室内全向移动机器人系统设计及导航方法研究

发布时间：2018-03-29 12:10

本文选题：全方位移动机器人　切入点：语义拓扑地图　出处：《哈尔滨工业大学》2017年硕士论文

【摘要】：21世纪以来,中国逐渐步入老龄化社会,出现“空巢老人”、老人无人陪护等社会问题。为家庭服务机器人行业提出巨大的市场需求,与此同时也对机器人提出了巨大的挑战。室内环境移动作业一直以来都是移动机器人研究的重点问题。由于机器人对于环境的理解局限于数字坐标信息,而人对于环境的理解处于语义区域层面,因此语义地图构建及语义导航成为移动机器人领域的研究热点。室内环境属于动态环境,因此动态环境导航也是一直以来研究的焦点。首先,设计基于Mecanum轮的全方位移动机器人系统。针对室内环境狭窄通过性差的特点,设计基于Mecanum轮的机器人构型,其中独立悬挂结构可增强机器人不平路面通过性及避免打滑;设计具有良好可替换性的伺服驱动模块,为机器人运动提供动力;通过激光和视觉传感器,建立立体的环境感知模型;软件系统基于ROS操作系统,可实现节点的弱耦合、多线程运行,提高程序运行效率。然后,建立机器人系统模型及环境地图模型。针对本文全方位移动机器人构型开展机器人运动控制以及位姿估计工作,分别建立运动控制模型和里程计模型;针对本文的主要传感器激光传感器进行建模修正及坐标系变换;面向语义导航需求,分别构建语义拓扑层地图和2D栅格层地图。其次,研究了机器人定位及语义导航问题。针对机器人定位,采用语义约束对AMCL定位算法的粒子分布进行指导,以此来加快机器人定位收敛速度,并减少误匹配现象;语义拓扑层路径规划采用基于连通区域的搜索方法,可以加快搜索的速度,并且增强机器人的狭窄环境通过性;基于行人预测模型,采用基于A*的二次规划方法进行动态环境机器人局部路径规划。最后,搭建基于Mecanum轮全方位移动机器人实验平台,分别就机器人工作稳定性、机器人室内环境定位、机器人语义导航和动态环境导航开展实验研究,验证实验平台的工作可靠性和上述算法的可行性。
[Abstract]:Since the 21st century, China has gradually stepped into an aging society, with social problems such as "empty nest old people" and the elderly being left unattended. It has put forward a huge market demand for the family service robot industry. At the same time, it also poses a great challenge to the robot. Indoor environment mobile operation has always been the focus of mobile robot research. Because the robot's understanding of the environment is limited to the digital coordinate information, Human understanding of the environment is at the level of semantic region, so semantic map construction and semantic navigation become the research hotspot in the field of mobile robot. Indoor environment belongs to dynamic environment. Therefore, dynamic environment navigation has always been the focus of research. Firstly, an omni-directional mobile robot system based on Mecanum wheel is designed. Aiming at the characteristics of narrow indoor environment, the robot configuration based on Mecanum wheel is designed. Among them, independent suspension structure can enhance the mobility of robot uneven road surface and avoid skid; design servo drive module with good substitutability to provide power for robot motion; through laser and vision sensor, The software system is based on ROS operating system, which can realize the weak coupling of nodes, multi-thread running, and improve the efficiency of the program. The robot system model and the environment map model are established. The motion control model and the odometer model are established respectively for the robot motion control and pose estimation for the omnidirectional mobile robot configuration in this paper. For the main sensor laser sensor modeling correction and coordinate system transformation, the semantic topology layer map and 2D grid layer map are constructed to meet the needs of semantic navigation. Secondly, The problem of robot localization and semantic navigation is studied. The particle distribution of AMCL localization algorithm is guided by semantic constraints in order to accelerate the convergence speed of robot localization and reduce the mismatch phenomenon. The path planning of semantic topology layer adopts the search method based on connected region, which can accelerate the speed of searching and enhance the passability of the narrow environment of robot, and based on the pedestrian prediction model, the path planning of semantic topology layer can improve the speed of searching. The local path planning of dynamic environment robot is carried out by using the quadratic planning method based on A *. Finally, the experimental platform of omnidirectional mobile robot based on Mecanum wheel is built to locate the stability of the robot and the indoor environment of the robot, respectively. The experimental research on robot semantic navigation and dynamic environment navigation is carried out to verify the reliability of the experimental platform and the feasibility of the above algorithms.
【学位授予单位】：哈尔滨工业大学
【学位级别】：硕士
【学位授予年份】：2017
【分类号】：TP242

【参考文献】