Distributed Multi-target Tracking by Multi-AUV Based on PMB Filter in Underwater

This paper focuses on the problem of multi-target tracking (MTT) by multiple Autonomous underwater vehicles (AUVs). It presents a distributed information fusion method to improve the robustness and performance of AUVs in MTT. In this method, each AUV locally obtains the posterior probability density function (pdf) of multi-target using local measurements based on the Poisson multi-Bernoulli (PMB) filter. In addition, a target-wise fusion rule based on generalized covariance intersection is proposed, allowing the locally obtained multi-target pdf to be shared between AUVs and fused effectively. The simulation results show that the proposed method performs well for underwater MTT by multi-AUV.