I’d like to introduce myself as Adithya Murali from UC Berkeley Robot Learning/Automation Sciences Lab. We’ve been heavily using OpenRave IKFast for the Raven II and PR2 and we’re attempting it on the da Vinci Research Kit (dVRK) . I’d be grateful for clarifications regarding our questions on IK Fast for the two slave arms. I’ve attached the collada, xml, URDF file and the python script used for IK. I could send a collada file with mesh information if needed.
The dVRK has a pair of identical 7 DOF arms (the left, right arms referred to as prefix ‘’one”, “two” ) which contain some mimic joints for pitch between the outer yaw and insertion joint. Since IKFast doesn’t solve for mimic joints, we separated them from the main TF tree (as seen in the "modified" files and the attached schematic of the frames in the tree). After the modification, an IK model was autogenerated using the code below. IKFast succeeded when we gave the current end-effector pose as input but the joint angles don’t look right. If we added a translational offset of greater than 1 mm to that pose, an IK solution was not found. I was wondering what is a good way to debug IKFast solutions.