Abstract
In recent years, a strong push towards driverless mobility solutions can be seen in many transportation sectors including railways. While the European Train Control System already specifies the necessary interfaces to open up the possibility of Automatic Train Operation (ATO) for mainline railway vehicles, required infrastructure-side upgrades of interlocking systems are time- and cost-intensive. Alternatively, a pure vehicle-side Automatic Train Operation solution can be conceptualized that relies on processing the same audio-visual input a human train driver would normally base his decisions on. This would require the vehicle-side detection of track-side railway signals to determine the vehicle’s movement authority and allowed maximum speed. Such a signal detection system could furthermore be employed as an Advanced Driver Assistance System (ADAS) or support autonomous shunting operations. To enable such a system, this paper presents GERALD, a novel dataset for a neural network based detection approach of railway signals. The dataset contains 5000 images from a wide variety of railway scenes as well as annotations for the most common types of German mainline railway signals. The material was gathered using publicly available cab-view recordings uploaded by railway enthusiasts on YouTube. Using a state of the art neural network architecture for evaluation, we notice promising detection accuracies despite GERALD being a comparably small dataset. The dataset is freely available for research and non-commercial purposes at: https://github.com/ifs-rwth-aachen/GERALD
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
More From: Proceedings of the Institution of Mechanical Engineers, Part F: Journal of Rail and Rapid Transit
Disclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.