Improving coordination with communication in multi-agent reinforcement learning | IEEE Conference Publication | IEEE Xplore