一种仿生四足机器人的系统设计及步态控制

CASIA OpenIR > 毕业生 > 硕士学位论文

	一种仿生四足机器人的系统设计及步态控制
	肖伟
	2016-05-23
学位类型	工学硕士
中文摘要	腿足机器人在自然环境下具有更好的机动性能，尤其是四足机器人，更易于适应各种复杂地形和环境。四足机器人在未来可用于灾害抢险、未知环境探索以及其他危险作业。本研究设计实现了一种仿生四足机器人（Biodog），主要包括四足机器人机构设计、驱动及传动计、控制电子及传感系统设计、系统集成及调试，并在此平台上进行控制算法理论和实验研究。基于稳定性和可靠性考虑，机器人每条腿设计有两个主动转动自由度和一个被动转动自由度，并在机构设计上为以后拓展自由度留下机械接口。针对步态控制，本文提出了一种平稳的步态变换算法以及一种融合中枢模式发生器和基于模型的控制方法，该控制策略充分集成了两种控制方法的优势。本文核心研究内容概要如下： 1.生物腿结构启发的腿足机构设计和紧凑型驱动及传动设计。通过连杆机构传动，将电机置于四足机器人本体，优化并实现了腿足结构的低惯量设计，提高了腿足的动力学性能。同时，受动物表层趾屈肌（SDF）腿足机理启发，设计了一种基于动物SDF 腿足机理的四足腿足结构，该结构具有减震、蓄能功能，并易于实现避障。基于低惯量腿足结构的要求，设计了直流伺服电机驱动的结构紧凑、拆装方便、全密封、易于维护的四足机器人同轴传动系统。为保证髋关节输出轴的强度和刚度，设计了一种双支撑的闭环强化结构。 2.基于CANopen 协议的控制电子及传感系统设计和机械及电控系统集成。四足机器人各关节伺服驱动器组成一个CAN 总线网络，基于CANopen 协议通讯，由ARM11 和DSP控制；同时还开发了基于Windows 的控制系统和软件，便于调试。四足机器人本体安装有陀螺仪，用于行走过程的位姿检测；足端内嵌FSR 力传感器，以检测足-地接触力。最终对低惯量腿足结构、控制电子、陀螺及力传感器、通讯软件、电机伺服驱动等进行了有效集成及测试。 3.步态控制算法研究。基于Hopf 振荡器，为实现四足机器人连续和平稳的步态变换，提出了一种可实现典型步态之间步态变换的控制算法，并从能源效率的角度分析了步态变换的条件。针对四足机器人自适应行走，提出了一种融合CPG 和模型的控制方法，结合四足机器人Biodog 分析了腿足运动模式和约束，以提高运动的稳定性和有效防止足端打滑。 4.步态变换仿真和融合算法实验研究。对于四足机器人步态变换算法，通过Adams和Matlab 联合仿真，实现了四足机器人Walk-Trot-Gallop 的连续平稳步态变换，并通过质心速度波动、控制信号极限环以及足端受力情况分析了步态变换的稳定性。为实现自适应行走的融合算法则基于Biodog 做了实验研究，通过比较一般的CPG 和改进后的融合算法实验分析了四足机器人行走过程在稳定性和足端防滑方面的性能改善。
英文摘要	Legged robots, especially for quadruped robots, excel in natural environment locomotion because of their higher mobility and better environment adaptability. The quadruped robot can be used in dangerous work in the future, such as disaster aid and exploration etc.. Enlightened by biomimetics, this thesis has designed a quadruped robot (named Biodog), including mechanism design, transmission design, mechatronics, system integration, and pays efforts on control algorithms and experiments on the Biodog. Each leg includes two active joints and one passive joint. This research also proposes a stable gait transition algorithm and synthesizes central pattern generator (CPG) and model-based algorithm. To be a low-inertia-leg quadruped, the robot should arrange all actuators on the trunk. Links are utilized for transmission, which significantly improves the leg dynamics. Inspired by animals’ leg muscle mechanism, we design a superficial digital flexor (SDF) for each leg to absorb shock and save energy, and this mechanism can also facilitate barrier crossing. The transmission system, using bevel gears and worm gears, has two coaxial driving outputs for the hip joint and knee joint link respectively, which are driven by DC servo motors. The whole gear case is sealed to prevent dust, and the structure is also convenient for maintenance. Using CANOpen protocol, motor controllers are connected through CAN bus and can communicate with ARM11 and DSP. We have also developed Windows-based control system and software to facilitate test. A gyroscope, installed on the trunk, is employed to measure posture, and force sensitive resistors (FSR), embedded in the feet, are used to sense foot-ground contact force. The control software, integrating sensor information and control algorithms, is compiled with C++ language. The quadruped Biodog has integrated mechanics, electronics, sensors, controller, control algorithms and software, which can be used as a platform for further research. The control algorithms include basic Hopf oscillator, gait transition algorithm and synthesized algorithm. Hopf oscillator is one of the CPG algorithms, and it prevails over others for its decoupling character and adjustable duty factor. The gait transition algorithm, based on the Hopf oscillator and considering energy efficiency, can achieve needed gait transition between every two gaits with a stable and natural transition process. This research also proposes a quadrilateral gait strategy and an algorithm that can synthesize kinematics and Hopf oscillator to facilitate motion planning for quadruped adaptive locomotion. The quadrilateral gait strategy is used to select a gait to better match with the environment and select footfall to get a relatively stable posture according to the environment model. The stride length and foot lift height are transmitted to the synthesized controller by quadrilateral gait strategy. Then the CPG controller, synthesized with inverse kinematics, can get the amplitude of the hip joint angle and trajectory of each joint. The gait transition algorithm is verified on a virtual prototype through the combined simulation with Adams and Matlab, which achieves continuous and stable walk-trot-gallop gait transition for the prototype. Simulation results demonstrate its stability through the fluctuation of center of mass, limit cycle of control signal and foot-ground contact force analysis. Finally, we conduct experiments on the quadruped robot Biodog, and experiment results demonstrate the advantages of the synthesized CPG over the normal one to effectively reduce foot sliding and improve the locomotion stability
关键词	四足机器人仿生腿中枢模式发生器步态控制步态变换
语种	中文
文献类型	学位论文
条目标识符	http://ir.ia.ac.cn/handle/173211/11672
专题	毕业生_硕士学位论文
推荐引用方式 GB/T 7714	肖伟. 一种仿生四足机器人的系统设计及步态控制[D]. 北京. 中国科学院大学,2016.

条目包含的文件
文件名称/大小	文献类型	版本类型	开放类型	使用许可
一种仿生四足机器人的系统设计及步态控制.（19534KB）	学位论文		限制开放	CC BY-NC-SA