This paper introduces a multi-agent coordinated maneuvering behavior decision-making method based on the MADDPG algorithm, for efficient collaborative navigation in urban battlefields. This method optimizes agent strategies based on the MADDPG algorithm, taking into account environmental factors such as traffic obstacles, topography, weather conditions, and damage levels. By defining a multi-agent state model and objective function, it achieves collaborative decision-making in terms of time efficiency, path optimization, and collision avoidance. The experiments were conducted based on open-source data from OpenStreetMap, simulating a scenario where three agents cooperate in reconnaissance. The results show that this method can rapidly improve the coordination efficiency of agents and achieve stable performance after about 30,000 iterations. The agents can dynamically adjust their paths, avoid obstacles, or select the best routes, demonstrating effective collaborative planning capabilities under non-strict obstacle avoidance conditions. In summary, the multi-agent coordinated maneuvering behavior decision-making framework proposed in this paper provides an effective solution to the problem of collaborative navigation in complex environments, significantly enhancing the collaboration effect between agents.