Box-Scan: An Efficient and Effective Algorithm for Box Dimension Measurement in Conveyor Systems using a Single RGB-D Camera

Abstract

Different algorithms for object dimension measurement have recently been proposed, leveraging the power of RGB-D cameras. However, most of these algorithms are not suitable for deployment in the logistics industry: they are only able to deal with non-moving objects and/or the usage of multiple cameras introduces time delays. In this paper, we introduce Box-Scan, a novel algorithm that enables real-time box dimension measurement in conveyor systems using a single RGB-D camera. We discuss the industrial setting in which our algorithm needs to operate, as well as a prototype that integrates the proposed algorithm. Furthermore, we provide an analysis of the effectiveness of our prototype as a function of the conveyor speed, demonstrating that the prototype built comes with a maximum measurement error of less than 5% at a conveyor speed of 3.4 km/h.