Stereovision based force estimation With stiffness mapping in roboticassisted Surgical tool insertion Using recurrent neural network

Abstract

Robotic-assisted surgery has great potential to transform the existing newlineconventional surgical practice, which offers many benefits as well as newlineintroduces innovation in robot-enhanced approaches that extend the newlineproficiency of the surgeon. Surgical robots are considered man-machine type newlinecollaborative robots that reproduce various surgical procedures based on the newlinecommand given by the surgeon and pre and intraoperative data to the surgeon newlinewithout any direct physical interaction with the patient. However, the field is newlinestill a nightmare for the medical community due to its nature of complexity. newlineThe lack of realistic force feedback in surgical robots is still an open challenge newlineto the research community, which further impedes the sophisticated use of newlinesuch robots.This research work proposes a novel method for estimating the newlinereaction force, the Stereovision-Based Force Estimation method (SBFEM), newlinewith deep learning techniques to predict the interaction force produced in newlinedifferent skin layers during the performance of various surgical procedures. newlineThe interaction force is estimated through SBFEM combined with computer newlinevision and neural networks instead of using direct force sensors due to the newlinedifficulty of adapting them to tools due to biocompatibility, sterilizability, and newlineintegration issues. The design of the force estimation model is intuitively newlineguided by the structure of the input and output data to be processed. The newlineproposed model processes both spatial and temporal information acquired newlinefrom the vision and tool data. The video sequence contains a spatiotemporal newlinestructure, whereas the interaction force and tool data contain a sequence of newlinetemporal information. newline newline