Shared‐control design for robot‐assisted surgery using a skill assessment approach

Abstract

Abstract In this article, a shared‐control system with skill‐based share weight allocation is proposed for a robot‐assisted minimally invasive surgery (MIS) procedure. A convolution neural network (CNN) is trained for online skill assessment, and the result is used to generate the share weights of robot autonomy and the user remote control. The control system can ensure synchronization of the two commands from the surgeon and robot autonomy and combine them to determine the motion of the surgical instrument. In this work, a contour‐tracking task is handled by the suggested shared controller to simulate a surgical cutting operation. Experimental results on a lab‐built robotic platform are presented to show the effectiveness of the proposed method. Multiple contour‐tracking experiments have been tested to compare the tracking performances of pure manual remote control and the proposed shared‐control method. Experimental results show that the shared controller achieved 34.5% improvement in tracking accuracy in comparison with pure manual control.