Reflex-action position control of robots: comparison with impedance control

Abstract

A technique for control of one-arm manipulators, called reflex-action position (RAP) control, is developed. The reflex action makes the manipulator dynamics independent of the possibly time-varying mass and inertia of the load and other loading effects. This is accomplished by sensing the load with a wrist force/torque sensor and computing, using a transposed Jacobian matrix, joint torques which support or drive the load. With this form of load decoupling, the position controller can be designed for good performance while maintaining robustness. Simulation results demonstrate typical performance of the RAP controller. The well-known impedance control method was designed so that the end-effector of a manipulator would exhibit (to an external force) a mechanical impedance specified by the desired apparent mass, damping, and stiffness of the end-effector. These three matrices must be chosen carefully to ensure stability. The use of positive force feedback was discovered experimentally as the only means to stabilize an impedance-controlled manipulator in contact with a rigid surface.< <ETX xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">&gt;</ETX>