In recent years, with the promotion of industry 4.0 and the development of e-commerce, multi-AGV task transportation systems have been increasingly used in intelligent warehousing. Multi-AGV path planning is the basic condition to ensure task scheduling and transportation, so it is very important to study multi-AGV path planning. The main contents of this paper are multi-AGV task scheduling and multi-AGV path planning. According to the source of the subject, the background and research significance of the research content of this paper for the subject adaptation are analyzed, and the different types of current path planning research algorithms are summarized. According to the types of different algorithms, the advantages and disadvantages of each algorithm are explained. This paper introduces the single AGV path planning algorithm, improves the heuristic function and search mode of A* algorithm by studying the principle and development status of A* algorithm, and completes the research on the bidirectional search of A* algorithm, verifies the feasibility of the bidirectional search algorithm of single path planning, and lays A foundation for multi-AGV path planning. Next, the multi-AGV task assignment strategy is studied. An improved task scheduling model of auction method is proposed. According to the actual working state of each AGV, the actual information of the task generated at the current moment and the scheduling table corresponding to the current AGV, the task is auctioned, and then the generated task is inserted into the scheduling table of the corresponding AGV. According to the actual allocation status of ham sausage production workshop, this part explores the actual bidding rules of auction method, and quickly completes the scheduling and allocation of multiple tasks. Finally, this paper introduces the improved auction task scheduling into the multi- AGC path planning system. For AGV path planning and avoid deadlock and collision problem, this article delves into the state of all sorts of all sorts of collision and deadlocks, and classifying all kinds of collision and deadlock, according to different types, set the collision avoidance strategy of want to use, and put forward A classification and A * algorithm based on collision placeholder multiple AGV path planning algorithm, through the experiment, proved the effectiveness of the proposed algorithm.