To improve the operation efficiency of automatic storage and retrieval system (AS/RS) at air freight station in term of job scheduling, on the basis of analyzing the effect of double unit load device (ULD) transport combination and anti-collision avoidance to the completion time of command sequences, a scheduling optimization model of AS/RS whose objective was to minimize the completion time of command sequences was established, and an improved ant colony algorithm was given to solve this model. To avoid trapping in local optimum in the search process, weight pheromone and random perturbation strategy were introduced. Besides, a state transfer parameter with a mutation probability was proposed to decide the moving direction of ants in the optimization process. Simulation results indicate that comparing with basic ant colony algorithm and genetic algorithm, the improved algorithm has better global search ability and solution precision. In comparison with the first-come-first-served scheduling strategy, the completion time of command sequences obtained by the scheduling optimization method proposed in this paper is improved by 37% at least.