基于粒子群优化的同步定位与地图构建  

Simultaneous localization and map construction based on particle swarm optimization

作　　者：曲丽萍[1,2] 王宏健[1] 曲永印[2] 

机构地区：[1]哈尔滨工程大学自动化学院,哈尔滨150001 [2]北华大学电气信息工程学院,吉林吉林132021

出　　处：《沈阳工业大学学报》2013年第4期419-425,共7页Journal of Shenyang University of Technology

基　　金：国家自然科学基金资助项目(50979017);教育部高等学校博士学科点专项科研基金项目(20092304110008)

摘　　要：针对基本FastSLAM算法的样本枯竭、估计精度下降等问题,提出了一种基于多样性启发因子的粒子群优化FastSLAM算法.利用粒子群搜索寻优重新分配粒子,使粒子的表示更加接近于真实的后验概率分布,并且采用粒子集多样性测度作为启发因子,引导粒子优化搜索过程,确保群体多样性水平最优,减轻粒子退化现象,驱动粒子集向后验概率较高的区域运动.对所提出的算法进行了仿真实验,验证了算法的可行性和有效性.仿真结果表明,该算法能够改进样本枯竭问题,并能够获得较高的定位精度、地图构建精度及较好的滤波估计稳定性.In order to solve such problems as the impoverishment of sample and decline of estimation precision for basic FastSLAM algorithm, a particle swarm optimization FastSLAM algorithm based on diversity heuristic factor was proposed. The optimum reassigning particles were searched with the particle swarm to make the particle representation more close to the real posterior probability distribution. In addition, the diversity measurement of particle set was taken as the heuristic factor, which could lead the optimizing and searching process of particles, ensure that the swarm diversity level was optimum and reduce the particle degeneration. Meanwhile, the particle set was driven to move towards the area where the posterior probability distribution was higher. The simulation experiment was performed for the proposed algorithm, and the feasibility and effectiveness of the algorithm were validated. The simulated results show that the proposed algorithm can avoid the impoverishment of sample, obtain the higher precision of both localization and map construction, and achieve the better filtering estimation stability.

关 键 词：同步定位与地图构建 粒子滤波器 粒子退化 样本枯竭 粒子群优化 多样性启发因子 有效样本大小 自适应重采样 

分 类 号：TP242.6[自动化与计算机技术—检测技术与自动化装置]