Research on the Problem of 3D Bin Packing under Incomplete Information Based on Deep Reinforcement Learning

Yupeng Wu, Liya Yao*

*Corresponding author for this work

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

3 Citations (Scopus)

Abstract

The Bin Packing Problem (BPP) in the logistics industry is a classic NP-hard problem. In practical applications, often only the size information of the current box can be obtained whereas getting the information of the subsequent boxes almost impossible. In consequence, an algorithm is very important for giving the packing position in the case of incomplete information. This paper used Deep Reinforcement Learning (DRL) algorithm and Monte Carlo Tree Search (MCTS), formed the state input shape for this problem to establish a model to solve the 3D bin packing problem under incomplete information. This model can achieve an average space utilization of 65%. The study's results proved that the model can solve the packing problem under incomplete information and has certain practical benefits.

Original languageEnglish
Title of host publicationProceedings - 2021 International Conference on E-Commerce and E-Management, ICECEM 2021
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages38-42
Number of pages5
ISBN (Electronic)9781665410250
DOIs
Publication statusPublished - 2021
Event2021 International Conference on E-Commerce and E-Management, ICECEM 2021 - Virtual, Dalian, China
Duration: 24 Sept 202126 Sept 2021

Publication series

NameProceedings - 2021 International Conference on E-Commerce and E-Management, ICECEM 2021

Conference

Conference2021 International Conference on E-Commerce and E-Management, ICECEM 2021
Country/TerritoryChina
CityVirtual, Dalian
Period24/09/2126/09/21

Keywords

  • Bin packing problem
  • Deep Q-learning
  • Monte Carlo tree search

Fingerprint

Dive into the research topics of 'Research on the Problem of 3D Bin Packing under Incomplete Information Based on Deep Reinforcement Learning'. Together they form a unique fingerprint.

Cite this