Torque Estimation-Based Decentralized Robust Control of Modular and Reconfigurable Manipulators with Kalman Filter via Generalized Momentum

Aiming at the joint torque estimation problems of the sensorless modular reconfigurable manipulators (MRMs), a torque estimation method with external wrench is proposed, with the basis of the generalized momentum-based Kalman filter. In the case of measuring only the motor torque, joint angle and joint velocity, the dynamic equation of the MRMs is modified by the generalized momentum, and the disturbance observer is impletmented as a classic Kalman filter. The proposed method not only reduces the measurement noise induced by the measurement of joint acceleration, but also systematically considers the uncertainties in the subsystem model. On account of the Lyapunov theory, the closed-loop manipulator system is demonstrated to be asymptotic stable with the decentralized robust controller. Ultimately, the avalibility and feasibility of the proposed approach is demonstrated by experiments.