| 摘要: |
| 针对视觉惯性里程计(Visual-Inertial Odometry,VIO)在特征高度重复的场合易产生较大误差以及Wi-Fi指纹精确度不高等问题,提出了一种基于VIO和Wi-Fi指纹技术的室内定位方法。该系统运用VIO和Wi-Fi指纹在系统层面的结合,利用Wi-Fi指纹的无漂移、成本低的特点和VIO在一定范围内的高精确度，先进行Wi-Fi指纹粗定位，后进行VIO精定位，将大面积切割成小面积从而有效提高系统室内定位精确度，降低误差。该系统能在精定位的同时进行对指纹数据库的更新。实验结果表明,该系统的定位误差小于单独使用VIO或Wi-Fi指纹的系统,平均误差达0.15 m，能够有效提高定位精度。 |
| 关键词: 室内定位 VIO技术 Wi-Fi指纹 多状态约束卡尔曼滤波 |
| DOI： |
|
| 基金项目:广东省科技计划项目（2016A010101009） |
|
| Design of an indoor localization system based on visual-inertial odometry and Wi-Fi fingerprint |
| WEN Zoutao,FENG Suili |
| (School of Electronic and Information Engineering,South China University of Technology,Guangzhou 510640,China) |
| Abstract: |
| Aiming at the problems that in the situations with highly repeated features,visual-inertial odometry(VIO) can easily generate larger error and the accuracy of Wi-Fi fingerprint is not satisfying,a novel system based on VIO and Wi-Fi fingerprint is proposed.Combining VIO and optimized Wi-Fi fingerprint in system level,this system takes advantages of low price of Wi-Fi fingerprint and high accuracy of VIO in certain range.In order to improve the accuracy,it processes coarse location using Wi-Fi fingerprint and then fine location using VIO so as to divide a large area into small ones.The system can also update the Wi-Fi fingerprint database when doing the fine location.Experiment results show that the proposed system has a lower localization error(mean error to 0.15 m) compared with that using VIO or Wi-Fi fingerprint merely.The performance of comprehensive localization can improve user experience in practice. |
| Key words: indoor localization visual-inertial odometry(VIO) Wi-Fi fingerprint multi-state constraint Kalman filter(MSCKF) |
