Carolina Diaz-Piedra1,2 (), Francesco Angioi1, Marcelo A. C. Fernandes1,3, Christophe Prat4, Jaka Sodnik5, Leandro L. Di Stasi1,6; 1University of Granada, 2Arizona State University, 3Federal University of Rio Grande do Norte, 4Commissariat à l’énergie atomique et aux énergies alternatives, 5University of Ljubljana, 6Joint Centre University of Granada - Spanish Army Training and Doctrine Command
Background: The development of a reliable monitoring fitness-to-drive system represents a key, safety-related issue for the future of partially-automated driving. Among several other physiological indices, both the saccadic main sequence and the stationary gaze entropy are of interest to include in a potential fitness-to-drive system. Both indices seem a valid and objective option to assess arousal levels in complex interactions, including driving. Here, we studied the effects of arousal variations due to the time-on-driving and resting conditions on both indices simultaneously. Methods: Eleven drivers underwent two simulated driving sessions in the early-morning of different days. Each time, they drove a monotonous 2-hour highway scenario, just after finishing a 24-hour shift. Throughout each driving session, eye movements were recorded using both electrooculography (400 Hz) and a 4-camera remote eye-tracker (120 Hz). After 90 minutes, operators were given the opportunity to take a 20-minute break. There were two break conditions: napping vs. supervising the automated system. In both cases, after the break, they took over the control of the vehicle and drove manually for a 30-minute period. The saccadic main sequence (velocity/magnitude relationship), gaze entropy, driving performance and subjective ratings of alertness were analyzed. Results: During the first 90 minutes of driving, gaze entropy increased, whereas saccadic velocity, performance, and perceived alertness decreased. After the break, only napping was able to restore saccadic velocity and perceived alertness. Gaze entropy and performance did not differ between break conditions. Conclusions: Our data show that saccadic main sequence is a more sensitive index of arousal levels than stationary gaze entropy in a driving task. Some features of the recording devices (e.g., wearable vs. remote) might in part explain these differences. Overall, our findings have the potential to improve the future development of an in-cabin driver monitor system by providing accurate measurements of fitness-to-drive.
Acknowledgements: Funded by HADRIAN (EU's Horizon2020 research and innovation prog: No 875597). This document reflects only the authors' view. CINEA is not responsible for any use that may be made of the information it contains. We thank Dr Gianfranchi, Ms Blasquez for their help during data collection and curation.