Learning-Based Discrete Hysteresis Classifier Using Wire Tension and Compensator for Flexible Endoscopic Surgery Robots

Abstract

The tendon-sheath mechanism can be applied to a flexible endoscopic surgery robot because of its flexibility and power transmission. However, the hysteresis, which is the inherent problem with this mechanism, affects the precision of the control of the surgical robot. Despite several studies that are aimed at tackling hysteresis, only a few literatures consider a practical circumstance such as initial unknown hysteresis, proper surgical procedure, and camera illumination. In this study, we propose a novel framework to reduce the hysteresis of a flexible surgical robot using the learning-based hysteresis classification and a feed-forward compensation based on practical scenarios. We empirically discretize and divide the hysteresis class based on its size and show the correlation between hysteresis and time-series wire tension experimentally to study its potential for use in real surgical robots. The results indicate that the hysteresis can be classified by utilizing the time-series wire tension data. Moreover, the proposed compensator could enhance the performance of a real-size flexible endoscopic surgery robot based on actual surgical environment.