人机交互力对下肢助力外骨骼关节期望角度的预测  

Prediction of Expected Angle of Joints of Lower Limb Assisted Exoskeleton with Human-Robot Interaction

作　　者：宋纪元[1,2] 朱爱斌[1,2] 屠尧 武鑫雨 张育林[3] 周旭 SONG Jiyuan;ZHU Aibin;TU Yao;WU Xinyu;ZHANG Yulin;ZHOU Xu(Institute of Robotics & Intelligent Systems, Xi’an Jiaotong University, Xi’an 710049, China;Shaanxi Key Laboratory of Intelligent Robots, Xi’an Jiaotong University, Xi’an 710049, China;Key Laboratory of Education Ministry for Modern Design and Rotor-Bearing System, Xi’an Jiaotong University, Xi’an 710049, China;Taiyuan Research Institute of China Coal Technology & Engineering Group Co. Ltd., Taiyuan 030006, China)

机构地区：[1]西安交通大学机器人与智能系统研究所,西安710049 [2]西安交通大学陕西省智能机器人重点实验室,西安710049 [3]西安交通大学现代设计及转子轴承系统教育部重点实验室,西安710049 [4]中国煤炭科工集团太原研究院有限公司,太原030006

出　　处：《西安交通大学学报》2021年第12期55-63,69,共10页Journal of Xi'an Jiaotong University

基　　金：国家自然科学基金深圳联合重点基金资助项目(U1813212);陕西省重点研发计划资助项目(2020GXLH-Y-007);山西省关键核心技术和共性技术研发攻关专项资助项目(2020XXX001)。

摘　　要：针对下肢助力外骨骼在穿戴者使用过程中人机交互力过大导致舒适性差的问题,对外骨骼关节角速度与腿部交互力产生的主动关节力矩间的映射关系进行研究,提出了将人体下肢摆动相腿部简化为以髋关节为支点的平面二杆模型,建立人体下肢摆动腿动力学模型,得到人体的重力补偿项。采用力传感单元获取人体与外骨骼之间的交互力,将外骨骼与人体间的腿部交互力转化为人体髋膝关节下一时刻运动的期望角度;分别通过S曲线映射和导纳控制模型建立了人体摆动相关节角速度与腿部交互力补偿人体重力分量后产生的关节力矩之间的非线性映射关系。实验结果表明:基于S曲线映射的期望角度预测算法和基于导纳控制模型的期望角度预测算法都实现了对外骨骼对人体运动意图的连续跟踪;基于S曲线映射的期望角度预测算法使得外骨骼跟随人体运动过程中大腿交互力峰值小于25 N,小腿交互力峰值小于20 N;基于导纳控制模型的期望角度预测算法使得外骨骼跟随人体运动过程中大腿腿部交互力峰值小于15 N,小腿腿部交互力峰值小于15 N,且关节角度曲线平滑,能够更好地识别人体运动意图,提升人机交互柔顺性。The mapping relationship between the joint angular velocity of the external skeleton and the joint torque generated by the interaction force of the legs is studied to solve the problem of poor human-robot interaction flexibility and discomfort caused by excessive human-robot interaction force in the use of lower limb assisted exoskeleton.A model to simplify the human lower limb swing phase leg into a plane two-bar model with the hip joint as the fulcrum is proposed,and a dynamic model of swing leg of human lower limb is established to obtain the gravity compensation item of human body.A force sensing unit is used to obtain the interaction force between the human body and the exoskeleton,and the leg interaction force between the exoskeleton and the human body is converted into the desired angle of the human hip and knee joint movement at the next moment.A non-linear mapping relationship between the swing phase joint angular velocity of human body and the leg interaction torque is established by means of the S-curve mapping and the admittance control model respectively.Experimental results show that both the expected angle prediction algorithm based on S-curve mapping and the expected angle prediction algorithm based on the admittance control model realize the continuous tracking of human motion intention from the exoskeleton.The expected angle prediction algorithm based on S-curve mapping makes the peak value of the thigh and leg interaction force less than 25 N and the peak value of shank and leg interaction force less than 20 N in the process of the exoskeleton following the human body movement;while the expected angle prediction algorithm based on the admittance control model makes the peak value of thigh-leg interaction force less than 15 N and the peak value of calf-leg interaction force less than 15 N in the process of the exoskeleton following human movement,and the joint angle curve is smooth,which can better identify human movement intention and improve the flexibility of human-robot interaction.

关 键 词：下肢助力外骨骼 人机交互力 人体运动意图 导纳控制 

分 类 号：TH117.1[机械工程—机械设计及理论]