Distributed sensor fusion using covariance intersection and particle filtering based on adaptive genetic algorithm

In this paper, a novel algorithm is proposed for target tracking with distributed sensors by combining particle filtering based on the adaptive genetic algorithm and the fast covariance intersection algorithm. The adaptive genetic algorithm is applied in the resampling process to overcome the problem of particle deprivation in the particle filtering. In each genetic particle filter, the particles with large weights are retained as a part of the offspring directly, while the other particles are hybridized or mutated with their weights and then we select the particles with larger weights as the other part of the offspring. It is different from the conventional adaptive genetic particle filtering. The distributed data fusion based on fast covariance intersection is processed on each sensor after the particle filtering. The simulation results show that the algorithm obtains more accurate estimation of the target comparing with SIR particle filtering.