Multi-agent Hybrid Architecture Design for Naval Warfare Game
摘要
Aiming at the needs of multi-agent confrontation games in the complex wide-area environment at sea, this paper proposes a multi-agent construction idea driven by knowledge rules and data learning, and designs a hybrid-driven multi-agent software framework. The overall software framework is modularized based on knowledge rules, and core action decisions are constructed based on data learning methods. In the end, a multi-agent instance was built, and the counterparty was built for confrontation verification in the sea scene.