Multi-Agents Interaction Approach based on Graph Network and Reinforcement Learning | IEEE Conference Publication | IEEE Xplore