Common data management platform for artificial sense training and testing for railway applications

@Article {
author = {Labayen, Mikel},
title = {Common data management platform for artificial sense training and testing for railway applications},
pages = {9},
keywds = {
AI Training and Testing; Artificial Intelligence; Common Data Management Platform; Railway
}
abstract = {

The railway sector is evolving towards digitalisation and automation. Automatic Train Operation (ATO) systems provide different degrees of operation autonomy according to the different Grade of Automation (GoA) levels. GoA3 (driverless) and GoA4 (unattended) levels remove drivers from railway operations, what makes necessary the introduction of Artificial Intelligence (AI) for perception tasks. The performance of AI models directly depends on the quality and diversity of the data used for training. Gathering, preparing and labelling these data is extremely costly and time-consuming, what can jeopardise the development of reliable AI-based systems. This paper describes a common data management platform devised by a consortium of stakeholders in the European railway sector to efficiently ingest and share the required data and presents a Proof of Concept (PoC) of it. 

}
date = {2023-01-01},
}