Predictive control application in arm manipulator coordination

Abstract

The coordination problem of a human arm and a robot manipulator is explored. The problem has recently been formulated to facilitate the use of manipulators in unstructured environments. This paper addresses the problem in the framework of predictive control techniques. An impedence model of the arm and a position-controlled manipulator are used. The transfer function from the manipulator input (position) to the arm output (force) is determined. The desired future set point for the arm force is set to zero. The predictive controller then computes the required control inputs by minimizing a cost function of future errors between estimated and desired set points. The control strategy allows the manipulator to take over the load from the arm such that the arm force gradually reduces to zero. The simulation results indicate that manipulator becomes highly responsive to movements initiated by the human arm.