Decentralized Fault Tolerant Control of Modular Manipulators System Based on Adaptive Dynamic Programming

This paper presented an original decentralized fault tolerant control approach for modular manipulators which based on adaptive dynamic programming (ADP) algorithm. First, the dynamic model of modular manipulators is established via joint torque feedback technique. Then, the fault tolerant controller is designed which composes of model-based compensation controller, observer-based fault tolerant controller and ADP-based optimal controller. According to ADP algorithm, the Hamiltonian-Jacobi-Bellman (HJB) equation can be tackled by critic neural network (NN). The closed-loop modular manipulators system is guaranteed asymptotic stable based on Lyapunov theory. Experiments are performed to verify the proposed method, and the results have guaranteed its effectiveness.