Global Neural Learning Finite-time Control of Robot Manipulators with Guaranteed Transient Tracking Performance

A global neural learning tracking control for robot manipulators to guarantee the transient performance and finite time convergence is investigated. The paper develops some auxiliary filtered variables and a nonsingular terminal slide mode (NTSM) surface to guarantee the tracking error converging to zero with finite time. An error transformation function is also introduced to deduce the control law, which ensure that the transient tracking error never violate the predefined boundedness. Furthermore, under the persistent excitation (PE) condition, the weights of neural networks (NNs) will converge to optimal values with finite time, which can be reused to decrease the computational cost. Compared experimental results are carried out to verify the superior performance of our controller.