custom cost function different solutions in pick-ik vs bio-ik

[buckleytoby]

Hey All,

I'm testing using only a cost function (with squared distance error to a target position), and when I run setFromIK with bio-ik I get the expected behavior; my robot state converges upon my desired target.

However, when running the same code with pick-ik active the joint_diff value grows quite large, causing my joints to saturate at their joint limits and stay there indefinitely. The result is a solution which never converges to my target.

A potential source of the issue is the linear step size estimation. There's no way to limit the joint_diff value. Can there be an option for a static joint_diff value?

Thanks