Help with Inverse Kinematics

Typically: "How do I... ", "How can I... " questions
Post Reply
stephen127
Posts: 6
Joined: 24 Sep 2016, 18:06

Help with Inverse Kinematics

Post by stephen127 »

Hello, I am very new to V-rep.

I followed the tutorial to build a clean model for an ABB_IRB_460 4 DOF robot arm. This model has some extra parallel linkages to support axis 3 and 4.

After I built the model, I was trying to follow the tutorial for inverse kinematics to set the robot tip to follow a target dummy. However, for some reason, my simulation did not work. The robot totally ignored the target dummy and pretended there was no such thing. It would just randomly moving around during the simulation.

I was trying to attached my simulation here, but did not see this option. Maybe I didn't build the joint hierarchy correctly so that the IK couldn't be solved. But I didn't get any errors.

Could anyone help me with it? Thank you so much!

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

Re: Help with Inverse Kinematics

Post by coppelia »

Hello,

you can share a scene via dropbox or similar. At the same time, have a look at the demo scene scenes/forwardAndInverseKinematics1.ttt, which illustrates a similar robot, once in FK and once in IK mode.

Cheers

stephen127
Posts: 6
Joined: 24 Sep 2016, 18:06

Re: Help with Inverse Kinematics

Post by stephen127 »

Hello,

Here is the link to my scene,

https://drive.google.com/file/d/0B0hmDx ... sp=sharing

I did notice some difference in the example scene and my settings. All the shapes' dynamic properties are not checked in the example. However, when I followed the tutorial of "Building a clean model tutorial", it mentioned that the body should be configured as dynamic and respondable. So now I am confused about which one I should follow now.

Also, I followed the example to change my settings and the robot did not fall apart during the simulation. But it still does not follow the target dummy when I moved the dummy around.



coppelia wrote:Hello,

you can share a scene via dropbox or similar. At the same time, have a look at the demo scene scenes/forwardAndInverseKinematics1.ttt, which illustrates a similar robot, once in FK and once in IK mode.

Cheers

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

Re: Help with Inverse Kinematics

Post by coppelia »

Building a purely kinematic model will be easier. So you do not need to turn every shape into a dynamic and respondable shape. Just leave the shapes as static and non-respondable.

Then, try to switch for your IK group to a damped resolution method. You will see that the end-effector will follow the target. If such a strange behaviour occurs (i.e. the pseudo inverse method does not work, but the damped method works), this usually points to an over-constrained IK group. You will have to carefully analyze which IK elements need to be solved simultaneously, and which can be sequentially solved.

One obvious problem in your case is that the robot end-effector does not have 6 degrees of freedom, yet, you set the IK group element's constraints as if it had 6 degrees of freedom.
Your robot is kinematically very very similar to the robots in the scene I pointed out, so you should be able to figure out.

Cheers

stephen127
Posts: 6
Joined: 24 Sep 2016, 18:06

Re: Help with Inverse Kinematics

Post by stephen127 »

Thank you for the clarification! Now I understand the difference.

I have managed to make the simulation work. Thank you for the help! :)
coppelia wrote:Building a purely kinematic model will be easier. So you do not need to turn every shape into a dynamic and respondable shape. Just leave the shapes as static and non-respondable.

Then, try to switch for your IK group to a damped resolution method. You will see that the end-effector will follow the target. If such a strange behaviour occurs (i.e. the pseudo inverse method does not work, but the damped method works), this usually points to an over-constrained IK group. You will have to carefully analyze which IK elements need to be solved simultaneously, and which can be sequentially solved.

One obvious problem in your case is that the robot end-effector does not have 6 degrees of freedom, yet, you set the IK group element's constraints as if it had 6 degrees of freedom.
Your robot is kinematically very very similar to the robots in the scene I pointed out, so you should be able to figure out.

Cheers

Post Reply