Kuka LBR iiwa Inverse Kinematics

I'm not sure there is a simple solution. A normal 6-axis robot can, worst case, have something like 16-32 different joint solutions for a single Cartesian location, and either calculates the least-effort disambiguation on the fly, or contains additional disambiguation data as part of the programmed location.

The issue with robots like the iiWA is that adding the 7th axis mid-arm creates a near-infinitude of valid joint poses for any given Cartesian coordinate set. Typically the iiWA resolves this by treating the mid-arm axis as separately controllable, rather like an integrated external axis. So the mid-arm axis position is explicitly controlled, and the IK solution is generated the same as for a 6-axis arm, just with the arm "shape" altered to match the programmed angle of the mid-arm axis.

When you get into some of the more unique abilities of the iiWA (like "bending" the arm in response to external forces without altering the TCP position/path, I'm not privy to the details, but I suspect the kinematics are resolved by limiting the solution space in realtime to poses very close to the current pose, and iterating this every interpolation cycle, rather similar to how a Linear motion is made up of many small Joint motions calculated on the fly.

Hello @Werther, I'm facing the same problem. I've tried different approaches but none of them is the right one.

Did you find a solution?

Thanks!

I had a closer look to the document mentioned by Werther.

In order to do the inverse kinematics the forward kinematics comes first. In this document the forward kinematics is explained very well.

The dh-parameters are given and the dh-transformation as well (page 3).

The dh-transformation (38) does not match the axes angles (37).

RoboDK is a good visualitation tool (selected robot KUKA LBR iiwa 7 R800)

If you want to follow I would be very happy otherwise I will do for myself