Adaptive control of robot manipulators via velocity estimated feedback

An approach for designing robust controllers via state-space feedback is presented. The robot model parameters are assumed to be unknown and velocity measurements are assumed to be unavailable. State observation and parameter adaptation are performed simultaneously. The adaptation law, the observer gains and the control law are designed on the reduced-order manifold which results from the invariance of the switching surface. With respect to the authors' previous work, the asymptotic stability of the closed-loop system resulting from the approach presented is not conditioned by a particular type of inertia matrix variations. On the other hand, the introduction of an adaptation loop may also be motivated by a reduction of chattering at control law level since the control law only depends on the estimated state and parameter vectors, and hence contains no terms proportional to discontinuities.<<ETX>>