A parallel algorithm for multi-AGV systems

Automated guided vehicles are widely used in various applications, especially in manufacturing. In this paper, we present a novel parallel algorithm for multi-AGV systems. The overall structure is composed of three parts: task assignment, path planning, and vehicle navigation. According to the priorities of AGVs, a greedy method is introduced to assign jobs. For path planning, a mixed-integer programming model of a multi-AGV system is formulated, which can be transformed into a series of sub-problems under certain conditions. Then, an improved routing method with a penalty item is adopted to generate the optimal paths of AGVs. To avoid collisions between vehicles, a simple and effective control method based on resource locking is proposed under the premise of parallelization. Furthermore, we design the experiments according to the warehousing systems. Various practical simulations are performed to illustrate the efficiency and robustness of the new algorithm.