双足动态行走机器人可控周期步态逆问题数值化方法与控制器设计

发布时间：2018-04-03 12:47

本文选题：双足机器人　切入点：被动行走　出处：《吉林大学》2016年硕士论文

【摘要】：双足机器人因其具有与人类相似的运动结构,并且能够拓展延伸人类的活动空间,所以受到了学者的广泛关注。而且经过多年的努力,双足步行机器人无论是在理论研究还是在实体样机开发方面都取得了骄人的成绩,但是高能耗、低效率、行走步态不自然限制了它在工程实际中的应用。于是学者们把目光转移到Mc Geer提出的基于被动行走原理的仿人机器人理论上,为双足机器人的研究打开了新的思路。但是基于被动行走原理的双足机器人具有行走步态和行走环境单一以及稳定性差等缺点,因此本文对丰富双足机器人的行走步态并提高稳定性进行了研究。主要包括以下几个方面:1)本论文的研究对象是在斜面上完全被动行走的Compass-like圆规机器人模型。首先利用拉格朗日力学方法对系统的连续摆动阶段进行了动力学建模,然后进行了基于Matlab软件平台的仿真及实验分析,为后续研究打下基础。2)利用数值仿真实验手段,探究了机器人行走过程中的周期步态、控制输入、初始条件之间的关系。在髋关节和踝关节同时施加模拟重力矩输入时,成功找到了周期步态、控制输入、初始条件之间的定量关系及表达式,为可控周期步态的逆问题研究打开了一条新思路。仿真试验结果表明得到的表达式具有普遍性。同时对髋关节和踝关节独立驱动时进行了定性分析并给出了一般性结论。3)采用机械能导数法,设计了基于参考轨道能量的控制器。机器人在稳定行走过程中,机械能是守恒的,通过数值拟合发现了机械能与行走步态之间的解析式形式,进而通过机械能导数法分别设计了髋关节控制器和踝关节控制器。仿真实验结果表明,设计的控制器提高了系统对环境的适应能力和抗扰能力。4)通过虚拟约束的步态控制策略,结合反馈线性化的运动控制方法,设计了状态依赖的有限时间稳定控制器,实现了机器人在一步之内充分镇定。因为支撑腿在行走过程中角位移的变化是单调的,避免出现多值现象,所以令支撑腿作为姿态变量,选择摆动腿作为同步姿态变量,于是虚拟约束中不显含时间,使系统形成自治系统。仿真实验结果表明,对于欠驱动机器人来说,虚拟约束是一种行之有效的步态控制策略。
[Abstract]:Biped robot has attracted much attention because of its similar motion structure to human beings and its ability to extend the living space of human beings.After years of effort, biped walking robot has made great achievements in both theoretical research and physical prototype development. However, high energy consumption, low efficiency and unnatural walking gait restrict its application in engineering practice.Therefore, the scholars turn their attention to the humanoid robot theory based on passive walking principle proposed by MC Geer, which opens a new way for the research of biped robot.However, biped robot based on passive walking principle has the shortcomings of single walking gait and walking environment and poor stability, so this paper studies the rich walking gait and improving the stability of biped robot.The research object of this paper is the Compass-like compasses robot model which is completely passive walking on the slope.Firstly, the dynamic modeling of the continuous swing stage of the system is carried out by using Lagrange mechanics method, and then the simulation and experimental analysis based on Matlab software platform are carried out, which lay the foundation for further research. 2) the numerical simulation experiment method is used.The relationship between periodic gait, control input and initial conditions of the robot is discussed.The quantitative relations and expressions of periodic gait, control input and initial condition are found successfully when the simulated weight moment input is applied to the hip and ankle joint at the same time, which opens a new way for the study of the inverse problem of controllable periodic gait.The simulation results show that the expression is universal.At the same time, the qualitative analysis and general conclusion of independent drive of hip and ankle joint are given. 3) the controller based on reference track energy is designed by using mechanical energy derivative method.The mechanical energy of the robot is conserved in the process of steady walking. The analytic form between mechanical energy and walking gait is found by numerical fitting, and then the hip joint controller and the ankle joint controller are designed by the mechanical energy derivative method.The simulation results show that the designed controller improves the adaptability and immunity of the system to the environment. 4) the gait control strategy with virtual constraints is combined with the feedback linearization motion control method.A state dependent finite time stability controller is designed to stabilize the robot in one step.Because the change of angle displacement of supporting leg is monotonous in the course of walking, the phenomenon of multi-value is avoided, so that the support leg is used as attitude variable and swing leg is chosen as synchronous attitude variable, so the virtual constraint does not contain time.To form an autonomous system.Simulation results show that virtual constraint is an effective gait control strategy for underactuated robot.
【学位授予单位】：吉林大学
【学位级别】：硕士
【学位授予年份】：2016
【分类号】：TP242

【相似文献】