结合粒子群寻优与遗传重采样的RBPF算法

发布时间：2018-06-15 03:10

  本文选题：同时定位与地图构建 + Rao-Blackwellized粒子滤波器　； 参考：《计算机工程》2016年11期

【摘要】：针对Rao-Blackwellized粒子滤波器(RBPF)重采样过程存在粒子衰竭、提议分布精确度不高的问题,提出一种改进的RBPF算法。为提高RBPF算法提议分布精确性,在改进的算法中将机器人里程计信息和激光传感器采集的距离信息进行融合,在算法中引入粒子群寻优策略,通过粒子间能效吸引力来调整采样粒子集,同时对重采样中权值较小的粒子进行遗传变异操作,缓解粒子枯竭现象,提高机器人位姿估计一致性,并维持粒子集的多样性。在基于机器人操作系统和配有URG激光传感器的Pioneer3-DX机器人平台上对改进RBPF算法进行可靠性验证。实验结果表明,改进算法在兼顾粒子集多样性的同时能显著提高机器人位姿估计精确性。
[Abstract]:An improved RBPF algorithm is proposed to solve the problem of particle failure and low accuracy of proposed distribution in the resampling process of Rao-Blackwellized particle filter (RBPF). In order to improve the distribution accuracy of RBPF algorithm, the robot odometer information and the distance information collected by laser sensor are fused in the improved algorithm, and the particle swarm optimization strategy is introduced in the algorithm. The sampling particle set is adjusted by energy efficiency attraction among particles, and the particles with small weight in resampling are operated by genetic variation, which can alleviate the phenomenon of particle depletion, improve the consistency of robot pose estimation, and maintain the diversity of particle sets. The reliability of the improved RBPF algorithm is verified on the Pioneer3-DX robot platform based on robot operating system and with URG laser sensor. Experimental results show that the improved algorithm can significantly improve the accuracy of robot pose estimation while taking into account the diversity of particle sets.
【作者单位】： 重庆邮电大学信息无障碍工程研发中心;
【基金】：国家科技部国际合作项目(2010DFA12160) 重庆市科技攻关项目(CSTC,2010AA2055)
【分类号】：TP242;TP18
，

本文编号：2020329