Abstract: To deal with the flexible production system scheduling problem, a new scheduling model was proposed based on the constrains of the single AGV and single buffer area. The strategy of the AGV scheduling was illustrated and the math model with the combination of job scheduling and AGV scheduling was constructed. To solve this model, a modified genetic algorithm was proposed to improve the average machine efficiency by optimizing its gene order. The result of scheduling is not only the jobs processing sequence in every workstation, but the move path of AGV. The experiment results show better computational performance of the proposed algorithm. Equipment utilization is also improved under the same experiment conditions.