A Cerebellum-Inspired Model-Free Kinematic Control Method with RCM Constraint

Abstract

Minimally invasive surgery requires to reduce trauma to patients during the movement of the inserted robot end-effector. The cerebellum is able to control limbs in many scenarios, with high precision and robustness. This article designs a cerebellum-inspired model-free scheme for the tracking control of redundant robot manipulators with remote center of motion (RCM) constraint. The scheme is formed by coupling liquid state machines (LSM) and zeroing neural network (ZNN). The ZNN is able to generate approximate joint angle commands as teaching signals without a perfect robot model to train the cerebellum model based on LSM. The output of the LSM is used as the control commands for current moment, which includes managing the constraint on RCM. Finally, demonstration simulations and experiments are conducted to verify the efficacy of the proposed control strategy.