This paper mainly studies a CPS-based multi-sorting mobile robot(multi-SMR) scheduling strategy. First, a novel time cost is added to the topology map model, realizing an accurate estimation of the uneven proportion of cargos and the congestion status of the robots. Then, an improved path planning algorithm is proposed, and the destination distance and time cost indicators are added to the path evaluation process. Finally, the complete scheduling process is deployed in the CPS model in a hierarchical structure, including time cost update and traffic control at the control layer, and distributed path planning and robot status update at the physical layer. Simulation experiments verify the reliability of the proposed method. Results show that the improved scheduling strategy further improves the system sorting efficiency, reduces the calculation cost, and solves the problem of robot congestion.