移动机器人同时定位与地图构建方法研究

Research on the method of simultaneous localization and map building for mobile robots

随着当前智能领域的不断发展,移动机器人成为当前研究和应用的重点,特别是在机器人定位和地图构建方面。但传统的基于粒子滤波的机器人定位和地图构建中,存在粒子群的权值会随着粒子数量的增加而逐步往权重高的方向搜索,最终导致粒子群种群变少,不利于定位和地图构建的问题,提出一种基于改进粒子滤波算法的机器人移动定位与地图构建方法。对此,文章首先对移动机器人的基本结构进行阐述,构建移动机器人的坐标系模型和运动模型;然后对传统的粒子滤波算法的基本原理和流程进行分析,并指出其存在的问题;结合上述的分析,提出首先采用K均值聚类完成对粒子移动的优化,让粒子往最优位姿靠近,以提高种群的种类,利用自适应的重采样方法,通过设定阈值的方式,完成对权重高或低粒子的重采样,以防止粒子退化。最后运用MATLAB软件对上述的方案进行验证,得出本文改进的算法与传统的EFK算法的误差要小,与原始路径更融合,由此验证了本改进方案的可行性。

With the continuous development of intelligent field,mobile robot has become the focus of current research and application,especially in robot location and map building.But based on the traditional particle filter localization and map building,existing particle swarm weight will increase with the number of particles gradually toward the direction of high weight search,eventually leading to particle swarm population less,is not conducive to the localization and mapping problem,a method based on mobile robot localization and map building method particle filter algorithm.In this regard,firstly,the basic structure of mobile robot is expounded,coordinate model and motion model of the mobile robot; then the basic principle and process of particle filter algorithm based on the analysis,and points out the existing problems; combined with the above analysis,first proposed by K means clustering complete optimization of the particle movement.Let the particle to pose near optimal,in order to improve the population type,using adaptive resampling method,by setting the threshold,complete the sampling of high or low weight particles,to prevent particle degradation.Finally,MATLAB software is used to verify the above scheme.It is concluded that the error between the improved algorithm and the traditional EFK algorithm is less,and it is more integrated with the original path,which verifies the feasibility of the improved scheme.