Abstract: In order to make full use of redundant data of multiple sensors to achieve high-precision tracking, an redundant data adaptive federated Kalman filter algorithm with outlier detection is proposed based on redundant measurement data. First, in the information distribution stage, an adaptive information sharing factor is designed for redundant information, which improves the information distribution efficiency. Secondly, in the information fusion stage, in order to reduce the influence of error data on tracking results, an outlier detection algorithm is proposed, which combines the judgment results of all filters through D-S evidence theory to evaluate whether the data is outlier data. Finally, the linear least square method is used to fuse and obtain a more accurate final estimation result. The simulation results show the proposed algorithm has better tracking accuracy and robustness than the existing models.  

Key words: federated filtering, collaborative tracking, D-S evidence theory, outlier detection