基于自主视觉的移动机器人运动规划方法研究

发布时间：2018-10-10 11:54

【摘要】：近年来,世界各国移动机器人研究迅猛发展,已开始应用于无人值守的仓库进行装配货物等复杂任务。移动机器人在无人环境中完成程序设置的任务已成为可能,得到了日益广泛的关注。而在人居环境中,要求机器人作为人的帮手,除了依靠自身设置的程序运作行动外,还需识别环境中的人,一方面为了安全避开人体;另一方面,能具备自主识别共居一室的人发出的工作指令,并“依计从事”。视觉系统作为计算机与用户之间交互的主要形式,为实现上述功能提供可能。本文针对具有可识别自然人肢体命令的自主视觉系统,并且能灵活全向运动的移动机器人开展研究。首先,建立实现了具有四麦克纳姆轮的可全方位灵活综合行进的机器人平台系统。可完成前移、后移、左移、右移、45°斜方位右移、45°斜方位左移、原地零半径顺时针回旋、原地零半径逆时针回旋的多种行进方式;其次,搭建了可识别自然人肢体语言动作的体表视觉系统,提出了识别人体姿态动作的空间向量算法,对肢体动作进行定义,达到识别肢体动作的目的;再次,搭建了基于SQL Server数据库服务器的移动机器人控制系统,通过服务器进行上、下位机之间的信息指令传输,实现实时指令传递控制功能;最后,进行了针对移动平台视觉识别效率和整体操控性能的测试。经实验验证,系统可高效识别人体表征肢体动作语言特征,操控者可通过自身动作灵活快速的控制移动机器人多方位全向运动。移动平台整体效果符合计划要求。
[Abstract]:In recent years, the research of mobile robots in the world has developed rapidly and has been applied to unattended warehouses to assemble goods and other complex tasks. It is possible for mobile robot to complete the task of program setting in unmanned environment, which has been paid more and more attention. In human settlements, robots are required to act as human helpers, in addition to their own programs, to identify the people in the environment, on the one hand, to avoid the human body, and, on the other, to avoid the human body. Able to independently identify the work instructions issued by a person living in one room, and "according to plan". As the main form of interaction between computer and user, visual system provides the possibility to realize the above functions. In this paper, an autonomous vision system with recognizable limb commands from natural persons and a mobile robot with flexible omnidirectional motion are studied. Firstly, a robot platform system with four McNum wheels is established. Can complete forward, backward, left, right, 45 掳oblique azimuth to the left, zero radius clockwise gyration, zero radius in situ counterclockwise gyration. Second, Build the body surface vision system that can recognize the body language movement of the natural person, put forward the space vector algorithm to recognize the human body posture action, define the limb movement, achieve the purpose of recognizing the limb movement; third, The mobile robot control system based on SQL Server database server is built. The information command transmission between the upper and lower computers is carried out through the server, and the real-time command transfer control function is realized. Finally, The visual recognition efficiency and the overall control performance of the mobile platform are tested. The experimental results show that the system can efficiently recognize the language features of the human body and the manipulator can control the multi-directional omnidirectional motion of the mobile robot flexibly and quickly through its own actions. The overall effect of the mobile platform meets the requirements of the plan.
【学位授予单位】：北京建筑大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】