Event-triggered-based Decentralized Optimal Control of Modular Robot Manipulators Using RNN Identifier

In this paper, an event-triggered-based decentralized tracking control method is proposed for modular robot manipulators (MRMs) using a recurrent neural network (RNN) and neuro-dynamic programming (NDP). The joint torque feedback (JTF) technique is introduced to model the MRM subsystems. The cost function of each subsystem consists of a tracking error fusion function and a term summarizing the RNN identifier errors. The event-triggered Hamiltonian-Jacobi-Bellman (ETHJB) equation is solved by constructing a critic neural network using NDP, and a decentralized optimal tracking control policy under the event-triggered framework can be obtained. The closed-loop MRM system is shown to be uniformly ultimately bounded under the Lyapunov stability theorem. Finally, the experimental results verify that the proposed control method is superior to the time-triggered optimal control policy and the observer-critic-based event-triggered optimal control policy proposed in the previous work of the author.