AI-powered logistics: 90% fewer errors in wagon tracking
ArcelorMittal is one of the world's leading steel and mining companies, supplying materials to industries such as automotive, construction, and manufacturing.
Client:
ArcelorMittal
Country:
Polen
Introduction
To reduce errors and manual work in its rail logistics, ArcelorMittal developed WagonUID in collaboration with Euvic. This is an AI-driven computer vision solution that automatically identifies and tracks wagons in real time. The result was up to 90% fewer errors, faster processes, and more efficient, data-driven logistics.
Challenge
ArcelorMittal faced growing complexity in managing its rail logistics. The company needed a way to streamline the tracking of moving depots and wagons across its operations. Manual data entry was not only time-consuming but also prone to human error, leading to inaccuracies in wagon identification and delays in processing.
The goal was clear: to automate the reading and registration of wagon data, improve accuracy, and save valuable time for logistics teams.
Solution
Together with Euvic, ArcelorMittal developed WagonUID, an AI-powered computer vision solution designed to automatically recognize and identify passing railcars in real time.
The system combines several advanced neural network models:
- Convolutional networks (YOLO) for detecting wagons in images.
- Transformer networks (ViT) for image classification and identifying gaps between carriages.
- Recurrent networks (LSTM) for reading and interpreting wagon numbers.
By processing visual data from multiple sources, WagonUID can accurately determine the railcar number, type, and cargo category. All information is processed and delivered in real time, giving operators instant access to reliable data without manual intervention.
The solution was built using a robust stack of Python, PyTorch, OpenCV, Scikit-learn, and Pandas. Ensuring both performance and scalability.
Result
- Employees no longer need to manually enter wagon data, freeing up time for higher-value tasks.
- The system reduced errors in wagon identification and tracking by up to 90%, creating a faster, more reliable logistics process.
- Real-time data availability improved decision-making and significantly enhanced the overall efficiency of wagon management, while reducing operational risk and human dependency.
Summary
WagonUID shows how AI and computer vision can transform industrial logistics by automating complex, error-prone tasks. Through the combination of cutting-edge neural networks and practical deployment, the solution delivers measurable improvements in accuracy, efficiency, and scalability.
With Euvic’s support, ArcelorMittal successfully modernized its wagon management system. Making operations smarter, safer, and ready for the future of AI-driven logistics.
