The invention discloses a deep reinforcement learning-based task allocation system and method for a multi-AGV system three-dimensional parking lot, and the method comprises the steps: carrying out the weighted processing of the state in the parking lot through a dynamic multi-attention mechanism, and carrying out the unified coding, and obtaining the spatial layout and vehicle flow characteristics. And constructing a task allocation model using a Dueling DQN neural network, and fusing the global average parking task execution time, the global average vehicle taking task execution time and the AGV load condition into a reward function for training. According to the training step number, dynamic self-adaptive multi-noise is provided for the distribution model in task training, so that the anti-interference capability of the task distribution model is improved. The method can adapt to parking lots of different scales, is more adaptive, can reasonably distribute different tasks and AGV scheduling strategies, reduces the average vehicle parking and taking time of the three-dimensional parking lot, optimizes the AGV load, and greatly improves the scheduling efficiency of the multi-AGV system three-dimensional parking lot.
本发明公开了一种面向多AGV系统立体停车场的基于深度强化学习的任务分配系统及方法,包括:采用动态多重注意力机制对停车场内状态进行加权处理并进行统一编码,获取空间布局和车辆流动特征。构建使用Dueling DQN神经网络的任务分配模型,并且将全局平均停车任务执行时间、全局平均取车任务执行时间和AGV负载情况融入奖励函数,进行训练。根据训练步数向任务训练中的分配模型提供动态自适应的多噪音,得以提升任务分配模型的抗干扰能力。本发明可适应不同规模停车场更具适应性,能够合理分配不同任务与AGV调度策略,减少立体停车场的平均停取车时间,优化AGV负载,极大提高了多AGV系统立体停车场的调度效率。
Deep reinforcement learning-based task allocation system and method for multi-AGV system three-dimensional parking lot
一种面向多AGV系统立体停车场的基于深度强化学习的任务分配系统及方法
2024-10-29
Patent
Electronic Resource
Chinese
IPC: | G06Q Datenverarbeitungssysteme oder -verfahren, besonders angepasst an verwaltungstechnische, geschäftliche, finanzielle oder betriebswirtschaftliche Zwecke, sowie an geschäftsbezogene Überwachungs- oder Voraussagezwecke , DATA PROCESSING SYSTEMS OR METHODS, SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL, SUPERVISORY OR FORECASTING PURPOSES / G06F ELECTRIC DIGITAL DATA PROCESSING , Elektrische digitale Datenverarbeitung / G06N COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS , Rechnersysteme, basierend auf spezifischen Rechenmodellen / G08G Anlagen zur Steuerung, Regelung oder Überwachung des Verkehrs , TRAFFIC CONTROL SYSTEMS |
Parking task allocation and trajectory planning system based on multi-agent reinforcement learning
European Patent Office | 2023
|Automatic parking method based on deep reinforcement learning
European Patent Office | 2020
|Intelligent driving vehicle autonomous parking system based on deep reinforcement learning
European Patent Office | 2024
|