Measuring drivers’ gaze concentration: Comparison of metrics and eye tracking systems with respect to gaze concentration
Examensarbete för masterexamen
Mobility engineering (MPMOB), MSc
This thesis aims to compare gaze concentration metrics from the literature using a remote and a wearable eye tracking system. Drivers’ gaze direction has been seen to narrow, also referred to as gaze concentration, for different driver states. These driver states could either be temporary such as cognitive distraction or lasting longer such as intoxication by drugs or alcohol. Gaze concentration can thus be used as an indication of these states. A small data collection was performed, including 5 test participants using both eye tracking systems at the same time. The participants were subjected to cognitive task during the drive to experience cognitive load. The gaze concentration metrics percent road center (PRC), percent area of interest (PAI), stationary gaze entropy (SGE), transitional gaze entropy (TGE), standard deviation of radial gaze (SDRG), standard deviation of horizontal gaze (SDHG), and standard deviation of vertical gaze (SDVG) were selected for analysis based on a literature review. The metrics were analyzed both for longer and shorter time windows. Results showed that SGE and TGE, which both uses area division of the gaze dispersion, were highly sensitive measures to the size and number of areas chosen. TGE also showed a large variation across participants and between the two systems, suggesting that it is not a robust metric. SDVG was shown to be less sensitive than the other metrics, suggesting that this metric was not suitable to capture gaze concentration during cognitive load. PRC, PAI, SGE, SDRG, and SDHG all showed similar trends, indicating that these metrics captured similar types of gaze behavior during cognitive load. Due to the small dataset used in this study, stating which metric was most sensitive in capturing gaze concentration was not possible. The analysis showed that there is a notable difference between the two eye tracking systems, indicating that comparing results from different studies should be done with care unless the same eye tracking system has been used.
Gaze concentration , Cognitive distraction , Eye tracking system , Metrics , Driver state