Online 3D Bin Packing with Constrained Deep Reinforcement Learning
Authors: Hang Zhao, Qijin She, Chenyang Zhu, Yin Yang, Kai Xu741-749
AAAI 2021 | Conference PDF | Archive PDF | Plain Text | LLM Run Details
| Reproducibility Variable | Result | LLM Response |
|---|---|---|
| Research Type | Experimental | We have conducted extensive evaluation showing that the learned policy significantly outperforms the state-of-the-art methods. A preliminary user study even suggests that our method might attain a human-level performance. |
| Researcher Affiliation | Academia | 1National University of Defense Technology 2Clemson University |
| Pseudocode | No | The paper describes the method but does not include any structured pseudocode or algorithm blocks. |
| Open Source Code | No | The paper does not provide any statement or link indicating that the source code for the described methodology is publicly available. |
| Open Datasets | No | The paper describes how training and test sequences were synthesized and generated (e.g., 'synthesized by generating items out of I', 'generate training sequences via cutting stock'), but it does not provide concrete access information (link, DOI, repository) for these specific datasets. |
| Dataset Splits | No | The paper mentions 'The training and test sequence is synthesized' but does not specify exact training, validation, and test splits (e.g., percentages or sample counts for each split, or a specific validation set). |
| Hardware Specification | Yes | We implement our framework on a desktop computer (ubuntu 16.04), which equips with an Intel Xeon Gold 5115 CPU @ 2.40 GHz, 64G memory, and a Nvidia Titan V GPU with 12G memory. |
| Software Dependencies | No | The paper states 'The DRL and all other networks are implemented with Py Torch (Paszke et al. 2019).' It mentions PyTorch but does not specify a version number for it or any other key software dependencies. |
| Experiment Setup | Yes | We set L = W = H = 10 in our experiments with 64 predefined item dimensions (|I| = 64). We also set li L/2, wi W/2 and hi H/2 to avoid over-simplified scenarios... We find through experiments that the following weights lead to consistently good performance throughout our tests: α = 1, β = λ = 0.5, and ω = ψ = 0.01. |