Real-Time Estimation of Drivers' Trust in Automated Driving Systems
DOI: 10.1007/s12369-020-00694-1
URL: https://doi.org/10.1007/s12369-020-00694-1
archive: archived pipeline: cataloged verified
Summary
Driving-simulator study (Int J Soc Robotics 2021) developing a Kalman-filter framework for real-time estimation of driver trust in an SAE Level 3 automated driving system. Trust dynamics are modeled from sensed behaviors (eye-tracking, system-usage time, performance on a non-driving-related task) rather than from periodic self-report. The trust model parameters are identified from human-subject data, then used to track trust across successive driver-ADS interactions.
Key finding
A Kalman-filter approach combining gaze, usage time, and NDRT-performance signals can produce continuous, real-time estimates of driver trust in a Level 3 ADS, enabling trust-aware automation that can mitigate undertrust and overtrust without interrupting the driver for surveys.
Methodology
Driving-simulator experiment with simulated SAE L3 automation; trust modeled as a hidden state; Kalman filter integrates eye-tracking, usage time, and NDRT performance as observation signals; model parameters identified from user-study data.
Sample size: N=80 drivers
Quality score: 5 / 5