Nonlinear H∞ optimal control of reconfigurable robot manipulator via adaptive torque estimation
Reconfigurable Robot Manipulators (RRMs) have extensive applications in power grids, industrial 4.0 and flexible manufacturing, disaster rescue and exploration, as well as complex environments such as earthquakes and fires. This study examines the performance of reconfigurable robot manipulators operating in uncertain environments, addressing the challenge of mitigating interference noise in Harmonic Drive (HD) signal transmission. Furthermore, it presents a simplified robust Adaptive Dynamic Programming (ADP) framework to implement H∞ control for RRMs subject to unknown external disturbances. By representing the RRM dynamics as an integrated set of joint subsystem models, the corresponding control is formulated as zero-sum game, allowing a closed-form solution in robotic systems. The research results show that the ADP algorithm proposed in this paper is effective. Its contributions are primarily reflected in problem modeling, theoretical framework, and algorithmic implementation. The developed algorithm solves the HJI equation via a critic neural network, which facilitates direct adaptation of the H∞ control pair with assured convergence.