Quantifying Drivers' Physiological Responses to Take-Over Requests in Conditionally Automated Vehicles

Before the introduction of fully autonomous vehicles with all their benefits and positive impact on quality of life (e.g., increased mobility options, reduced carbon footprint, road safety), researchers propose an era of conditionally automated vehicles where the driver must take over (resume control of the automated vehicle) in critical situations. In terms of human-computer
interaction (HCI) during the take-over process, the driver's physiological signals seem promising as they could be read and understood by the vehicle. In this paper, we quantify the physiological responses to take-over requests (TOR), i.e., we determine their amplitudes, delays, and durations. We measured and examined drivers' heart rate, pupil diameter, horizontal gaze dispersion, blink rate, skin conductance response, and skin temperature. Values before the TOR were compared with values after the TOR, averaged over different time intervals. In addition, the duration until the first noticeable change in each physiological response (delay) and the duration until the signals stabilized to their normal values (duration) were measured. The results showed that the relatively greatest effect of TOR was observed in skin conductance (from -62% to 142%). The fastest response (on average) to TOR was observed in pupil diameter (2.24 s ± 2.48 s), followed by skin conductance and heart rate. Manual or automatic artifact correction has not yet been performed and should be included in further analysis.