Author Topic: Multiple set of {A,B,C} for same orientation and position  (Read 3322 times)

0 Members and 1 Guest are viewing this topic.

Offline ravi.joshi53

  • Jr. Member
  • **
  • Thank You
  • -Given: 2
  • -Receive: 0
  • Posts: 62
Multiple set of {A,B,C} for same orientation and position
« on: January 27, 2015, 05:25:20 PM »
Hi,
I have noticed that sometimes KUKA shows multiple set of {A,B,C} for same orientation and position. Below is an example-

I kept KUKA in A=0, B=90, C=90 orientation and just created a point P1 from teach pendent. This point is the first PTP run (BCO Run). I noticed that even without any change in position and orientation (no motion in end-effector), KUKA tool A,B,C values keep on changing by releasing and pressing the motor drive button in the teach pendent.

I am attaching all these values in a screenshot of KUKA teach pendent. Please have a look.

Also somebody please explain the reason behind this variations and how to handle these situations, since I am trying to control KUKA from RSI in this orientation.


Offline SkyeFire

  • Global Moderator
  • Hero Member
  • *****
  • Thank You
  • -Given: 32
  • -Receive: 947
  • Posts: 6429
Re: Multiple set of {A,B,C} for same orientation and position
« Reply #1 on: January 27, 2015, 11:28:22 PM »
  Certain positions that are combinations of 0deg, 90deg, and/or 180deg, are essentially "singularities" (in the mathematical sense).  Due to the nature of Euler engles, robot orientations that are close to these "singularities" will show very large swings in the ABC values even if their actual position change in physical space is very small.  What you're seeing is essentially the "noise" of the robot position, amplified by your proximity to an Euler singularity.  This is similar to how, if you program a LIN motion with Axis 5 too close to 0, A4 and A6 begin rotating faster and faster (and eventually exceed their speed/accel limits).  But fortunately for us, this Euler singularity does not (normally) produce such fatal motion issues.


Offline ravi.joshi53

  • Jr. Member
  • **
  • Thank You
  • -Given: 2
  • -Receive: 0
  • Posts: 62
Re: Multiple set of {A,B,C} for same orientation and position
« Reply #2 on: January 28, 2015, 03:26:00 PM »
SkyeFire,

Thank you so much for this wonderful explanation. I am new to rotation matrix and Euler Angles. I have the knowledge about them, but I am not aware of their implementation in KUKA robot. Can you please provide some material (or some brief explanation), showing the implementation in KUKA. If possible, please provide some KRL code, so that It can be verified by running in robot.

-
Thanks

Offline Fubini

  • Hero Member
  • *****
  • Thank You
  • -Given: 4
  • -Receive: 169
  • Posts: 559
Re: Multiple set of {A,B,C} for same orientation and position
« Reply #3 on: January 29, 2015, 06:55:42 AM »
Hi,

KUKA uses Z-Y-X EulerAngles (A,B,C). It is exactly the same as the often so called roll-pitch-yaw (RPY) convention. See

http://en.wikipedia.org/wiki/Euler_angles

or

http://www.robot-forum.com/robotforum/kuka-robot-forum/abc-3-point-base-measurement-how-does-algorithm-works/msg67183/#msg67183

or

http://www.robot-forum.com/robotforum/abb-robot-forum/orientation-questions/msg21927/#msg21927

...

Basically using the forum search with keyword "Euler" shows many different explanations and examples.

Fubini
« Last Edit: January 29, 2015, 07:03:38 AM by Fubini »