DeepPack3D: A Python package for online 3D bin packing optimization by deep reinforcement learning and constructive heuristics
Year: 2024
Type: software-paper
Abstract: The rapid advancement of industrial robotic automation has increased the significance of online 3D bin packing optimization for applications, like palletization and container loading. Despite numerous learning-based methods emerging for informed decision-making in this process, the absence of a standardized benchmark makes it challenging to experience the process and validate new algorithms. To bridge this gap, we introduce DeepPack3D, a software package that integrates deep reinforcement learning and constructive heuristic approaches for online 3D bin packing optimization. DeepPack3D provides a foundation for benchmarking, allowing users to evaluate performance using customizable item lists and lookahead values, thereby facilitating consistent research advancements. • DeepPack3D is an open-source Python package for online 3D bin packing optimization. • Deep Q network and four different constructive heuristics are supported. • Lookahead value can be customized to support a flexible online optimization process.
Source: Software Impacts
Authors: Y.P. Tsang, D.Y. Mo, K.T. Chung, C.K.M. Lee
Language: English
Subfield: Industrial and Manufacturing Engineering
Field: Engineering
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $480
Funder: Hong Kong Polytechnic University
OpenAlex ID: W4405836594