基于PDR和蓝牙加权定位的卡尔曼滤波算法  

Kalman Filter Algorithm Based on Weighted PDR and Bluetooth Positioning

作　　者：何梦园 刘小瑜 刘正熙[1] 熊运余[1] 袁钰涵 HE Meng-yuan;LIU Xiao-yu;LIU Zheng-xi;XIONG Yun-yu;YUAN Yu-han(College of Computer Science, Sichuan University, Chengdu 610065)

机构地区：[1]四川大学计算机学院,成都610065

出　　处：《现代计算机》2019年第10期3-8,共6页Modern Computer

基　　金：国家自然科学基金(No.61471250)

摘　　要：随着智能手机的性能提高,各种传感器的性能也得到较大改善,在进行室内定位时,使用手机自带的传感器可做到行人航迹推算(PDR),但是PDR存在累计误差,需要其他定位信号对其进行修正,针对此情况,提出基于卡尔曼滤波的PDR与蓝牙定位融合定位方法。该方法有效克服惯性导航累计误差以及单点定位精度低的问题。针对蓝牙信号跳变比较大的问题,提出多信号位置加权的方法。通过在iPhone 6智能手机上测试,在行走状态下蓝牙定位平均误差为5.64m,PDR平均误差为13.44m,融合滤波后平均误差下降到1.85m,且最大误差由蓝牙定位的14.98米降低到3.6米。With the improvement of the performance of smart phones, the performance of its various sensors has been greatly improved. The pedestrian dead reckoning (PDR) can be achieved by using the sensor of mobile phones. However, the accumulated errors of PDR exist, which need to be corrected by other positioning signals. In view of this situation, proposes a positioning fusion method of PDR and Bluetooth based on Kalman filter. This method effectively overcomes the cumulative error of inertial navigation and the low accuracy of single point positioning. Aiming at the problem of large jump of Bluetooth signal, proposes a method of weighting each signal location. Through testing on the iPhone 6 smartphone, the average error of Bluetooth positioning is 5.64m and PDR is 13.44m. After Kalman filtering, the average error of positioning is reduced to 1.85m, and the maximum error is reduced from 14.98m (Bluetooth) to 3.6m.

关 键 词：室内定位 PDR 蓝牙定位 卡尔曼滤波 

分 类 号：TN925[电子电信—通信与信息系统] TN713[电子电信—信息与通信工程]