基于IGA算法的FastSLAM算法研究  

Research On FastSLAM Algorithm Based On IGA Algorithm

作　　者：刘坤坤[1] 

机构地区：[1]上海理工大学光电信息与计算机工程学院

出　　处：《软件导刊》2017年第9期55-60,共6页Software Guide

摘　　要：为了使自主移动机器人在SLAM（同步定位和地图创建）上更加准确,分析了粒子滤波器（Particle Filter,PF）的FastSlam算法在粒子退化和粒子早熟两方面的不足,提出了一种改进算法（IGA算法）。该算法通过替代原有的重采样过程,改善了粒子多样性,提高了预测精度。在粒子早熟方面采用模拟退火思想对遗传算子进行改进,避免了遗传算法中的遗传算子易陷入局部最优解产生“早熟”现象问题。仿真结果表明,IGA算法使粒子保持的多样性更加持久,算法精度持续时间更长。In order to make the autonomous mobile robot more accurate in SLAM, an improved algorithm （IGA algorithm） is proposed based on the shortcomings of Particle Filter （PF） FastS/am algorithm in particle degradation and early maturing of par- ticles. The algorithm is used to replace the original resampling process to improve the particle diversity and improve the prediction accuracy. The genetic algorithm is improved by using simulated annealing in the early maturing of particles, which avoids the genetic algorithm in the genetic algorithm which is easy to fall into the local optimal solution and produce ＂premature＂ phe- nomenon. The simulation results show that the IGA algorithm makes the diversity of particles more durable and the algorithm has longer duration.

关 键 词：SLAM Fast SLAM IGA遗传算法粒子滤波 粒子退化 粒子早熟 

分 类 号：TP312[自动化与计算机技术—计算机软件与理论]