燕山大学学报2024，Vol.48Issue(3)：215-222,8.DOI:10.3969/j.issn.1007-791X.2024.03.003
基于强化学习的机器人眼科手术适应性虚拟夹具设计
Design of adaptive virtual fixture for robot-assisted ophthalmic surgery based on reinforcement learning
摘要
Abstract
Virtual fixture plays a significant role in enhancing manipulation performance and safety for man-machine collaborate surgery.However,the virtual fixture is pre-fixed and hand-designed,unable to adapt to the operator-specified skills.Based on the learning of surgical data,a general framework for virtual fixture control of robot-assisted ophthalmic surgery based on reinforcement learning policy training method is proposed to assist doctors to complete surgical tasks.The virtual fixture enforcement module is redesigned to reinforces compliant impedance rather than a fixed one and generates dynamic Cartesian force feedback on the client side.The skill learning module is carried out the virtual fixture adjusting and configuration optimization based on reinforcement learning method.The proposed method is evaluated by simulated cataract capsulorexis experiments.With the assistance of the proposed method,the performance of experienced doctors is enhanced in precision by 50.3％,to fresh doctors the precision is increased by 45.5％.The performance is enhanced without spending more execution time,which shows the proposed method enables the cutback of virtual fixture redundancy and the improvement of adaptability of virtual fixture to both experienced and fresh doctors.关键词
强化学习/虚拟夹具/适应性/机器人辅助眼科手术Key words
reinforcement learning/virtual fixture/adaptability/robot-assisted ophthalmic surgery分类
信息技术与安全科学引用本文复制引用
宿耀光,李旭昊,刘卫朋..基于强化学习的机器人眼科手术适应性虚拟夹具设计[J].燕山大学学报,2024,48(3):215-222,8.基金项目
河北省自然科学基金资助项目(F2020202009) （F2020202009）
河北省"三三三人才工程"资助项目(A202002001) （A202002001）
河北省研究生创新能力培养项目(CXZZSS2020029) （CXZZSS2020029）
