Correlation-Driven Task Assignment for Multi-UAV Networks via Imitation Learning

In this paper, a cooperative task assignment problem is addressed in networks of multiple Unmanned Aerial Vehicles (UAVs), where UAVs have limited resources and energy constraints. Tasks are distributed across various Points of Interest (PoIs) and the completion of tasks requires the collaboration of multiple UAVs with specific resource capabilities. To optimize task assignment and improve UAV operational efficiency, a correlation framework is introduced that quantifies the compatibility between UAVs and task requirements. A distributed task-to-UAV assignment is then modeled as a coalition formation game (CFG), where UAVs form coalitions according to their capabilities and task attributes autonomously. A novel coalition formation algorithm is proposed that incorporates behavioral imitation learning, enhancing convergence speed and adaptability in dynamic environments. The algorithm ensures joint optimization of energy consumption and task completion by enabling the UAVs to make efficient decisions in finite iterations. Simulation results demonstrate the effectiveness of the proposed method in terms of energy consumption and time cost for task completion.