Observation-Driven Multi-UAV Distribution for Localization and Tracking of a Noncooperative Ground Moving Target

This article considers the problem of localization and tracking enhancement for a noncooperative ground moving target by multiple uncrewed aerial vehicles (UAVs). Based on the Cramér-Rao lower bound and Fisher information matrix, an observation metric characterizing the upper bound of target state estimation is derived in the multi-UAV scenario for the first time. To optimize this metric, an observation-driven guiding vector field is proposed, which continuously drives the UAVs toward respective locally optimal spatial distributions while satisfying formation constraints. Compared with previous studies, the cooperative multi-UAV observation approach eliminates the need for high maneuverability required in single-UAV observation, effectively leveraging the redundancy among multiple UAVs to achieve more stable state estimation of the target through adaptive spatial repositioning. The proposed approach is validated through simulation of multiple quadrotor UAVs performing localization and tracking task for ground moving target. The result demonstrates that, through the designed guiding vector field, the variance of the state estimation error is reduced, thereby significantly improving the quality of target localization and tracking performance.