基于SVR和粒子滤波的室内导航方法

Indoor Navigation Method Based on SVR and Particle Filter

摘要:

针对室内环境复杂性造成的室内定位精度不足问题,提出一种基于支持向量机回归和粒子滤波的室内导航方法；离线阶段通过采集室内接收信号强度并利用统计学习方法,构建室内RSS与物理位置之间的支持向量回归映射模型,定位阶段使用智能移动设备采集加速度、方向角等运动状态信息和Wifi模块感知的环境信息,并利用粒子滤波将运动数据和回归结果进行融合处理,推算移动用户运动轨迹；室内实验结果表明,本方法最大定位误差为1.891 m,平均误差为0.669 m,有效地提高了室内定位导航精度。

Abstract:

Aiming at the lack of indoor positioning accuracy caused by the complex indoor environment, a method of indoor navigation based on support vector regression (SVR) and particle filter is proposed. At the offline stage, the indoor received signal strength (RSS) is collected, and statistical learning methods are used to construct support vector regression model indoor mapping between received signal strength and physical location. At positioning stage, intelligent mobile equipment is used to capture acceleration, direction angle and environmental information perceived by WIFI module. Based on the particle filter, the motion data and the regression result are fused to calculate the motion trajectory of the mobile user. Experimental results show that the maximum location error of proposed method is 1.891 m, while the average error is 0.669 m, and indoor location and navigation accuracy can be effectively improved.