量子粒子群优化下的RBPF-SLAM算法研究  

Research on RBPF-SLAM algorithm based on quantum-behaved particle swarm optimization

作　　者：伍永健[1] 陈跃东[1] 陈孟元[1] WU Yongjian;CHEN Yuedong;CHEN Mengyuan(Anhui Key Laboratory of Electric Drive and Control,Anhui Polytechnic University,Wuhu 241000,China)

机构地区：[1]安徽工程大学安徽省电气传动与控制重点实验室,安徽芜湖241000

出　　处：《智能系统学报》2018年第5期829-835,共7页CAAI Transactions on Intelligent Systems

基　　金：2016年度安徽高校自然科学项目(KJ2016A794);2016年安徽工程大学研究生实践与创新基金项目(Y040116004)

摘　　要：为了解决传统Rao-Blackwellized粒子滤波(RBPF)存在提议分布精度不高以及重采样过程出现的粒子退化和多样性丢失问题,提出一种量子粒子群(QPSO)优化下的Rao-Blackwellized粒子滤波同时定位与地图构建(RBPF-SLAM)算法。将机器人运动模型和观测模型融合作为混合提议分布,提高提议分布的精度;在重采样过程中引入量子粒子群优化算法更新粒子位姿,根据权值划分粒子种类,引入自适应交叉变异操作,对所得粒子集进行优化、调整,有效地防止粒子退化以及保持粒子的多样性。利用本文算法不仅用MATLAB进行仿真实验,而且结合了旅行家2号移动机器人在机器人操作系统(ROS)上进行实际验证。结果表明,本文算法能以较少粒子数精确估计出机器人的位姿和高精度的地图,误差和运行时间也大大降低了。The traditional Rao-Blackwellized particle filter(RBPF)is associated with a low distribution accuracy as well as particle degeneracy and loss of diversity during resampling.To solve these problems,a combination of RBPF and simultaneous localization and mapping(RBPF-SLAM)algorithm based on quantum-behaved particle swarm optimization(QPSO)is proposed.A fusion of robot motion model and observation model is proposed as a hybrid distribution to improve accuracy.The QPSO algorithm updates the pose of particles in the resampling process according to the weight measurement of particle type,and an adaptive crossover and mutation operation is introduced to optimize and adjust the particle set to effectively prevent particle degradation and maintain particle diversity.To verify the effectiveness of the improved algorithm,a simulation experiment is performed on MATLAB,as well as a Voyager-II mobile robot in a robot operating system(ROS).The results show that the proposed algorithm can accurately estimate the position and pose of the robot and a high precision map,and error and running time are also greatly reduced.

关 键 词：RAO-BLACKWELLIZED粒子滤波 同时定位与地图构建 提议分布 量子粒子群优化 交叉变异 移动机器人 机器人操作系统 

分 类 号：TP21[自动化与计算机技术—检测技术与自动化装置] TP24[自动化与计算机技术—控制科学与工程]