基于粒子滤波的无线传感器网络辅助同步定位与地图创建方法研究  

On WSN-Aided Simultaneous Localization and Mapping Based on Particle Filtering

作　　者：李阳铭[1] 孟庆虎[1] 梁华为[1] 李帅[1] 陈万明[1] 

机构地区：[1]中国科学院合肥智能机械研究所仿生感知与控制研究中心

出　　处：《机器人》2008年第5期421-427,434,共8页Robot

基　　金：国家自然科学基金重点资助项目(60535010)

摘　　要：提出了一种新颖的无线传感器网络(WSN)辅助的移动机器人同步定位与地图创建(SLAM)方法,解决了传统SLAM方法难以解决的求解问题空间维数高和多数据关联困难两大问题.为该WSN辅助的SLAM方法建立了模型,并进行了噪声分析;在此基础上,提出一种适用本方法的分布式粒子滤波数据融合算法.着重分析了粒子初始化、预测、序贯重要性采样和重采样等关键步骤,并通过仿真实验分析验证了该方法的正确性和高效率.实验结果表明,采用粒子滤波算法,并综合无线传感器网络进行辅助导航,可以极大地降低求解问题空间维数,解决多数据关联错误问题,可以完全不依赖锚节点完成盲节点高精度定位;同时,还能够有效地提高移动机器人定位与地图创建精度,特别是在不要求机器人路径闭合的情况下可以有效抑制惯性导航的误差累计.A novel WSN-aided SLAM （simultaneous localization and mapping） method is proposed for mobile robot to solve two main problems in traditional SLAM methods, i.e., high dimensions in problem spaces and difficulties in multitarget data association. Model for the WSN-aided SLAM method is.built, and noises are analyzed. Then a distributed particle filtering （PF） data fusion algorithm suitable for this method is developed. The key steps, such as particle initialization, prediction, sequential importance sampling, resampling, are particularly analyzed, and the validity and efficiency of the method are testified by simulation experiment. The experiment results demonstrate that, when the PF algorithm is used and the WSN is integrated for aided navigation, the dimensions of problem spaces can be greatly reduced, the multi-target data association problems can be solved and blind nodes can be located with high precision independent of the anchor node. The precision of localization and mapping for mobile robots can be effectively improved, and the error accumulation of inertial navigation can be effectively suppressed especially when the robot closed-loop track is not required.

关 键 词：同步定位与地图创建 无线传感器网络 粒了滤波 移动机器人 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]