A Physical Human-Robot Interaction Algorithm Based on Fuzzy Reinforcement Learning for Minimally Invasive Surgery Manipulator

In order to achieve the surgical gesture adjustment of the minimally invasive surgical robot, a variable admittance model based on fuzzy reinforcement learning for physical human-robot interaction is proposed. The manipulation characteristics of the operator are taken into account in the physical human-robot interaction process by an online learning method, which can adaptively modify the admittance model to respond to the operator's control intention. An experimental verification is carried out on a self-developed minimally invasive surgical robot, and the experiment results show that the pose adjustment of manipulator can be implemented naturally and smoothly by the variable admittance model based on fuzzy Sarsa(λ) learning. The proposed control strategy can meet the requirements of damping change in each stage of the physical human-robot interaction, and has high controllability and stability.