Neuro-adaptive Backstepping Control Of Robotic Arms With Rigid And Flexible Joints

Robotic systems are profoundly nonlinear coupled with uncertainties. Such multi-input multi-output systems, if considering measurement error, unmodelled dynamics, external disturbances, or even actuator failures, become rather complex. As modeling uncertainties and actuation failures are inevitable in practice, it is therefore is theoretical and practical significance to investigate the trajectory tracking control problem of robots considering the coexistence of those anomaly factors. The primary objective of this research is to investigate the trajectory tracking control problem of robotic arms with rigid and flexible joints. The main contributions of this work can be summarized as follows:(1) A novel RBF NN-based robust adaptive backstepping controller is proposed for the rigid robotic manipulators considering actuator dynamics. In contrast to the traditional backstepping method, the design procedure in our method is much simpler. Moreover, this method is able to deal with the known non-affine feature of the actuator.(2) To deal with nonlinear parameterized uncertainties and unknown non-affine feature of the actuator, a control method that blends RBF NN with robust adaptive control is proposed. Since NN unit is used to approximate the upper bound of unknown lumped nonlinearity instead of the uncertain vector itself, the resultant control scheme is much simpler in structure and less expensive in computation, as compared with most existing methods.(3) By taking into account the joint flexibility, actuator fault, actuator saturation simultaneously, a RBF NN-based DSC (dynamic surface control) fault-tolerant controller is proposed. In every step of the design, a first-order integral filter is introduced to calculate the derivative of virtual control, which avoids the expansion of the entry number caused by derivation of the virtual control in the process.Rigorous proof of the stability of the proposed control methods are provided based on Lyapunov stability theory. It is shown that with the proposed control scheme the closed-loop stability is always ensured and high tracking precision is achieved. A series of simulation studies are conducted that verify the effectiveness of the proposed control algorithms. It is worth mentioning that the developed control schemes, with suitable modification, can be applied to a larger class of nonlinear systems including robotic systems.