This repository contains the TS-BP framework code for solving three-dimensional packing problems. This framework combines a two-stage scheme with a bidirectional packing method to enhance space utilization in large-container packing problems.
This code is ran on a machine with Python version 3.8.0 and PyTorch 2.4.0.
Run main.py to train the TS-BP framework. If you wish to change the item quantity and container sizes, modify the nof_item, L, and W values in config.py.
If you wish to modify container sizes other than 100×100, 200×200, and 400×200, you must also modify the convolutional network in Net.py.
Our paper, A Learning-Based Two-Stage Bidirectional Packing Framework for 3D Packing Problems, is accepted at IEEE Transactions on Automation Science and Engineering.
lf our work is helpful for your research, please cite our paper:
@ARTICLE{11153660,
author={Yin, Hao and Chen, Fan and He, Hongjie},
journal={IEEE Transactions on Automation Science and Engineering},
title={A Learning-Based Two-Stage Bidirectional Packing Framework for 3D Packing Problems},
year={2025},
volume={22},
number={},
pages={21541-21551},
doi={10.1109/TASE.2025.3607410}}