ADAPTIVE DEAD-BEAT CONTROL LAW FOR TRAJECTORY TRACKING OF ROBOTIC MANIPULATORS

The purpose of this paper is to propose an adaptive dead-beat controll er for the trajectory tracking of a robotic manipulator. The dead-beat compensator is self-tuned to a linearized discretized model whose par ameters are identified on-line through a Kalman-like estimator. To imp rove the convergence of the estimator and to obtain good control perfo rmances even in the case of time-varying parameters, the state covaria nce matrix of the Kalman filter is adapted to the observed statistics of the innovation process. Numerical results have been obtained in a s imulation context and refer to various operating conditions. They show that very good control performances in terms of maximum error are rea lly obtainable. A comparison with minimum variance control is also rep orted.