Triangular Topology Sequence-Based Multi-Target Association for Aerial-Ground Unmanned Systems

Abstract

The heterogeneous nature of aerial-ground unmanned systems has a considerable advantage in terms of full-area awareness. However, it also presents a great challenge in the multi-target association domain. The almost orthogonal observation view results in large parallax of the aerial-ground sensors large, rendering the trajectory association and visual feature association methods inapplicable. To address this problem, this paper proposes a multi-objective association method based on triangular topology from the relative position relationship of targets. First, the algorithm extracts the distance and azimuth of the target relative to the Unmanned Ground Vehicle based on its position in the sensor and constructs a triangular topology. Then, a bicircle matching model satisfying the triangular topological sequence association constraints is proposed to transform the constrained structure matching problem into an unconstrained one. Finally, an association algorithm based on the similarity association matrix is proposed to complete the association of topologies immediately by obtaining the “optimal path” in the matrix. Simulation experiments show that the correlation accuracy is above 80% when the interference target is less than 20%, and above 90% when the observation error is less than 2.0[Formula: see text]m. The overall performance of this method outperforms similar algorithms in comparative tests.