Enhancement of Manipulator Interactivity through Compliant Skin and Extended Kalman Filtering

In this paper we discuss control algorithms for real-time interaction between humans and robot manipulators sharing a common workspace. Unlike traditional robotic manipulators, we assume that the interaction between human and robot is not restricted to the wrist/end-effector of the robot, but that it can happen anywhere along the kinematic chain. Interaction forces are measured in some directions, and estimated in others via an extended Kalman filter. Sensory measurements used are traditional shaft encoders and also 1 dimensional force sensors via robotic "skin" placed on the robot links. We present simulation results with a CRS A465 showing the performance of our algorithms that compare the impedance response in the presence and absence of force measurements. We also show planned experimental validation on an actual robot using "Quickskin", a piezo-electric skin patch prototype in our lab.