HUMAN-ROBOT INTERACTION OF A CRANIOTOMY ROBOT BASED ON FUZZY MODEL REFERENCE LEARNING CONTROL.

In this paper, we design a variable admittance controller and propose a variable admittance human-robot cooperative control method based on fuzzy model reference learning. The method is intended to improve the flexible adaptive capability of the robot to assist the surgeon in accomplishing different stages of the task during a craniotomy. First, the method establishes the autoregressive integrated moving average-Kalman filtering-blood pressure (ARIMA-Kalman-BP) model for the drag force prediction by taking the features of natural human arm motion as the reference model of fuzzy learning control, which solves the problem of the features of natural human arm motion being difficult to model. Then the tuning parameter rules for variable virtual damping and virtual mass of the fuzzy conductivity controller are trained by the learning mechanism. Subsequently, the variable conductivity control method based on the tuning of virtual damping and virtual mass parameters is developed by using the robot acceleration and the robot velocity as inputs, and the robot desired velocity and desired acceleration as outputs. The experimental results show that the method can meet the requirement of flexibility; the maximum error of human-machine cooperative velocity is 0.0014 m/s, and the maximum error of human-machine cooperative acceleration is lower than 0.0021 m/s2. Compared with the fuzzy control based on the variable admittance parameter alone, this method has better tracking velocity and acceleration. [ABSTRACT FROM AUTHOR]