Multivariable self-tuning control of redundant manipulators

An efficient self-tuning control design method for redundant robotic manipulators is presented. With the nominal torques compensating the nonlinear interactions among the joints, the complex nonlinear dynamic equations are linearized and a linear input-output perturbation difference model is developed. Based on which two self-tuning control schemes are proposed in joint and Cartesian spaces. The controller parameters are estimated by the recursive least squares identification algorithm, and the modification torques are generated by minimizing a generalized cost function. Some typical simulation results are presented to demonstrate the effectiveness of these self-tuning control schemes