A study on hierarchical modular reinforcement learning for multi-agent pursuit problem based on relative coordinate states | IEEE Conference Publication | IEEE Xplore