Distributed joint sensor registration and resolvable-group target tracking based on hypergraph matching theory
Guchong Li, Gang Li, You He · IET conference proceedings. · 2024
The paper addresses the problem of distributed resolvable-group target tracking (RTT) over a peer-to-peer sensor network with different coordinates. In the local filtering stage, each sensor node employs the labeled multi-Bernoulli (LMB) filter. In the fusion stage, the presence of coordinate system differences, also known as drift, hinders direct fusion and can lead to diverging tracking performance. To address this issue, we propose a scheme called the distributed joint sensor registration a nd RTT (DJSR-RTT). Specifically, we first perform state estimates and then minimize the Wasserstein distance of different RFSs to find the matched targets, which is a prerequisite for estimating the drifts. To effectively estimate the drifts, we combine the hypergraph matching (HM) theory with Hungarian assignment. Lastly, the pairwise fusion under the generalized covariance intersection (GCI) rule is performed based on the calibrated coordinates. Simulation experiments are provided to validate the effectiveness of the proposed approach.