Intelligent adaptive control of robotic manipulators based on a multiestimation scheme

A multiestimation-based adaptive controller is designed for robotic manipulators. Multiestimation based techniques can improve the transient response of adaptive systems by the appropriate switching between the various estimation schemes running in parallel. The estimated parameter vector which parameterizes the adaptive controller is generated by a convex linear combination of the individual estimated vectors from multiestimation scheme. The switching process, which is subject to a minimum dwell or residence time in order to guarantee stability, is determined by the updating of the values of the coefficients of such a linear combination. The robot is controlled by using an impedance type control.