A Concept for Camera-based Classification of Load Carriers

Abstract

Due to growing environmental awareness, the Circular Economy and in particular the concept of Reverse Logistics (RL) are more and more becoming the focus of industry, yielding ecological as well as economic advantages. However, the successful implementation of the concepts requires that several challenges be met. One of the most common challenges is the lack of information within RL. One proposed solution is to use more Automatic Identification Systems (Auto-ID) to track returning goods and close the information gaps between RL participants. Currently available identification systems are often limited in their field of application, as they can be very expensive, require a huge change in current logistics processes or suffer from physical characteristics, such as electromagnetic absorption or limited visual contact. With this paper, we introduce our novel concept for load carrier classification and quantification using Time-of-Flight (ToF) cameras in combination with color images, beginning with a general overview of the system architecture and process structure. This is followed by an in-depth analysis of the process steps, starting with triggering camera records followed by image pre-processing, classification and finally a quantification of the loaded cargo.