Research on the Unmanned Tank Cooperative Maneuvering Strategy Based on Deep Reinforcement Learning

This paper proposes a cooperative maneuvering strategy for unmanned tanks to address the problem of planning movement paths and firing positions for effective target engagement in tank confrontations. Firstly, a firing-angle-driven strike effectiveness model is constructed by the De Marre formula. Secondly, the deep reinforcement learning algorithm Multi-Agent Deep Deterministic Policy Gradient (MADDPG) is employed to design the cooperative maneuvering strategy. To address the limitations of traditional MADDPG-such as insufficient exploration, local optima, and sparse rewards - an intrinsic curiosity module (ICM) is introduced. Finally, comparative simulation experiments in typical combat scenarios validate the feasibility of the proposed algorithm and its superior performance in completing maneuver-based combat tasks efficiently.