Computational & Technology Resources
an online resource for computational,
engineering & technology publications
engineering & technology publications
Civil-Comp Conferences
ISSN 2753-3239
ISSN 2753-3239
CCC: 15
PROCEEDINGS OF THE SEVENTH INTERNATIONAL CONFERENCE ON RAILWAY TECHNOLOGY: RESEARCH, DEVELOPMENT AND MAINTENANCE
PROCEEDINGS OF THE SEVENTH INTERNATIONAL CONFERENCE ON RAILWAY TECHNOLOGY: RESEARCH, DEVELOPMENT AND MAINTENANCE
Edited by: J. Pombo
Paper 10.8
An Augmented Reality and Edge AI Framework for Intelligent Railway Pantograph Maintenance Integrating BIM and CMMS
A. Malta1,2,3
1, Research Centre for Asset Management and Systems Engineering, Coimbra, Portugal
2, Electromechatronic Systems Research Centre, Covilhã, Portugal
3Polytechnic of Coimbra - ISEC, Portugal
Full Bibliographic Reference for this paper
A. Malta, "An Augmented Reality and Edge AI Framework for Intelligent Railway Pantograph Maintenance Integrating BIM and CMMS", in J. Pombo, (Editor), "Proceedings of the Seventh International Conference on
Railway Technology:
Research, Development and Maintenance
",
Civil-Comp Press, Edinburgh, UK,
Online volume: CCC 15, Paper 10.8, 2026, doi:10.4203/ccc.15.10.8
Keywords: railway maintenance, augmented reality, edge AI, MobileNet, BIM, CMMS, object detection.
Abstract
The digital transformation of railway maintenance is increasingly driven by the integration of intelligent technologies capable of improving situational awareness, operational efficiency and knowledge management. Augmented Reality (AR), Artificial Intelligence (AI), Building Information Modelling (BIM) and Computerized Maintenance Management Systems (CMMS) have emerged as key enabling technologies for the development of next-generation maintenance environments aligned with Industry 4.0 and Industry 5.0 paradigms.
This paper presents an integrated augmented reality maintenance assistant combining BIM-based digital models, CMMS maintenance data and real-time object detection using a lightweight MobileNet architecture deployed on mobile devices. The proposed system creates a digital bridge between physical railway assets, AI-based perception and maintenance information systems.
The framework was validated using the pantograph subsystem of the Portuguese Alfa Pendular high-speed train. A MobileNet MultiHW AVG I384 model was trained using operational data and deployed using TensorFlow Lite and MediaPipe for real-time inference.
The model achieved an AP50 of 0.9459 and an F1-score of 0.8253. Quantization-Aware Training reduced model size but significantly degraded performance when detecting small industrial components.
The results demonstrate the feasibility of deploying lightweight deep learning models in augmented reality maintenance environments and highlight the potential of integrating AR, edge AI and maintenance information systems for intelligent railway maintenance.
download the full-text of this paper (PDF, 1664 Kb)
go to the previous paper
go to the next paper
return to the table of contents
return to the volume description