摘要
FastSLAM算法中粒子滤波存在粒子退化问题,严重影响了滤波的精度。为解决此问题,提出了基于容积粒子滤波的FastSLAM算法。将容积卡尔曼滤波作为粒子滤波的重要性密度函数,获得所需要的加权粒子,进而通过计算粒子均值,获得系统状态的最小均方误差估计,能够很好地逼近系统状态后验概率,抑制粒子退化问题。在无人机应用环境下对该算法进行仿真验证,结果表明该算法在SLAM的估计精度和一致性方面都显示出较好的性能。
In FastSLAM,particle filter may induce particles degeneration,and affect the filtering precision.FastSLAM based on cubature particle filtering algorithm is proposed to solve this problem. The CPF uses the cubature Kalman filtering as the importance probability function to obtain the required particles with weights. By calculating the mean,then the minimum square error state estimation is obtained. This algorithm can approach the system posterior probability and restrain the particle degeneration. Simulation research is carried out by proposed algorithm on uninhabited aerial vehicle（UAV）. Result shows that this method can get better performance in estimation accuracy and consistency.
出处
《系统仿真技术》
2014年第3期234-238,共5页
System Simulation Technology
关键词
无人机
FASTSLAM
粒子滤波
容积卡尔曼滤波
uninhabited aerial vehicle
fast simultaneous location and mapping
particle filtering
cubature Kalman filtering
