Cooperation of robot manipulators with motion constraint by real-time RNN-based finite-time fault-tolerant control

Abstract

The nonlinear cooperative dynamics of m robots with motion constraint and faulty input saturation is addressed. To the best of our knowledge, the motion constraint for the cooperative control of multiple robots is first discussed as one of the designed concepts such that the proposed cooperative control system is more practical and its performance is also improved. Together with a finite-time concept, a real-time recurrent-neural-network-based finite-time fault-tolerant cooperation constraint control (RT-RNN-FTFTCCC) is designed to tackle the aggregately dynamic uncertainties brought about by motion constraint error, cooperation uncertain dynamics, disturbance torque, faulty input saturation, unsatisfying skew-symmetric matrix condition, and others. Moreover, the stabilized learning law includes two terms to cover all the range of filtering cooperation errors such that learning weights converge even in uncertain circumstances. Finally, the stability analysis of the overall cooperation control system is achieved by Lyapunov stability theory. In comparison to adaptive compensation control, recurrent-neural-network-based control without the motion constraint compensation, and different stabilized learning laws, the cooperative control of bi-robotic arms of 3-link in-plane with new constraint validates the improvement and robustness of the proposed control.