Role-Based Collaboration with E-CARGO Modeling: A Reinforcement Learning Framework for Enhanced Coordination in Intelligent Games
Yuanbai Li, Yinqi Quan, Xinyuan Wan, Ziyuan HUANG, Yuxiang Sun, Xianzhong Zhou, Liang Geng · 2025
Researches in management often involves complex decision making environments. Intelligent decision making games, crucial for simulating such environments, assist decision makers in strategy evaluation and resource allocation. However, traditional decision making games based on deep reinforcement learning (DRL) face challenges such as prolonged training periods, slow convergence, and multi-agent coordination difficulties. To overcome these issues, an enhanced framework integrating Role-Based Collaboration (RBC) and the E-CARGO model (Environment, Class, Agent, Role, Group, Object) has been proposed. This framework assigns distinct roles to agents, followed by training through reinforcement learning. Simulation experiments on the Winning-first platform demonstrate that this method surpasses traditional reinforcement learning in terms of convergence and intelligence, addressing the aforementioned challenges and enhancing agent intelligence.