IV: Noticeable difference in gripping accuracy between MoveL and MoveJ
As I wanted to test the accuracy of a vision job I set up using the integrated vision of a Yumi robot, I mounted a pointing-tool to my robot and used a target with a marker.
To make sure my TCP is defined right, I aligned the tool with the world coordinate-system and rotated the wrist angle of the robot (to see if there is any change of the TCP while rotating-> there was none). I configured the quaternions of the robtarget for the grasping so that the tool is perpendicular to my gripping surface (which is parallel to the base).
While testing I noticed something strange:
If the robot approached the target with a MoveJ (using the offs command) and then moves to the target with a linear movement, the pointing tool sometimes missed the marker with an error greater than 1.5 mm, depending on the orientation of the target.
If it approached the target with a MoveL instead of a MoveJ, this error was much less.
AbsAcc is on. I also checked if something is wrong with the AbsAcc values in the MOC.cfg (and did a calibration using the calibration routine).
Shouldn't the results be comparable (within a range of 0.6 mm as this is the maximal error of absolute accuracy for the IRB14000), or am I missing something?
Thank you in advance!
Kris
To make sure my TCP is defined right, I aligned the tool with the world coordinate-system and rotated the wrist angle of the robot (to see if there is any change of the TCP while rotating-> there was none). I configured the quaternions of the robtarget for the grasping so that the tool is perpendicular to my gripping surface (which is parallel to the base).
While testing I noticed something strange:
If the robot approached the target with a MoveJ (using the offs command) and then moves to the target with a linear movement, the pointing tool sometimes missed the marker with an error greater than 1.5 mm, depending on the orientation of the target.
If it approached the target with a MoveL instead of a MoveJ, this error was much less.
AbsAcc is on. I also checked if something is wrong with the AbsAcc values in the MOC.cfg (and did a calibration using the calibration routine).
Shouldn't the results be comparable (within a range of 0.6 mm as this is the maximal error of absolute accuracy for the IRB14000), or am I missing something?
Thank you in advance!
Kris
0
Comments
-
Hi @unni2007
Is very strange that the same robtarget point to two different positions using the linear instead of joint movement;
Is the tool properly define (both the TCP and the load) and the robot arm right calibrate?
Probably the calibration of the vision system is not so accurate, try to move the target under the camera without rotating it and see if the robot follow it.
Hope this help.
0 -
Hello,Do you use fine zonedata?0
-
Maybe the mounted tool is not fastened and wobbles a bit. Does the same happen, when moving slowly?
0
Categories
- All Categories
- 5.5K RobotStudio
- 394 UpFeed
- 18 Tutorials
- 13 RobotApps
- 297 PowerPacs
- 405 RobotStudio S4
- 1.8K Developer Tools
- 249 ScreenMaker
- 2.7K Robot Controller
- 309 IRC5
- 59 OmniCore
- 7 RCS (Realistic Controller Simulation)
- 785 RAPID Programming
- AppStudio
- 3 RobotStudio AR Viewer
- 18 Wizard Easy Programming
- 105 Collaborative Robots
- 4 Job listings