Online 3D Bin Packing Algorithm Based on Heuristic Deep Reinforcement Learning

In the field of logistics and transportation, the three-dimensional bin packing problem(3D-BPP) is one of the core challenges in optimizing transportation resource utilization. Given the superior learning and optimization capabilities of deep reinforcement learning (DRL) in complex decision-making, applying deep reinforcement learning to solve the online three-dimensional bin packing problem has become a key research direction in intelligent logistics. This study proposes a deep reinforcement learning algorithm integrating an Event Point trigger and a dynamically updated Empty Maximal Space heuristic rule, along with a multi-objective combined reward function. Experimental results demonstrate that the improved model significantly outperforms traditional deep reinforcement learning algorithms in online three-dimensional bin packing, achieving a 9% higher average space utilization and a 27.3% faster convergence speed.