The invention discloses an AGV task scheduling method based on a simulated annealing algorithm, and belongs to the field of automatic warehousing system task scheduling. The method comprises the stepsof 1, determining position coordinates and endurance mileage of an AGV and position coordinates of a to-be-carried goods shelf and a sorting workstation; 2, setting parameters of a simulated annealing algorithm; 3, generating an initial task allocation scheme; 4, starting iterative search to generate candidate solutions; 5, the feasibility of the candidate solutions is checked, and if not, the step 4 is executed to regenerate the candidate solutions; 6, judging whether to accept the candidate solution or not, and if so, replacing the current solution with the candidate solution; 7, if the preset number of iterations is not reached, skipping to the step 4 to carry out the next iterative search; and 8, if the algorithm termination condition is met, outputting the current solution, otherwise, reducing the annealing temperature and skipping to the step 4 to start a new round of iterative search. An AGV task scheduling problem in an intelligent warehousing system environment can be rapidlysolved, and a task allocation scheme with the shortest or approximately shortest completion time is obtained.