I perform a MoveL or MoveC.
I am encountering errors with a UR5/CB3.1 where both MoveC and MoveL will overshoot their targets by a considerable margin. In the case of MoveC, moving between the following three sets of joint angles, which form a 30 degree arc, results in an overshoot of three or four degrees.
-104.342079, -105.270730, -113.423561, -135.392874, -29.432259, 136.049416
-110.914384, -100.749880, -119.571584, -135.593098, -50.957413, 138.275169
-120.712291, -99.309278, -121.514807, -135.727312, -75.725615, 139.589409
It is only when performing a subsequent MoveJ back to the last target that the robot ends up where it should be. Note that rounding is set to -1 (ie off) at the start of the Python program. Further, I am encountering similar issues with MoveL where the robot overshoots the end point by several millimetres - a clearly obvious amount.
I initially thought it might be caused by the way my Python program adds frames and targets to the RoboDK station tree before referring to the target items by name when performing movements. Saving the RoboDK file to HDD prior to running online appeared to solve the issues, but now it occurs despite doing this. Further, this overshooting issue is intermittent, despite running the exact same code with no modification. Some times the robot stops right on the target, and other times it overshoots.
At this point I am not sure where the issue lies - with RoboDK, PolyScope, or my code. Is this an issue anyone else has encountered with a UR5?
EDIT
The only thing that may be causing it is that all of my targets are Cartesian, and I usually force the robot into a specific pose using MoveJ before I perform a MoveL or MoveC.
