Tank Behaviour Decision Based on Behaviour Tree and SA-QL

Abstract

With the wide application of artificial intelligence technology in the military field, unmanned intelligent equipment will play an increasingly important role in the future equipment combat system, and autonomous decision-making ability will directly affect the combat performance of equipment. Therefore, the research on decision-making ability of unmanned intelligent equipment has important practical significance. This paper introduces the principle of behavior tree and SA-QL algorithm, proposes a modeling method of autonomous behavior decision based on behavior tree and SA-QL, and applies it to the construction of tank behavior decision model. By constructing the tank behavior tree model, the autonomous behavior decision is realized, and the tank model has the ability of continuous learning and evolution by using Q learning. Simulation results show that the built tank model can learn new decision instructions according to environmental changes and make behavioral decisions autonomously, which also proves the effectiveness of the proposed modeling method.