The real-time strategy game represented by StarCraft is the frontier and hot spot of Artificial Intelligence (AI) research in recent years, in which micromanagement is the basic and core key problem. The micromanagement technological breakthrough has great and wide application prospects in the field of swarm intelligence game confrontation. At present, in the research of micromanagement in StarCraft, the verification scenarios proposed by various researchers for their own concerns are scattered and independent, and lack of a unified training and testing scenario testbed. In this regard, we propose a unified testbed named micromaps. According to reasonable principles, we systematically and comprehensively design a variety of homogeneous and heterogeneous micromanagement scenarios, and support the comprehensive comparison of AI algorithms in many aspects, such as winning rate, generalization and convergence speed. Finally, the future research direction of StarCraft 2 micromanagement is prospected.