Preparations reduce drivers' perceived risk when resuming control from conditionally automated vehicles.

Area of Science:

• Human-Computer Interaction
• Automated Driving Systems
• Cognitive Psychology

Background:

• Unexpected vehicle control transfers in automated driving increase driver risk perception and negatively affect driving behavior.
• Monitoring requests (MRs) preceding takeover requests (TORs) improve takeover performance, but their effect on perceived risk is unclear.

Purpose of the Study:

• To examine the effect of MRs on drivers' perceived risk during automated driving.
• To investigate if physiological indicators can predict perceived risk.
• To explore the association between perceived risk and takeover performance.

Main Methods:

• Sixty participants were randomly assigned to either an MRs + TORs group or a TORs-only group in a simulated automated driving experiment.
• Perceived risk, physiological indicators (eye-movement, heart rate, skin conductance, respiration), and takeover performance (braking, steering) were recorded.
• A random forest model predicted perceived risk using pre- and post-TOR physiological data.

Main Results:

• Drivers receiving MRs before TORs reported lower perceived risk and showed improved physiological responses compared to the TORs-only group.
• The perceived risk prediction model demonstrated strong performance with a mean macro-precision of 0.757 and a mean macro-F1 of 0.758.
• Lower perceived risk correlated with reduced braking input and steering velocity during takeovers.

Conclusions:

• MRs preceding TORs are associated with reduced driver perceived risk and enhanced takeover performance.
• Pre- and post-TOR physiological indicators show potential for predicting driver perceived risk.
• Findings can inform the development of driver-state detection and human-machine interaction strategies in automated driving.