My IK simulation is unstable

Typically: "How do I... ", "How can I... " questions
Post Reply
lfself
Posts: 4
Joined: 28 Jun 2018, 04:37

My IK simulation is unstable

Post by lfself »

I follow the ABB IRB 360.ttt and Quattro.ttt model .The joints will break when I don't set dependent joint in simulation. If I add dependent joint, The link will tremble in simulation ,then I drag the ik_target, link will break and run in a mass,My model is https://pan.baidu.com/s/17nhXgSPXGMbXIef0GKtiLA, The password is 7w81.

coppelia
Site Admin
Posts: 10339
Joined: 14 Dec 2012, 00:25

Re: My IK simulation is unstable

Post by coppelia »

Your file host is super slow and always fails at downloading the file. Also it is impossible to understand the Chinese text for me. If possible, use something more robust/common..

Cheers

lfself
Posts: 4
Joined: 28 Jun 2018, 04:37

Re: My IK simulation is unstable

Post by lfself »

Hi,I reupload it from google drive. https://drive.google.com/open?id=162mgw ... D4gI58QZpa.Thank you for your replay.

coppelia
Site Admin
Posts: 10339
Joined: 14 Dec 2012, 00:25

Re: My IK simulation is unstable

Post by coppelia »

Do you want to control the robot in IK mode only, or also in FK mode? If IK mode only, the contstruction of the model is much more straight-forward...

Cheers

lfself
Posts: 4
Joined: 28 Jun 2018, 04:37

Re: My IK simulation is unstable

Post by lfself »

I want control it both in IK mode and FK mode. I follow the irb 360.ttt and quattro.ttt. In here , I want drag the ik_target to solve the IK problem.

coppelia
Site Admin
Posts: 10339
Joined: 14 Dec 2012, 00:25

Re: My IK simulation is unstable

Post by coppelia »

Ok, there are several things that appear wrong. Here a few suggestions:
  • Make sure you do not use any damped resolution method for your IK chains. If your construction is correct with enough degrees of freedom, then it should be stable with an undamped resolution method.
  • Try to go step-by-step. Typically, create first two IK branches: Branch A: starting at the base of arm 1, ending at the base of arm 2. Branch B: starting at the base of arm 1, ending at the base of arm 3. Once that is stable and looks right, add the 3 branches for the parallel rods of the secondary arms. Those can actually also end at the extremity of the primary arms. This gives you 5 IK elkements of the main task (inside one IK group)
  • Then you need to add a 6th IK element into that IK group for the case when you want to control the arm in IK mode. You'll have to programmatically enable/disable that 6th IK element to switch between IK and FK mode (you'll also have to switch the mode of the 3 base joints in that case: from Ik mode to passive mode.
  • In your scene, if you delete all IK groups except for the main, and set the resolution method to pseudo inverse, then it is stable. But you'll see some small initial movement due to the fact that you joints are not positioned/oriented precisely. Also, it is important to have the exact same pose for linked dummies.
  • My suggestion: make the FK mode first work. To that end, temporarily disable the last IK element. And set the 3 base joints into passive mode. Then you can test things with a sinusoid movement on them
  • ...
It is difficult to explain, and debug such a robot. It is actually much easier to start from scratch and constantly test the sub-steps.

Cheers

lfself
Posts: 4
Joined: 28 Jun 2018, 04:37

Re: My IK simulation is unstable

Post by lfself »

Thank you for your reply.I managed to make it stable in some place.I will work on.
Cheers.

Post Reply