Deep Reinforcement Learning Based on Multi-head Attention for the Multi-Objective Task Allocation in Fire Fighting

With the expansion of human activity areas, the threat of forest fires has greatly increased. Therefore, the quality and decision-making speed of fire-fighting programs, as well as the utilization rate of fire-fighting resources, directly determine the final fire-fighting effect. Aimed to solve the task allocation in forest fire-fighting, this paper proposes a multi-objective optimization model with an unfixed number of vehicles, and develops a deep reinforcement learning (DRL) method based on multi-head attention to solve the problem. Experimental results show that, compared with the Genetic Algorithm (GA) and Particle Swarm Optimization (PSO), the allocation scheme obtained by the proposed method incurs lower costs and requires less computation time.