IK Solver failing

Typically: "How do I... ", "How can I... " questions
Post Reply
r_dhak
Posts: 9
Joined: 02 Jan 2019, 15:10

IK Solver failing

Post by r_dhak »

Hello,

So I have a custom manipulator. And I am trying to make the end-effector follow a defined path using Inverse Kinematics.

However the Ik solver is failing from the start, the end-effector is following the path however the end-effector is just lagging behind the target continuously.

Also I noticed when I used the simulation steppe,r that the first step is a sudden jump in the end-effector position.

I am just using the IK implementation from one of the examples of CoppeliaSim.
I tried keeping all the revolute joins cyclic as well (instead of defining a range), but in both cases the Ik solver fails.

I am attaching the scene https://1drv.ms/u/s!Ah8Cgzv3fsXqni2VJj5 ... b?e=B0uI9G

Thank You.

coppelia
Site Admin
Posts: 9049
Joined: 14 Dec 2012, 00:25

Re: IK Solver failing

Post by coppelia »

Hello,

actually the IK solver appears to be successful. Except for the error message. The reason is that your manipulator doesn't have the ability to place the end-effector in position and orientation (it is lacking additional degrees of freedom), you have however specified simIK.constraint_pose instead of simIK.constraint_position for the constraints.

Cheers

r_dhak
Posts: 9
Joined: 02 Jan 2019, 15:10

Re: IK Solver failing

Post by r_dhak »

Yes that was the mistake that I only wanted a 3DOF but was giving 6DOF as constraint.
Thank you. That solved the warning problem.

However whats with the lagging?
When the simulation time-step is higher the lag is more, which is understandable i.e. the pseusoinverse approximation is not accurate. Still in order to keep the lag negligible I have to set the simulation time-step to 1ms. Any suggestion to resolve it?

Also the sudden jump after 1st step.
The path following should start from path control point 0, which is where i am keeping my end-effector as the initial state. So there should not be any problem.

Could you check it with the simulation stepper and in the 1st step you will notice the jump.

coppelia
Site Admin
Posts: 9049
Joined: 14 Dec 2012, 00:25

Re: IK Solver failing

Post by coppelia »

this is because your manipulator is dynamically enabled, and each joint has a motor with controller: those can only accelerate with the given force/torque. To verify, simply set the whole manipulator model as Model is not dynamic in the model dialog and you'll see the difference.

Cheers

Post Reply