基于粒子群优化的移动机器人SLAM方法  

SLAM Method for Mobile Robot Based on Particle Swarm Optimization

作　　者：马家辰[1,2] 张琦[1] 谢玮 马立勇 

机构地区：[1]哈尔滨工业大学航天学院,黑龙江哈尔滨150001 [2]哈尔滨工业大学(威海)信息与电气工程学院,山东威海264209

出　　处：《北京理工大学学报》2013年第11期1151-1154,1202,共5页Transactions of Beijing Institute of Technology

基　　金：山东省自然科学基金资助项目(ZR2011FM005)

摘　　要：针对传统Rao-Blackwellized粒子滤波器存在的粒子消耗问题,提出了一种基于粒子群优化的移动机器人同步定位与制图方法.该方法在粒子重采样过程中利用粒子群优化算法获得机器人位姿的建议分布,并引入遗传算法中的交叉和变异操作对求得的粒子集进一步优化、调整.改进后的粒子分布保持了粒子的多样性,有效提高了机器人位姿估计的一致性.仿真结果表明,本文提出的方法与传统Rao-Blackwellized粒子滤波器相比,能有效解决粒子耗尽问题,使机器人获得更精准的定位和更准确的地图,具有可行性、实用性.To solve the particles degeneracy phenomenon of the Rao-Blackwellized particle filter（RBPF）,an improved method of RBPF based on particle swarm optimization（PSO）is presented to solve simultaneous localization and mapping（SLAM）problem of mobile robot.During the particle re-sampling process,the proposed distribution of mobile robot＇s pose is acquired by PSO.The crossover and mutation operation of genetic algorithm is applied to optimize and adjust the obtained particle sets.The new distribution of particles maintains the diversity of the particles and improves the consistency of robot＇s pose estimation effectively.Experiment of mobile robot in a particular environment has been implemented.The simulation results demonstrate that the presented approach solves the problem of particles degeneracy effectively,improves the accuracy of SLAM and has the characteristics of feasibility and availability.

关 键 词：移动机器人 同步定位与制图 RAO-BLACKWELLIZED粒子滤波器 粒子群优化 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]