Effects of Road Geometry and Driving Automation on Pupil Diameter: Evidence from a Simulated Driving Study

Abstract

Pupillometry is increasingly adopted to infer drivers’ psychophysiological states, as pupil diameter is sensitive to variations in cognitive workload, visual demand, and arousal. However, in automated driving contexts, pupillary responses may be simultaneously influenced by road geometry and by changes in the driver’s control role, making interpretation sharp.,This study investigates whether pupil diameter is modulated by road geometry and by the level of vehicle automation during simulated driving. A controlled experiment was conducted using a driving simulator, in which 39 drivers completed two traversals of the same rural road alignment. The first traversal was entirely manual, while the second included a manual-automated-manual sequence across three consecutive road sections. Horizontal alignment complexity was characterized by curve type (easy vs. sharp).,Pupil diameter was analyzed using linear mixed-effects models to account for repeated measurements and inter-individual variability. To properly reflect the sequential experimental design, both a global analysis and targeted within-subject comparisons were performed.,The results showed significant effects of curve type and road section on pupil diameter, as well as a significant interaction between geometry and section. In contrast, no significant main or interaction effects involving the level of automation were observed, even when directly comparing the same road section driven under manual and automated conditions.,Overall, the findings indicate that pupillometry is highly sensitive to infrastructure-induced visual and cognitive demand, while steady-state supervised automation does not produce a robust pupillary signature.

Keywords: Pupillometry, Road geometry, Driver workload, Autonomous driving, Driving simulator, Human factors