SIA OpenIR  > 机器人学研究室
主被动柔性机器人关节研究
Alternative TitleStudy on Active Passive Flexible Robot Joint
林光模1,2
Department机器人学研究室
Thesis Advisor刘光军 ; 赵新刚
Keyword柔性关节 串联弹性驱动器 刚度控制 螺旋动态优化 人机交互力估计
Pages113页
Degree Discipline模式识别与智能系统
Degree Name博士
2018-08-31
Degree Grantor中国科学院沈阳自动化研究所
Place of Conferral沈阳
Abstract医疗康复与辅助类机器人是一种对人的运动行为和力量进行补偿和增强以达到训练或完成日常生活活动目的的机器人。由于需要直接接触人的身体从而提供所需的驱动力,相比传统工业机器人,对这类机器人的安全性提出了更高的要求。传统机器人的设计往往使用基于电机驱动的刚性关节作为活动部件,并希望其关节具有尽可能大的刚度,从而能够提供更好的伺服性能。然而对于存在物理人机交互的机器人,往往希望其关节具有较低的机械阻抗,同时能够接受一定程度的伺服精度损失。柔性关节作为一种新型驱动方式,其动力学模型与控制方法都有别于传统刚性关节的机器人。并且由于其操作对象为具有主观运动意图的人体,而人体的运动也会影响机器人的自身状态,这也不同于传统机器人操作静止物体的情形。这些问题对柔性关节机器人的实际应用提出了挑战。为解决上述问题,本文在国家高技术研究发展计划和国家自然科学基金的联合资助下,研制了具有被动柔性元件的串联弹性关节实验系统,并深入研究了弹性元件参数优化设计、主动刚度控制方法和人机交互力估计方法,具体内容如下:首先,弹性元件是柔性关节系统中的关键因素。现有的弹性元件设计方法通常依赖于设计人员的经验与灵感,或采用简单的迭代修改方法进行设计,这样获得的柔性关节难以达到结构的最优化。本文在阿基米德螺旋线型平面涡卷弹簧的基础上,建立了弹簧设计参数与性能指标之间的参数化模型。针对弹簧的变形分析,提出了基于三次样条模型的弹簧收缩与舒张极限形状求解方法,并得到了两种极限情况下的弹簧变形量指标。以弹簧外形尺寸最小化为优化目标,采用改进的螺旋动态优化算法进行弹簧参数的优化设计,并在有限元仿真平台上验证了该优化方法的可行性。根据优化得到的弹簧参数,进行了被动柔性机器人关节系统的机构设计和控制系统开发。采用直流电机作为驱动装置,并设计了蜗轮蜗杆减速机构,利用其自锁特性,阻隔了从负载端到电机端的反向驱动,故可将电机当成理想的位置源。采用差分方式安装的绝对位置传感器用于测量弹簧的变形角度,以获取输出连杆受到的驱动力大小,从而实现力矩传感功能。控制系统部分采用基于Simulink Real-Time实时系统的快速控制原型开发平台,为后续控制方法的研究提供了实验基础。其次,对被动柔性关节的主动刚度控制方法进行了研究。由于被动柔性关节自带弹性元件,因此其动力学模型将围绕弹性元件进行分析。根据机器人关节刚度特性的定义,得到关节等效刚度与关节弹性元件物理刚度、以及弹簧输入端和输出端之间的动态模型。通过在线构建等效刚度的表达形式,得到基于输入输出加权反馈的等效刚度控制方法。通过实验分析得到不同刚度调整模式下的等效刚度控制效果,验证了该方法的有效性。最后,对人机交互力估计方法进行了研究。在人机交互过程中人机间相互作用力的大小可作为判断人体运动意图的依据,并能用于提高机器人控制系统的稳定性和鲁棒性。本论文采用基于时间多项式的模型作为人机交互过程中的交互力动态模型。采用自适应卡尔曼滤波方法进行交互力的在线估计,同时也对动态系统的噪声统计特性进行在线修正。并根据交互力及其各阶导数的估计结果对交互力模型参数进行在线自适应调整,使得交互力模型具有更好的适应能力。实际人机交互实验验证了该方法的有效性。本论文工作对被动柔性关节的设计、刚度控制及人机交互过程进行了深入研究,为被动柔性关节机器人的理论研究与实际应用提供了新思路。
Other AbstractMedical rehabilitative and assistive robot is a robot that compensates and strengthens the behavior and strength of human movement to achieve the purpose of training or completing daily life activities. Since it needs direct contact with human body, so as to provide the driving force needed, the safety requirement of this kind of robot is higher. The design of the traditional robot often uses motor driven rigid joints as moving parts, and it is hoped that the higher the joint stiffness is, the high precision servo performance can be provided. However, for the robot with physical human-robot interaction, it is usually expected that its joint has a lower mechanical impedance and can receive a certain degree of servo precision loss. As a new driving mode, flexible joints are different from traditional rigid joints in terms of their dynamic models and control methods. And because the operation object is the human body with subjective motion, the motion of the body also affects the state of the robot, which is different from the operation of a static object. These problems pose a challenge to the practical application of flexible joint robots. To solve the above problems, under the joint support of national high technology research and development program and National Natural Science Foundation, this paper developed a series elastic joint experimental system with passive flexible components, and three key methods of elastic component parameter optimization, active stiffness control and human-robot interaction force estimation are studied, the specific contents are as follows: First, the design of elastic elements is a key factor in flexible joint system. The existing design methods of elastic components usually depend on the experience and creativity of designers, or use simple modification attempt method to carry out iterative design. Therefore, the design of flexible joints based on this method is difficult to achieve structural optimization. Based on Archimedes spiral type plane torsional spring, the correspondence between spring design parameters and performance indexes is established in this paper. In view of the deformation analysis of the spring, the spring shrinkage and diastolic limit shape based on the spline model is proposed, and the spring deformation angle under two limit conditions is obtained. Minimization of spring shape size as an optimization target, a modified spiral dynamic optimization algorithm is used to optimize the spring design parameters, and the feasibility of the optimization method is verified on the finite element simulation platform. According to the optimized spring parameters, the mechanism design and control system development of the joint system of the passive flexible robot are carried out. The DC motor is used as the driving device, and the worm gear and worm deceleration mechanism is set up. Using its self-locking characteristic, the reverse drive from the load end to the motor end is blocked, so the motor can be regarded as an ideal position source. The absolute position sensor installed in the differential mode is used to measure the deformation angle of the spring, thus the driving force received by the output link can be obtained, and the torque sensing function is realized. The control system adopts the rapid control prototype development platform based on Simulink Real-Time system, which provides experimental basis for subsequent control methods. Secondly, the active stiffness control method of passive flexible joints is studied. Since the passive flexible joints have elastic elements, their dynamic models will be analyzed around the elastic elements. According to the definition of the joint stiffness characteristic of the robot, and the differential form of the relation between the spring input end and the output end, the physical stiffness of the joint elastic element and the joint equivalent stiffness can be obtained. An equivalent stiffness control method based on input output weighting feedback is obtained by constructing equivalent stiffness expression online. The equivalent stiffness control effect of different stiffness adjustment modes is obtained through experimental analysis, which verifies the effectiveness of the method. Finally, in the process of human-robot interaction, the interaction force can be used as a basis to judge the motion intention of the human body, and can be used to improve the stability and robustness of the robot control. Since the contact force sensor is usually only with limited coverage area and range, and its structure is complex, this paper proposes a human-robot interaction force estimation method based on the human-robot interaction dynamic model. The time polynomial based model is used as a dynamic interaction model in the process of human-robot interaction. The adaptive Kalman filtering method is adopted to estimate the interaction force online, and the online statistical correction of the dynamic system's noise is also corrected online. According to the estimation results of interaction force and its derivatives, the dynamic parameters of interaction force are adjusted adaptively so that the interaction force model has better adaptability. The effectiveness of the proposed method is verified in the actual human-robot interaction experiment. In this paper, the design, stiffness control and human-robot interaction of flexible joints are studied, and a new idea is provided for the theoretical research and practical application of passive flexible joint robot.
Language中文
Contribution Rank1
Document Type学位论文
Identifierhttp://ir.sia.cn/handle/173321/23652
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
林光模. 主被动柔性机器人关节研究[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
主被动柔性机器人关节研究.pdf(7548KB)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
[林光模]'s Articles
Baidu academic
Similar articles in Baidu academic
[林光模]'s Articles
Bing Scholar
Similar articles in Bing Scholar
[林光模]'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.