External Information Aided Urban Target Tracking with UAVs

This paper addresses the challenging problem of airborne target tracking in complex urban contexts by incorporating external information. Both extrinsic and intrinsic elements are mathematically formulated to constrain the motion of the target. A Bayesian method, aided by a data-driven pattern matching mechanism, is proposed for recognizing the target's behavior and allocating appropriate constraints to the ground target. A quadratic programming problem is formulated to integrate the constraint relations into the target tracking process. The target's distribution is obtained using Generalized Covariance Intersection (GCI) to fuse the state distribution calculated from each local tracker. The proposed method's advantages are analyzed and validated through numerical simulations.