Non-cooperative space multi-target passive tracking for multi-spacecraft with field-of-view constraint

This article addresses the critical problem of multi-target tracking for distributed spaceborne bearing-only sensors with limited fields of view (FoV). To overcome the issues of trajectory loss in partially overlapping FoV, we propose an innovative FoV mapping partition fusion method for the labeled multi-Bernoulli filter. The labeled Bernoulli component (LBC) is partitioned via a dynamic mapping criterion that incorporates the position, attitude and detection range of sensors. This strategic partitioning allows LBCs to participate in fusion only within local partitions, rather than across the entire sensor network, which effectively prevents information degradation commonly caused by global consensus weighting. To further enhance robustness against clutter and improve computational efficiency, a partial consensus mechanism is extended to the LMB filter. Furthermore, a measurement-driven adaptive birth model suitable for bearing-only FoV is employed to reduce dependence on initial values and improve tracking performance for newborn targets. The proposed methodologies are rigorously validated via a high-fidelity Hardware-in-the-loop (HIL) system. Simulation results validate the efficacy of the proposed approach, demonstrating significant improvements in tracking robustness and accuracy for distributed spaceborne tracking systems operating in challenging scenarios.