Adaptive Critic based Redundancy Resolution scheme for Robot Manipulators

A novel adaptive critic based kinematic control scheme for a redundant manipulator has been proposed in this paper. The redundancy resolution has been formulated as a discrete-time optimal control problem. A Takagi-Sugeno (T-S) fuzzy based critic network is proposed to predict the global costate dynamics as fuzzy average of local costate dynamics. The integral cost function is used in the literature earlier, to achieve global optimum in contrast to instantaneous cost functions which gives local optimum. But both the approaches require the computation of pseudo-inverse which is computationally complex and suffers from numerical instability. In contrast, the proposed scheme does not require computation of pseudo-inverse of the Jacobian which makes the method computationally efficient. The proposed scheme is tested on 7 degree of freedom (7DOF) PowerCube manipulator from Amtec Robotics.