基于表面肌电信号的人机动作信息传递研究

发布时间：2018-05-11 12:07

本文选题：服务机器人 + 变阻抗控制　；参考：《哈尔滨工业大学》2017年博士论文

【摘要】：随着社会生产和生活方式的快速变化,传统机器人已无法满足生产的高柔顺性以及智能性、灵巧性的要求和人们生活环境的复杂多变需求。因而新一代机器人需具有较好的人机友好界面,能够快速、灵巧地具备人的认知及操作技能、能够与人共享工作空间,提高生产效率、质量等。本文以人的阻抗调节机理和机器人阻抗控制技术为基础,基于表面肌电信号(sEMG)对人的肢体动作信息,即阻抗和运动学信息进行建模和提取,通过构建人机动作信息协调控制系统,实现人的动作信息向机器人的自然、柔顺传递。人体肢体运动与传统的机器人运动相比,具有较好的柔顺性、环境适应性和复杂的技巧性。这些特性得益于人类长期进化产生的机械阻抗调节技能来适应交互环境的变化。因而人机动作信息传递只有同时包含运动学信息和阻抗信息的传递,才能实现人的动作特性向机器人的传递。本文首先对目前人体阻抗调节和机器人柔顺性人机交互技术发展现状进行了综述,总结了机器人在面临新的生产方式和生活方式上的不足,归纳了下一代机器人相对于传统机器人更高的柔顺性要求,进而引出本文的研究目的和意义。接着对人的上肢运动灵巧性,柔顺性的动作技能产生机理,即阻抗调节,技能学习优化进行了系统综述,尤其在与机器人技术结合,实现人机动作技能传递等方面进行了深入概括。在此基础上,对近些年来仿人阻抗调节的机器人关节设计及控制方法进行了综述。接着对基于人体阻抗调节的机器人人机交互系统及相关方法进行了类比,分析和总结。为本文开展研究奠定了前期理论和工程基础。基于人体上肢阻抗调节的生物学原理,本文进行了基于sEMG信号的人体上肢刚度建模、辨识。采用二阶系统模型对人体上肢3D笛卡尔空间刚度模型进行了系统辨识,验证了人体阻抗调节特性。对于人体上肢末端的阻抗辨识,本文系统阐述了人体肌肉笛卡尔空间刚度向关节空间刚度的变换,以及关节空间向笛卡尔空间的逆变换。设计了sEMG信号平滑包络技术算法和人体上肢刚度辨识算法。上肢雅可比矩阵是刚度空间变换的重要参数,本文基于机器人D-H坐标和人的上肢相应关节动作匹配,对人的手臂运动学进行了建模,基于视觉和惯性测量单元(IMU)对人体上肢的雅可比矩阵进行提取,为实现人体上肢的刚度在线辨识奠定了基础。为实现人体阻抗信息和运动学信息向机器人的传递,本文对机器人变阻抗控制原理进行了深入研究,设计了机器人变阻抗控制器模型,构建了基于人体上肢阻抗调节的机器人人机动作信息传递控制模型,可使机器人快速模仿人的肢体动作。为提高人和机器人交互的自然性和柔顺性,本文引入触觉反馈对人机动作信息传递进行了对比试验研究。本文开展了基于sEMG信号的人机动作信息传递应用研究,如人体关节抗干扰和机器人自适应抗干扰实验研究,提重物实验,机器人阻抗写字示教实验和人-机器人-人写字技能训练实验及遥操作实验研究。实现了人的阻抗调节信息和运动学信息向机器人的自然传递,验证了本文提出的人机动作信息传递的有效性和可行性。
[Abstract]:With the rapid change of social production and life style, traditional robots have been unable to meet the requirements of high compliance, intelligence, dexterity and the complex and changeable needs of people's living environment. Therefore, the new generation of robots must have a good human-machine friendly interface, fast and dexterity with human cognition and operation skills. In this paper, based on the human impedance regulation mechanism and the robot impedance control technology, based on the surface EMG signal (sEMG), this paper builds and extracts the human body movement information, namely the impedance and kinematics information, through the construction of the human machine action information coordination control system, and realizes the human being. Movement information transfers to the nature and flexibility of the robot. Body movement of human body has better flexibility, environmental adaptability and complex skill than traditional robot motion. These characteristics benefit from the mechanical impedance adjustment skills produced by human long-term evolution to adapt to the change of interactive environment. In this paper, the current status of human body impedance regulation and robot compliant human-computer interaction technology is summarized, and the shortcomings of the robot in the face of new production mode and life style are summarized. A generation of robots is more compliant than traditional robots, and then leads to the purpose and significance of the research. Then, the mechanism of the dexterity of the movement of the upper limb and the mechanism of the flexibility of action skills, that is, the impedance adjustment, and the optimization of skill learning are systematically reviewed, especially in combination with the robot technology to realize the man-machine action skill transmission. On the basis of this, the joint design and control methods of humanoid impedance regulating robot in recent years are summarized. Then the robot human-computer interaction system based on human impedance regulation and the related methods are analogous, analyzed and summarized. The preliminary theory and engineering are laid for this study. Based on the biological principle of the impedance regulation of the human upper limb, this paper carries out the modeling of the stiffness of the upper limb based on the sEMG signal. The two order system model is used to identify the 3D Cartesian spatial stiffness model of the human upper limb, and the impedance characteristics of the human body are verified. The transformation of space stiffness of human muscle Descartes space stiffness to joint space stiffness and the inverse transformation of joint space to Descartes space are expounded. The sEMG signal smoothing envelope technique and the stiffness identification algorithm for human upper limb are designed. The upper limb Jacobi matrix is an important parameter for the spatial transformation of stiffness. This paper is based on the D-H coordinates of the robot and the upper part of the human body. The kinematics of the arm is modeled by the corresponding joint of the limb. The Jacobi matrix of the upper limb of the human body is extracted based on the visual and inertial measurement unit (IMU), which lays the foundation for the realization of the on-line identification of the stiffness of the upper limb. The principle of impedance control is deeply studied. A robot variable impedance controller model is designed. A robot man-machine action information transfer control model based on the impedance adjustment of the body upper limb is constructed, which can make the robot imitate human movements quickly. In order to improve the nature and flexibility of human and robot interaction, the tactile feedback is introduced in this paper. The study of man-machine action information transmission is carried out by comparative experiment. In this paper, the application of man-machine action information transmission based on sEMG signal is carried out, such as anti interference of human joint and Experimental Research on Adaptive Anti-jamming of robot, experiment of lifting heavy objects, experiment of robot impedance writing instruction, training experiment of human and human writing skill and tele operation It realizes the natural transmission of the human impedance adjustment information and kinematics information to the robot, and verifies the validity and feasibility of the man-machine action information transmission proposed in this paper.

【学位授予单位】：哈尔滨工业大学
【学位级别】：博士
【学位授予年份】：2017
【分类号】：TP242

【相似文献】