Multi-Sensor Integrated Navigation Algorithm Using Adaptive Federated Kalman Filter for MAVs

In this paper, a novel adaptive information-distributing algorithm is proposed for the federated Kalman filter (FKF) with application to the multi-sensor integrated navigation system of commercial micro aerial vehicles (MAVs). A new matrix-form information distribution coefficient is constructed based on two new vector-form indicators, i.e., bias and amplitude. The bias indicator evaluates the deviation of the local filters from the global optimal estimations. The amplitude indicator is closely related with the oscillation magnitude of local filters' state estimations, which is caused by either the external noise or the sensor's faults. Based on the two vector-form indicators, the information distribution coefficients are adjusted dynamically according to the different characteristics of the separate state estimation in each local filter. Then an adaptive federated Kalman filter (AFKF) model integrating the inertial navigation system (INS) and other sensors is established. Simulation results demonstrate that the proposed method improves the precision and disturbance rejection performance of the FKF effectively.