2nd ICAI 2022

International Conference on Automotive Industry 2022

Mladá Boleslav, Czech Republic

The Use of The DNN YOLO Architecture in The Analysis of Transport Operations in Intralogistics Process – Case Study

Marek Karkula 1 , Robert Mazur 2 AGH University of Science and Technology Faculty of Management ul. Gramatyka 10, Kraków, 30-067 Poland e-mail: mkarkula@agh.edu.pl 1 , rmazur@agh.edu.pl 2

Abstract Popularization of deep neural networks in image classification and object detection, combined with common image recording measures in an industrial environment, create space for application research in the field of improvement and reduction of data acquisition costs for intralogistics operations. This paper presents the results of research on the application of the object detection architecture from the YOLO group (You Only Look Once) in the analysis of transport performance of intralogistics system, operating with full-pallet loads. The paper contains a detailed description of the developed solution, an estimate of the efficiency of the solution with the use of commercial computers, and an assessment of the observer’s influence on the measurement result for the presented solution and classic methods of mapping transport operations in intralogistics. The purpose of this paper is to indicate the application utility of the presented solution in the internal supply chain, for the transportation system in the plant of an automotive supplier. Keywords: dnn, intralogistics, object detection, transfer learning JEL Classification: C810, C880, L620, L900 1. Introduction The development of modern intralogistics, based on the Industry 4.0 concept, challenges the interdisciplinary scientific approach in the analysis and description of logistic processes. The market position of companies is now not only based on the development of new products and their distribution, but also on the improvement of processes in external and internal supply chains using modern digital tools. Optimization of internal procurement and planning processes through process reorganization is a key element to improve the competitiveness of companies. Modern development of intralogistics 4.0 requires simultaneous production and storage infrastructure, transport and storage techniques and technologies and information support based on material flows of the company (Rosi & Lerher, 2017). The idea of Industry 4.0 assumes the interdependence of people, machines, products and systems at the level of management and information processing. The Industry 4.0 model assumes that these units are intelligent objects capable of independent management and communication with the environment. Thanks to this it is possible to develop a model of virtual reality, which reflects the processes conducted within the system (Radivojević & Milosavljević, 2019). The main

108