Intelligent systems for automated and optimized packing operations in the e-commerce industry

Abstract

The e-commerce industry often deals with items of different shape, size and weight. Based on submitted orders, items are collected in boxes before delivery.

The boxing process is usually entrusted to human operators, as this task, also known as 3D bin-packing task of irregular objects, requires complex decision-making capabilities.

This project aims at automating the task by developing an intelligent system capable of performing the boxing process, employing robots to manipulate and correctly pack the items. The system will integrate robotic tools with state-of-the-art Artificial Intelligence algorithms for Computer Vision, along with Deep-Learning-based strategies for solving 3D bin-packing problems.

Project Goals

This project, in particular, aims at:

• identifying a characterization strategy for unknown items through Computer Vision, as information on incoming items is usually scarce;
• developing a 3D bin-packing algorithm to find an optimized configuration of the items inside the box;
• defining optimal dimensions for the box to be used to reduce empty spaces;
• defining a manipulation strategy for the pick-and-pack problem, designed to recreate the configuration given by the 3D bin-packing algorithm.

In collaboration with: IMA S.p.A.

---

Do you want to know more about this project? Send us a message!