A potential game approach to multiple UAV cooperative search and surveillance

Abstract

In this paper, we developed a game theoretic formulation for multiple unmanned aerial vehicle (UAV) cooperative search and surveillance. The cooperative search problem is decomposed into three sequential tasks: coordinated motion, sensor observation, and cooperative information fusion. Firstly, the coordinated motion is designed as a multi-player potential game with constrained action sets. Then the binary log-linear learning is adopted to perform motion control, which guarantees optimal coverage. Then a consensus based fusion algorithm is introduced to construct the probability map to guide the following coordinated motion. Finally, simulations are performed to validate the effectiveness of our proposed approach. The modular framework enables the separate design of utility functions and learning algorithms, which offers a flexible way to accommodate different global objectives and underlying physical constraints.