A New Adaptive Federated Kalman Filter for the Multi-sensor Integrated Navigation System of MAVs
Ge-Hang Lei, Zhejun Fang, Bing Luo, Pei-Feng Qi · 2018
This paper proposes a novel adaptive federated Kalman filter (AFKF) with a new matrix-form information distribution coefficient for enhancing the accuracy and reliability of multi-sensor data fusion. The key innovation of the AFKF is that two new vector-form indicators, called the bias and the amplitude, are presented and they are both used to compute the matrix-form information distribution coefficient. The bias and the amplitude evaluate the deviation and oscillation magnitudes of each separate state component of local filters, respectively. On this basis the matrix-form coefficient has the adaptive ability to assign different weights to different state components in data fusion. A multi-sensor integrated navigation model developed for the commercial micro aerial vehicles (MAVs) is constructed based on the proposed AFKF. Experimental results show the proposed AFKF is effective in the accuracy improvement, and its anti-jamming performance is superior to the traditional fixed-coefficient federated filter.