Source: own collage based on images from the project report.

ATOSenseData is a dataset of human train driver performance measurements in a simulator. The task was to detect visual stimuli on the tracks. It was created as part of a DZSF project ATO Sense. Bahnbetrieb und Infrastruktur (BBI) at the TU Berlin in cooperation with the German Aerospace Center (DLR), Siemens Mobility and DB Systemtechnik constituted the project consortium. It contains a total of 424 recorded human decisions from 25 train drivers in the DLR study and 287 recorded human decisions from 18 train drivers in the BBI study. In addition to the cleansed data, this dataset also contains raw data of the measurements by DLR and raw data of the measurements by BBI.

Code Repository

The scripts used for parsing and plotting the data in the IAVVC paper can be found in the official openCode repository here.

Supplementing Literature

• R. Tagiew and P. Balaji, "Measuring Train Driver Performance as Key to Approval of Driverless Trains", arXiv, 2025.

• B. Thomas-Friedrich, E. Bosch, B. Cogan, H. Wasle, and A. Naumann, “Menschliche Wahrnehmung als Maßstab für zukünftige ATO-Systeme,” EI - DER EISENBAHNINGENIEUR, September 2023.