GMM-Based Heuristic Decision Framework for Safe Automated Laparoscope Control

Bin Li,Yiang Lu,Wei Chen,Bo Lu,Fangxun Zhong,Qi Dou,Yun-Hui Liu
DOI: https://doi.org/10.1109/lra.2024.3352308
IF: 5.2
2024-02-01
IEEE Robotics and Automation Letters
Abstract:Automated laparoscope field of view (FoV) control in minimal invasive surgery (MIS) poses challenges, as existing solutions failed to address dynamic surgical FoV requirements across different phases and they neglected the misorientation effect or potential obstacles during the control process which raised safety concerns. In this letter, we propose a Gaussian mixture model (GMM)-based heuristic decision framework that can achieve safe automatic laparoscope control to provide the phase-specific requirements for surgeons. Leveraging GMM to fit the domain knowledge of instrument distributions, we formulate a general nonlinear constraint optimal control model to optimize the laparoscope motion. To overcome the time-consuming issue of raw nonlinear optimization solvers, we first employ a decoupling approach for misorientation control under remote center of motion (RCM) constraints and propose a novel heuristic decision algorithm for predicting the optimal target and the future trajectory in real-time while ensuring effective collision avoidance. Furthermore, a unified primitive motion controller is developed accordingly for the FoV control which is applicable to mechanical/soft-programmed RCM surgical robotic systems. Extensive validations on the surgical dVRK platform and the general UR5 robotic laparoscope system demonstrate the feasibility, versatility, and superiority of our framework for safe automated laparoscope control, providing personalized views based on the surgeon's recorded clinical videos.
robotics
What problem does this paper attempt to address?