An Indoor Mobile Robot Positioning Algorithm Based on Adaptive Federated Kalman Filter
Xiaobin Xu,Fenglin Pang,Yingying Ran,Yonghua Bai,Lei Zhang,Zhiying Tan,Changyun Wei,Minzhou Luo
DOI: https://doi.org/10.1109/jsen.2021.3106301
IF: 4.3
2021-01-01
IEEE Sensors Journal
Abstract:In order to avoid the problem of inaccurate positioning of single indoor sensor, an adaptive federated Kalman filter (AFKF) algorithm is proposed in this paper. According to the error covariance matrix, residual, trace and other information of each subfilter, the sharing factors of information fusion and distribution in federated Kalman filter are adaptively adjusted. And the fault detector is added to detect extreme abnormal condition. The simulation was performed and the experiment was carried out in indoor environment. The results show that the AFKF is not affected by abnormal signals of the subfilter, and the accuracy compared with other AFKF algorithms is improved by 10.76% and 10.84% in ${X}$ and ${Y}$ directions. And even if the subfilter fails the proposed AFKF can also work normally. All of these indicates the feasibility and effectiveness of the proposed AFKF.