ASAM OpenLabel standard now also used in rail sector for development of automated driving

In order to support research and development on automated driving in railroad operations, the German Center for Rail transport Research has published the multisensor dataset “Open Sensor Data for Rail 2023”. This data is intended to help train automated driving functions using machine learning techniques, thus improving object recognition in railroad environments. To enable comparability and reusability of the data, the dataset uses the standardized annotation format ASAM OpenLABEL.

Hoehenkirchen, Germany - Just as the automotive industry is developing autonomous driving, the rail industry is working on the realization of fully “automatic train operation” (ATO). One major challenge is object detection: Train paths must be monitored and obstacles must be detected early and reliably so that emergency braking can be initiated in time if needed. In order to develop this, machine learning methods are used. They require training data that are collected by different types of sensors. The availability of multi-sensor data for rail traffic has been limited to date.

Aim of the project

"Open Sensor Data for Rail 2023" (OSDaR23) is the first open multi-sensor dataset for railroads in mainline traffic that provides information from various sensors for training, validation and testing purposes. The data was collected by infrared cameras, color imaging cameras, as well as LiDAR, radar, position and acceleration sensors.

The goal of publishing this is to train, test and validate software systems with machine learning techniques and to improve the automated environment perception on railways. To be able to exchange and reuse the data, it must be readable and understandable by all users. To ensure this, the project group has chosen ASAM OpenLABEL as the format to annotate the data. The Standard specifies how object information must be categorized and described in order to provide autonomous driving systems with a common fundamental and in-depth understanding of their environment. Thus, communication problems between systems which can lead to accidents in real life are thus avoided.

Vicomtech’s role

Vicomtech had two main objectives in this project. On one hand, it aimed to rectify existing problems and assist in the seamless transformation of its annotations into the widely recognised OpenLabel format.

Vicomtech's involvement has focused on rectifying the inaccuracies and anomalies present in the dataset. Through analysis and data processing expertise, Vicomtech's team has worked to ensure the integrity and reliability of the information.

And finally, Vicomtech has leveraged its technological prowess to develop a web application, known as WebLabel. This innovative solution is fully compatible with OpenLabel standards and has been specifically designed to provide an intuitive and accessible platform for the visualization of annotated data. By employing state-of-the-art visualization techniques, WebLabel allows data to be explored and analyzed efficiently, making it easier for both experts and the general public to obtain valuable information.

This project brings us one step closer to the development of automated driving for all types of transport, and shows us that, thanks to artificial intelligence, we will be even more capable of detecting anomalies in the driving environment and preventing avoidable accidents.

Through this link you can visit the github of the WebLabel. 

(*Picture: Sensor configuration on the data collection vehicle. Source: DB Netz AG)