Transition to manual: Comparing simulator with on-road control transitions

Area of Science:

• Human-computer interaction
• Automotive engineering
• Transportation psychology

Background:

• Limited research exists on driver behavior transfer in automated driving compared to manual driving.
• Most human-automation interaction studies occur in controlled lab or test track environments.

Purpose of the Study:

• To assess the correlation between driving simulator research and real-world driving conditions for non-critical control transactions in highly automated vehicles.

Main Methods:

• Twenty-six drivers used a highway scenario in a driving simulator with automated driving mode.
• Twelve drivers operated a Tesla Model S with Autopilot activated on a public motorway.
• Participants were prompted to transfer control to and from automation in both settings.

Main Results:

• Drivers resumed control faster in on-road conditions.
• Strong positive correlations were observed for control transfers between simulator and on-road driving.
• No significant differences in workload, perceived usefulness, or satisfaction were found between simulator and on-road drives.

Conclusions:

• Driving simulators demonstrate high relative validity for automated driving research.
• Simulators are a viable research tool for studying human-automation interaction in automated vehicles.