<p>As a critical part of modern industry, assembly line plays an important role in mass-producers because of its great advantages in production efficiency and construction cost. With the increasing demand for customized production, the fixed processing flow of assembly line is becoming more and more difficult to adapt to the changes. The introduction of automated guided vehicles (AGVs) as the material handling equipment improve the flexibility of assembly line, which also derives a complex AGV scheduling problem. In the recent years, because of the limitation in place, no-buffer assembly line appears in some plant, which derives a new problem, processing sequence conflicts (PSC). This paper builds a Petri net model to deal with the AGV scheduling problem and processing sequence conflicts in the no-buffer assembly line. A genetic algorithm based look-ahead scheduling algorithm (LASA) with processing sequence conflict-free is developed to minimize the total processing makespan in a no-buffer assembly line by introducing prospective tasks. Chain task strategy is integrated to further improve the performance of LASA in the case of AGVs deficiency. Two groups of simulation experiments are carried out to test the performance of the proposed algorithms in different scenarios. The results show the proposed algorithm has advantages in processing makespan, AGVs’ efficiency and machines’ efficiency. A robot experiment is carried out in a real no-buffer assembly line. The processing makespan of experiment is close to the theoretical lower bound.</p>