【作者】 王晓龙；

【导师】 李祖枢；

【作者基本信息】 重庆大学 ， 控制理论与控制工程， 2010， 硕士

【摘要】 仿人机器人技术综合涉及机械学、计算机、电子学、力学、生物学、控制论、系统工程、人工智能等多学科领域,已成为21世纪最具代表性的高技术,是目前科技发展最活跃的领域之一,代表了机电一体化的最高成就。仿人机器人研究一直是机器人研究领域的热点之一,在仿人机器人实现稳定行走研究中,如何增强仿人机器人对环境的适应能力一直是研究的重点之一。本文以ZMP稳定性判据作为仿人机器人步态行走的稳定性判据,主要研究如何设计并实现仿人机器人脚底传感系统在线获取ZMP,并对机器人进行稳定性控制。首先,论文回顾和总结了国内外仿人机器人的研究历史及其发展现状,对各主要研究机构的仿人机器人脚底传感器设计和应用情况进行了分析和对比,阐述了本课题的研究背景、研究任务以及应用前景,提出了本文研究的主要内容。其次,从分析步行方式的分类与比较入手引入ZMP(零力矩点)、稳定区域、稳定裕度的概念,介绍了ZMP点的计算公式及其性质,分析了ZMP点与稳定的关系,进而引出了ZMP稳定性判据,并以此为仿人机器人行走的稳定性判据。用Lagrangian方法建立了仿人机器人的动力学模型并以此为研究对象,从理论上详细分析得出在无反馈信息的条件下,机器人无法通过调整踝关节实现稳定行走,进而说明了研制脚底传感系统检测ZMP点位置,并引入脚底力信息反馈控制的必要性。再次,分析了目前仿人机器人控制系统的控制结构和硬件结构以及其局限性,提出在现有硬件结构的基础上建立方便和原控制系统连接且易于扩展的外部传感器信息采集通道,形成开放的分层信息采集与处理结构,为脚底传感系统的信息采集通道以及多传感器信息的融合提供了依据。然后,设计并实现了适用于本课题的脚底传感系统,并对其整体结构以及各个部分的功能及设计(包括机械设计、硬件设计、软件设计)做了详细的阐述,通过实验证明了该系统的有效性。最后,在仿人机器人简化模型的基础上以前向调整为例,给出了脚底传感信息在实际中如何应用,通过仿真得到在引入了脚底传感信息的条件下,使用改进遗传算法和分段PD控制结合的方法对仿人机器人进行控制的可行性及其良好的控制效果。更多还原

【Abstract】 Humanoid robot technology involving mechanics, computer, electronics, dynamic, biology, cybernetics, systems engineering, artificial intelligence and many other technologies, has become the most representative high technologies of 21st century. It represents the highest achievements of mechanical and electrical integration, and is the most active fields of scientific and technological development.As a hot research, how to enhance the humanoid robot’s ability to adapt to the unknown environment has been the focus of the study in humanoid robot research. Through designing and realizing foot sensor system, this research is order to on-line access ZMP point, and use it as walking stability criterion to control humanoid robot achieve stable walking.Firstly, by reviewing and summarizing the history and development status of humanoid robot, this paper compares the foot sensor system of humanoid robot designed by different research institutions, and explains the research background, tasks and application prospects, at last gives out the frame of the paper.Secondly, by analysing the classification and comparison of the walking mode , the concept of ZMP (Zero Moment Point), stable region, stability margin and the formula and nature of ZMP is introduced. This paper analyses the relationship of ZMP and stable, then educe the ZMP stability criterion, and uses it as the stability criterion of humanoid robot. Base on the Lagrangian method, the general dynamic model of humanoid robots is upbuilded. As the research object, it is conclude that the robot can not achieve stable walking by adjusting the ankle in the absence of feedback conditions. Therefore it is important to develop foot sensor system to detect ZMP and introduce anti-force information feedback control.Thirdly, by analyzing the limitations of current humanoid robot’s control system、hardware structure, the external sensor data acquisition channel is built to link with the original control system and easy to extend. It forms an open layer information collection and processing structure, and provides a basis for the channel of foot sensor information collection and integration of multi-sensor information.Fourthly, the foot sensor system is designed and implemented, and its whole structure and various components function (including mechanical design, hardware design, software design) is detailed. Through the experiment, it is proved that this system is effective in practice.Finally, as front adjustments for example, how to use the foot sensor information in practice is introduced basing on the simplified model of humanoid robot. Under conditions of the introduction foot sensor information feedback, through simulation it is obtained the feasibility and excellent control effect of the improved genetic algorithm and sub-PD control methods.更多还原