Design and Accuracy Improvement of Distributed Data Fusion Algorithms for UAV Swarms

Abstract

In the case of collaborative perception in UAV swarms with dynamic topology, communication constraint, there are challenges in accuracy and consistency for data fusion. A hierarchical fusion framework is designed in this paper. The bottom layer reaches intra-cluster consensus update using improved diffused Titin Kalman filter with dynamic weights based on the communication quality as well as on the historical errors predictive for adaptive weighting fusion. The top layer uses virtual pheromone diffusion mechanism to direct information of high confidence between the clusters. Experimental results show that the proposed algorithm is 0.68 m to 1.08 m accurate, convergence time is less and topology changing speed is fast, the algorithm’s performance is stable, so as to greatly improve the estimating accuracy and collaborative consistency of the swarm under complicated environment.