Inverse Kinematics for Barrett Hand

Typically: "How do I... ", "How can I... " questions
Post Reply
Posts: 10
Joined: 02 Feb 2016, 21:33

Inverse Kinematics for Barrett Hand

Post by Matt » 02 Apr 2016, 01:19

Hi there,

I have an application where I've collected many different 3-D positions of fingertip points from moving the Barrett Hand. I'm now working on the inverse problem, of moving the hand to the collected positions. The problem is kind of similar to the scenario in the tutorials: ... asics6.jpg

In the current scene, I've inverted the Barrett Hand so that fingerTip1 is defined to be the model base. I place this base at one of the given positions and orientations, and then perform simultaneous IK on the remaining 2 fingers (0 and 2), and use "inversePivot1" to control rotation of the hand from the initial contact spot. Here's an image/video of what it looks like: ... sp=sharing

For the IK, I have X,Y,Z, Gamma enabled, and I'm using the DLS solver. The problem is that the process very rarely succeeds - it often gets stuck like the poses in that folder. Is it just a matter of playing with the IK weights from here? I could always try expanding the search a bit more around the object, but this seemed like the simplest approach.


Site Admin
Posts: 7020
Joined: 14 Dec 2012, 00:25

Re: Inverse Kinematics for Barrett Hand

Post by coppelia » 04 Apr 2016, 07:51

Hello Matt,

your approach should work, but I would have done it by first placing the base of the gripper in a pose that minimizes the distance between the finger tips and the desired finger tip positions. Then apply simultaneous IK on the 3 fingers. But for that, it is important to have an initially open gripper. Another point to remember is that IK will not work properly if the tip/targets are too far away (since IK works by linearizing around current joint state).

Once you have the fingers on the desired points, you could keep IK on while trying to improve the position/orientation of the gripper base.


Post Reply