Scheduling method, device and system of mobile robot

The invention provides a scheduling method, device and system for a mobile robot, and the method comprises the steps: inputting a current scheduling state into a scheduling strategy neural network for analysis, and obtaining a plurality of scheduling strategies; determining a plurality of target scheduling states according to the plurality of scheduling strategies; inputting the plurality of target scheduling states into a state evaluation neural network for analysis to obtain a plurality of time costs, and performing Monte Carlo tree search according to the current scheduling state, the plurality of target scheduling states and the plurality of time costs to obtain a plurality of node values; determining the scheduling strategy corresponding to the child node with the maximum node value as the optimal scheduling strategy; all the mobile robots are controlled to execute the optimal scheduling strategy; and the steps are repeated at least once in sequence until all the mobile robots complete the corresponding