一种基于粒子滤波的月球车同时定位与地图创建方法  

A Simultaneous Localization and Map-Building Method for Lunar Rover Based on Particle Filter

作　　者：李秀智[1] 居鹤华[1] 

机构地区：[1]北京工业大学电子信息与控制工程学院,北京100124

出　　处：《宇航学报》2009年第5期1891-1895,共5页Journal of Astronautics

基　　金：国家863计划资助项目(2006AA12Z307)

摘　　要：针对月球探测车位置与姿态的准确估计问题,提出了一种基于RBPF粒子滤波的月球车SLAM方法。该方法基于高速激光雷达对环境地貌的观测,通过激光测距值与所建栅格地图之间的匹配实现。鉴于车体的位置估计仍难以彻底消除累积误差,提出了对于月球车平动速度的估计方法,并通过仿真实验验证了该速度估计方法的正确性。为提高粒子滤波的实时性能,提出所有粒子共同维护一幅地图的处理方法,实验表明该方法在明显提高SLAM实时性的同时,并未对滤波效果产生明显影响。For an accurate position and orientation estimation of Lunar Rover, a method of SLAM based on Rao- Blackwellised Particle Filter is introduced in this article. In this approach environment measurements obtained by high speed Laser Range Finder is aligned with the constructed grid map. As the position estimation by SLAM based on particle filter is hard to eliminate the error accumulated in traveling, we proposed an estimation method for translation velocity of Lunar Rover, which is validated by simulation experiments. To improve real-time performance, we propose that all the particles maintain a common map. Simulation experiment reveals that this map management method does not degenerate filter effect while dramatically enhancing its real-time performance.

关 键 词：月球车 SLAM 粒子滤波 定位 栅格地图 激光雷达 

分 类 号：TP24[自动化与计算机技术—检测技术与自动化装置]