SIA OpenIR  > 机器人学研究室
基于人工肌肉的仿人手臂关键技术研究
Alternative TitleReasearch on the Key Technologies in Artificial Muscle-based Humanoid Arm
张道辉1,2
Department机器人学研究室
Thesis Advisor韩建达
Keyword形状记忆合金 气动人工肌肉 仿人手臂 主动模型控制 力与刚度控制
Pages126页
Degree Discipline模式识别与智能系统
Degree Name博士
2018-11-27
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract当前机械臂技术从刚性本体非本质安全、与人非接触及物理空间隔离的现状正朝着柔性本体本质安全、与人紧密协调合作及同一自然空间的人机共融新趋势发展。传统电机、液压等刚性驱动存在功率密度比低、质量大、柔顺性和安全性差等缺点,而新型人工肌肉柔性驱动具有功率密度比高、结构简单、柔顺性和安全性好、类似人体骨骼肌特性等优点。因此,基于人工肌肉柔顺驱动研制轻质、灵活、柔顺、安全、低成本的仿人手臂,可以推动机械臂技术朝着与人共融的新方向发展。然而,人工肌肉材料响应慢、易疲劳、稳定性差,存在强迟滞、蠕变等非线性因素使得难以对其进行建模与控制,人工肌肉固有的柔顺性对仿生关节及仿人手臂系统的控制也带来了新的挑战。本文针对以上问题展开深入研究,主要内容包括人工肌肉主动建模与控制,仿生关节力与刚度控制,以及仿人手臂柔顺控制与抓取规划。具体研究内容如下:首先,针对形状记忆合金和气动人工肌肉的强非线性、迟滞和时变特性造成难以建立精确模型及设计高性能控制器的问题,提出了基于主动模型的控制方法。根据形状记忆合金和气动人工肌肉的物理特性和驱动机制分别建立其数学模型,将由外部扰动、时变参数以及未建模动态引起的不确定性造成的模型差引入到系统的状态方程中,利用联合估计技术对包含模型差的系统状态方程进行在线估计,同时估计出系统状态和模型差。基于在线更新的主动模型设计自适应控制器及其补偿控制策略,来提高对形状记忆合金和气动人工肌肉的控制性能。其次,针对柔性仿生关节难以实现力与刚度独立控制的问题,建立了一种新的PAM等效弹簧模型以及关节力与刚度模型,设计了一种双输入双输出控制器。PAM拮抗关节系统数学模型复杂且难以精确描述时变的实际系统,关节位置、力和刚度高度耦合,这都给关节力和刚度独立控制带来很大困难。而基于PAM等效模型建立的关节力和刚度模型可以实现对力和刚度解耦,进而由双输入双输出控制器实现对它们的独立控制。以此为基础,通过引入位置PID前馈环建立位置与力的关系,还可以实现对关节位置和刚度的独立控制。最后,针对被动柔顺性的不可控制性和主动柔顺性的过于依赖反馈控制的问题,提出一种主、被动柔顺性相结合的控制策略,将人工肌肉固有的被动柔顺性与基于控制的主动柔顺性相结合,实现仿人手臂与人或环境的安全交互。研制了一款基于人工肌肉的仿人手臂系统,设计无模型自适应阻抗控制器对仿人手臂进行了柔顺控制实验。另外,基于单目视觉开展了仿人手臂抓取物体实验。实验验证了所研制仿人手臂系统的综合性能以及所提出方法的有效性。本文的工作依次从驱动、关节及整个手臂系统三个层面对基于人工肌肉的仿人手臂技术进行了深入研究,为当前机械臂技术所面临的安全性、轻量化及低成本的挑战提供了一条新的解决思路,为后续实现仿人手臂与人共融提供了一定的理论基础和技术支撑。
Other AbstractNowadays, manipulator technology is developing towards a new trend of human-machine coexistence and cooperation from non-essential security of rigid ontology, non-contact and physical space isolation with humans to essential safety of flexible ontology, close coordination and cooperation and the same natural space with humans. The traditional motor, hydraulic, and other rigid actuators have the disadvantages of low power density, heavy weight, poor flexibility and safety, etc., while new artificial muscle flexible actuators have the advantages of high power density, simple structure, good flexibility and safety, and similar to the characteristics of human skeletal muscle. Therefore, the research and development of a lightweight, flexible, compliant, safe and low-cost humanoid arm based on artificial muscle flexible actuators can promote the development of the manipulator technology towards a new direction of coexistence and cooperation with human beings. However, artificial muscle materials are slow in response, easy to fatigue, poor in stability; and there exist strong hysteresis, creep and other nonlinear factors that make it difficult to model and control; and the inherent flexibility of artificial muscles brings new challenges to the control of bionic joint and humanoid arm. In order to solve the above issues, this paper will conduct an in-depth study, mainly including active modeling and control of artificial muscles, force and stiffness control of bionic joint, and compliance control and grasping planning of humanoid arm. The concrete contents are as following: Firstly, an active model-based control scheme is proposed to solve the problem that it is difficult to build an accurate model and design a high-performance controller due to strong nonlinear, hysteresis and time-varying characteristics of shape memory alloy and pneumatic artificial muscle. According to the physical properties and driving mechanism of shape memory alloy and pneumatic artificial muscle, their mathematical models are established respectively. The model error caused by the uncertainties owing to external disturbance, time-varying parameters and unmolded dynamics is introduced into the system state equation. And the joint estimation technique is utilized to online estimate the system states and the model error simultaneously. The adaptive controller and its compensation control strategy are designed based on the online updating active model to improve the control performance of shape memory alloy and pneumatic artificial muscle. Secondly, considering the problem that it is hard to achieve independent force and stiffness control for flexible bionic joint, a new PAM equivalent spring model and joint force and stiffness model are established, and a dual-input and dual-output controller is designed. The mathematical model of antagonistic joint system driven by PAM is complex and difficult to accurately describe the time-varying actual system. And joint position, force and stiffness are highly coupled. All these bring great difficulties to joint force and stiffness control. However, the joint force and stiffness model based on the PAM equivalent model can achieve the decoupling of force and stiffness, and their independent control is realized by a dual-input and dual-output controller. Furthermore, the relationship between position and force can be established by introducing the position PID feedforward loop, and then the independent position and stiffness control of bionic joint can be done as well. Finally, in view of the uncontrollability of passive compliance and the over-reliance on feedback control of active compliance, a control strategy combining active and passive compliance is proposed, which combines the inherent passive flexible of artificial muscle with the active compliance based on control and realizes the safe interaction between humanoid arm and human or environment. An artificial muscle-based humanoid arm is developed, and a model-free adaptive impedance controller is designed to control it. In addition, we carried out the object grabbing experiment based on monocular vision. The experiments test the comprehensive performance of the humanoid arm prototype and verify the effectiveness of the proposed methods. The work of this paper carried out an in-depth research on artificial muscle-based humanoid arm technology from the three levels of actuator, joint and whole arm system in turn. It provides a new solution to the challenges of safety, lightweight and low cost faced by the current robotic arm technology. It also provides some theoretical foundation and technical support for the subsequent realization of humanoid arm and human coexistence and cooperation.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/23650
Collection机器人学研究室
Affiliation1.中国科学院沈阳自动化研究所
2.中国科学院大学
Recommended Citation
GB/T 7714
张道辉. 基于人工肌肉的仿人手臂关键技术研究[D]. 沈阳. 中国科学院沈阳自动化研究所,2018.
Files in This Item:
File Name/Size DocType Version Access License
基于人工肌肉的仿人手臂关键技术研究.pd(6606KB)学位论文 开放获取CC BY-NC-SAApplication Full Text
Related Services
Recommend this item
Bookmark
Usage statistics
Export to Endnote
Google Scholar
Similar articles in Google Scholar
[张道辉]'s Articles
Baidu academic
Similar articles in Baidu academic
[张道辉]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[张道辉]'s Articles
Terms of Use
No data!
Social Bookmark/Share
All comments (0)
No comment.
 

Items in the repository are protected by copyright, with all rights reserved, unless otherwise indicated.