Stereoscopic warehouses improve logistics efficiency and reduce costs through automation, making it necessary to optimize their storage management systems. This paper focuses on the problem of stacker picking delay in the storage management system of a stereoscopic warehouse, aiming to improve storage and production efficiency through technology and process optimization. Firstly, a path planning model is established to optimize the path length and response speed. Then in order to enhance the global search capability and make the algorithm jump out of the local optimum, the MOEA/D algorithm is adapted to incorporate the adaptive Lévy flight operator into it. The superiority of the improved MOEA/D (IMOEA/D) is verified through a case study.

This paper investigates the automatic guided vehicle (AGV) route planning problem. AGVs perform simultaneous pickup and delivery services to workstations at specified times. An AGV route planning model with minimum transportation cost as the optimization objective is constructed. We add a time window constraint to the model to ensure the overall operational efficiency of the workshop. Meanwhile, a hybrid genetic algorithm with improved variable neighborhood search (GA-VNS) is designed to solve the problem. The algorithm combines the strong local search ability of VNS algorithm to optimize GA. We redesigned the coding method of chromosomes to generate chromosomes that meet the needs of optimization. Five types of neighborhood structures are also designed to improve the algorithm’s optimality seeking ability. We designed two sets of experiments for the characteristics of the problem. The experiments test the example in the literature and the actual AGV logistics transportation data of a workshop. The algorithm compares experimental data with other algorithms to verify the effectiveness and applicability of the proposed algorithm.

For the problem of collaborative decision-making, we propose a multi-agent deep reinforcement learning collaborative behavior decision-making algorithm. Firstly, a discrete state space and a greedy strategy-based action space are established in the context of multi- agent collaborative attack, the conditions for successful collaborative siege are given for the requirements of rapidity and collocation. Secondly, the Markov Decision Process (MDP) framework is established based on the multi-agent collaborative behavior decision algorithm, we introduce the experience replay to train the neural network using gradient descent. Finally, a centralized training and distributed execution architecture is used to complete the training of collaborative behavioral decision making, in which the agents share the same strategy and execute actions independently. The simulation shows that the deep reinforcement learning algorithm is able to realize the multi-agent collaborative decision. It can be placed in a real environment.

In order to increase the number of parking spaces, AGV (automated guided vehicle) intelligent parking lot sets the channel as a two-way single lane. In this paper, a general parking space allocation model is proposed to solve the problem that parking AGVs are prone to fall into deadlock in two-way single lane, and a conflict probability calculation model for AGV paths is further designed. By planning and allocating the target parking spaces of vehicles before the route planning, the potential deadlocks can be avoided reasonably. According to the conflict probability calculation model proposed above, the conflict probability of parking spaces sequence calculated by genetic algorithm is lower than serial number sequence method, from near to far method and alternate allocation method. The simulation results show that the conflict probability is positively correlated with the total transportation time of AGV intelligent parking lot. With the increase of parking AGVs number, it begins obvious that reducing conflict probability to improve the transportation efficiency of AGV intelligent parking lot. Therefore, reasonable planning and allocating the target parking spaces of vehicles can effectively improve the transportation efficiency of AGV intelligent parking lot.

The virtual AGV fully automatic parking system is theoretically researched based on the parking management problems of traditional parking lots such as shortage of parking spaces, poor service quality, and low satisfaction. In this paper, a virtual intelligent parking system based on the signal request mechanism is designed. In theoretical research, corresponding solutions are proposed for the selection of parking spaces, the path planning of the parking AGV and the deadlock processing of the AGV. The selection mechanism of free parking spaces is based on the principle of proximity, path planning is based on Dijkstra's algorithm, and the deadlock conflict resolution method is based on the signal request mechaLnism. The core part is the deadlock conflict processing of intersection nodes. Based on the original priority of transportation tasks, this paper takes into consideration factors such as remaining distance, transportation type, waiting time, time peaks. This paper introduces a new type of dynamic priority which effectively improves the efficiency of solving the backlog of parking AGV caused by deadlock conflicts.