A Four-Dimensional Spatiotemporal Bin-packing Problem in the Cyber-Physical Internet: A Deep Reinforcement Learning
Loading...
Author(s)
You, Yi
Li, Ming
Advisor(s)
Editor(s)
Collections
Supplementary to:
Abstract
Bin-backing problem is a combinatorial optimization problem of loading smaller items into larger bins that have to satisfy geometric constraints. It is categorized into three dimensions: One-Dimensional (1D), Two-Dimensional (2D), and Three-Dimensional (3D). In the field of logistics, the 3D bin-packing problem is also commonly referred to as the Container Loading Problem (CLP). With the establishment of Physical Internet (PI), goods are packaged in smart containers of modular dimensions that are reusable or recyclable, i.e., PI(Π) containers. It standardizes containers and promotes the research on weakly hetergeneous 3D loading problems. Monitoring through PI identifiers facilitates tracking and tracing, akin to internet dataq packets, thereby digitizing the physical bin packing process. The emergence of Cyber Physical Internet (CPI) provides real-time information to the network layer, propelling the transition from 3D spatial packing problems to four-dimensional spatiotemporal packing problems. In contrast to PI, CPI further demands rapid alignment between freight demand and logistics resourcesw, necessitating real-time decision-making within large-scale PI network. Rather than planning for the future, decision are made at the moment of data acquisition. However, when dealing with extensive datasets, executing pre-forecasted packing scenarios according to a predetermined plan becomes challening, leading to resource wastage and reduced network efficiency.
Sponsor
Date
2024-05
Extent
Resource Type
Text
Resource Subtype
Poster