| 题名 | 基于张拉整体的可变结构体机器人移动步态研究 |
| 作者 | 杜汶娟 |
| 学位类别 | 博士 |
| 答辩日期 | 2017-05-31 |
| 授予单位 | 中国科学院沈阳自动化研究所 |
| 授予地点 | 沈阳 |
| 导师 | 马书根 ; 王越超 |
| 关键词 | 可变结构体机器人 六压杆张拉整体结构 几何拓扑 有限元法 遗传算法 |
| 其他题名 | Study on Gaits of a Structure Variable Robot |
| 学位专业 | 机械电子工程 |
| 中文摘要 | 本文以国家自然科学基金面上项目“可变结构体机器人多步态多相型运动机理研究”为依托，针对六压杆可变结构体机器人构型设计与多步态运动机理展开研究。具体研究内容包括： 一、构型选择与结构参数优化：本文通过比较不同张拉整体结构特性，选择出形变能力强、结构对称易于分析的初始构型用于机器人设计；并通过结构参数优化降低关节受力，避免机器人结构崩溃。可变结构体机器人主要依赖自身变形产生各类步态，而其初始构型和结构参数对其变形能力有极大影响；结构参数影响关节受力，而过大的关节受力会损伤关节，导致机器人结构崩溃。但现有可变结构体机器人构型设计时，并未对上述两类问题进行研究，在构型选择和结构参数选择时具有一定的盲目性。本文提出基于拓扑几何的构型选择方法，与基于FEM(Finite Element Method)的结构参数优化方法解决上述问题。 二、样机设计与仿真平台搭建：为验证机器人多类步态运动控制方法，本文设计了I代气动样机与高度集成、拉索全驱动的II代电动实验样机，并分别搭建了ODE仿真平台与MATLAB/C++仿真平台，进行辅助验证。现有样机包括气动、电动两类，其中气动样机设计简单，但集成度差，外置的气动元件限制了机器人的机动性；电动样机集成度高，但受限于驱动器性能，现有研究多通过减少驱动器数量以减小负载，因而未能实现拉索全驱动设计。本文综合考虑驱动器性能，配合优化结构设计，搭建了高集成、拉索全驱动设计的II代电动样机。 三、滚动步态研究：本文针对六压杆可变结构体机器人滚动步态运动控制方法展开研究。由于张拉整体结构的多输入、非线性、耦合度高等特性，目前针对滚动步态运动的研究多基于穷举试验，而未进行深入的理论分析。滚动步态运动速度较快，适合在平整地形下运动，其利用自身形变、改变重心位置，在不同稳态中顺序翻滚而达到运动的效果，因此，滚动步态分析的核心问题是分析自身变形情况。本文提出了基于FEM的滚动方向控制方法，分析结构在不同驱动参数作用下的变形情况，从而控制滚动运动方向；同时，提出了基于FEM的驱动参数优化方法，以能耗为优化目标，对驱动参数进行优化。 四、蠕动步态研究：本文针对六压杆可变结构体机器人的蠕动步态运动控制方法展开研究。由于六压杆结构较三压杆结构更为复杂：其一，自由度、驱动器数量更多，增加了建模难度；其二，结构类似球形，极易发生滚动，需解决稳定性问题，因此，现有蠕动步态研究中仅有针对三压杆结构的研究，而未有对六压杆结构的研究。蠕动步态通过模拟多足动物的爬行步态，通过小幅变形改变机器人与地面接触点的摩擦力大小，实现机器人蠕动运动；蠕动步态虽移动速度较小，但由于变形幅度较小，运动十分稳定，较滚动步态更适合在斜坡等复杂地形下运动。本文提出基于遗传算法的蠕动步态控制方法：首先，利用牛顿-欧拉Newton-Euler法建立其动力学模型，基于该模型分析其初始构型稳定性与可使机器人稳定蠕动的驱动参数范围；之后，选择合适的目标函数，利用遗传算法获得可使机器人蠕动前进的驱动参数最优解。 上述滚动、蠕动步态控制方法均在仿真平台及实验平台完成验证。 |
| 英文摘要 | This work is supported by the National Natural Science Foundation of China “Study on Movement Mechanism of Multi-gaits and Multi-configuration of Structure Variable Robot”, and focuses on the multi-gaits of a 6-strut structure variable robot. The main parts of this thesis are shown below. 1. Selection of tensegrity type and optimization of structure parameters: by comparing the characteristic of various tensegrities, the tensegrity with high deformable capacity and symmetric structure is chosen as the model for the configuration design of the structure variable robot. Moreover, the structure parameters are optimized to decrease the force on joints to avoid collapse of structure. The structure variable robot can generate multiple gaits by self-deformation, and the initial tensegrity model and the structure parameters have great influence on the deformable ability. Meanwhile, large force on the joints of the robot may cause damage of joints and make the robot collapsing. These two problems mentioned above, however, are not studied in the existing researches on the structure design of the structure variable robot, and the initial tensegrity model and the structure parameters are chosen without theoretical support. To solve the two problems mentioned above, this paper proposes an initial configuration selection method based on the geometry and topology, and an optimization method for the structure parameters based on the FEM (finite element method). 2. The design of the prototypes and the establishment of the simulation environments: for validating the motion control methods for different gaits, the I-generation pneumatic prototype and the II-generation electric prototype (highly integrated, all-cable-drive) are built, and the dynamic simulation environments are built on the platform of ODE and MATLAB, respectively, to verify the validity of our researches. The existing prototypes can be divided as two types: the pneumatic prototype and the electric prototype. The design of the pneumatic prototype is simple, however, the external pneumatic components limit the mobility of the robot as the difficulty of integration. The electric prototype is highly integrated, however, due to the limitation of the performance of the actuators, the existing researches mostly reduce the number of actuators to reduce the load, and thus fail to realize the all-cable-drive design. We build the highly integrated and all-cable-driven electric prototype by considering the power-to-weight ratio of the actuators, and optimizing the design of the mechanism. 3. Rolling gait: this paper focuses on the motion control method for the rolling gait of the 6-strut structure variable robot. As the tensegrity structures is a multi-input, nonlinear and highly coupled system, the existing researches on the rolling gait are mostly based on experimental experiences, yet without in-depth theory analysis. The robot can change its center of gravity with self-deformation, and then generate rolling motion by converting among different stable states. Hence, the key problem of analysis of the rolling motion is to analyze the deformation of the structure. We propose a rolling motion control method based on the FEM to control the rolling direction by analyze the deformation with different driving parameters, and an optimization method for the driving parameters based on the FEM to optimize the driving parameters to decrease the energy consumption. 4. Crawling gait: this paper focuses on the motion control method for the crawling gait of the 6-strut structure variable robot. As the 6-strut tensegrity is more complicated than the 3-strut tensegrity: Firstly, the difficulty of modeling is increased as the increasing of the number of DOFs and actuators. Secondly, the 6-strut tensegrity model is like a sphere and can roll easily, which challenges the stability of the crawling motion. Hence, the study on the crawling gait of the robot based on the 3-strut tensegrity already exists, but the one based on the 6-strut tensegrity has not been studied. The crawling gait is generated by imitating the crawling gait of the multi-legged animals, and the structure variable robot can crawling by changing the friction between the ground and the nodes contacting with the ground with small self-deformation. The speed of the crawling gait is smaller than that of the rolling gait, but the small deformation makes the robot move more stable and make the crawling gait more useful in complex terrain like a slope. We propose a crawling gait control method based on the genetic algorithm. Firstly, we model the dynamics of the robot by Newton-Euler method, and derive the stable initial configuration and stable range of driving parameters. Then, by choosing an appropriate objective function, the optimal solution of driving parameters that can make the robot crawling forward is selected by the genetic algorithm. The control methods for the rolling and crawling gaits are validated on both the simulation platforms and the robot prototypes. |
| 语种 | 中文 |
| 产权排序 | 1 |
| 内容类型 | 学位论文 |
| 源URL | [http://ir.sia.cn/handle/173321/20562]  |
| 专题 | 沈阳自动化研究所_机器人学研究室 |
| 推荐引用方式 GB/T 7714 | 杜汶娟. 基于张拉整体的可变结构体机器人移动步态研究[D]. 沈阳. 中国科学院沈阳自动化研究所. 2017. |
| 个性服务 |
| 查看访问统计 |
| 相关权益政策 |
| 暂无数据 |
| 收藏/分享 |
除非特别说明，本系统中所有内容都受版权保护，并保留所有权利。
修改评论