Human Behavior and Cognitive States Characterization (Post-Event)

To build safer, human-centered Physical AI systems, it is essential to understand how users perceive and behave during critical situations, integrating these human factors into both early-stage design and algorithmic control.

My research focuses on driver behavior and cognitive states across both manual and automated driving environments:

• Manual Driving: I examine driver responses during hazardous scenarios, specifically quantifying metrics such as perception time and steering/pedal reaction times [J1]; and investigate how distinct emotional states impact driving performance [J2].

• Automated Driving: I study driver behavior under system fallibility, such as when vehicles deliver inaccurate or misleading information [J5, J6], behavioral patterns driven by high versus low trust in automation [J7], and driver reactions when automated driving styles deviate from user expectations [J9].

Methodology: To capture these complex behavioral and cognitive dynamics, my experimental setups combine high-fidelity driving simulators with multimodal tracking methods, including eye-tracking, heart rate, galvanic skin response (GSR), and 3D skeleton motion capture.