Driver Behavior Analysis of Older Adults at Road Intersections Using Naturalistic Driving Data

Abstract

In this study, I have understood driving behavior difference between drivers with Mild Cognitive Impairment (MCI) and drivers without Mild Cognitive Impairment (Non-MCI) and understood the relationship between cognitive abilities of different individuals and their driving behavior. I have developed different methodologies to extract different measures representing driving behavior at road intersections. Multiple driving individuals residing in MCI and Non-MCI were recruited and their driving data and physiological data were recorded. Driving behavior was represented in two domains (Physiological domain and Vehicular domain). First goal of this study was to find out driving behavior difference between MCI and Non-MCI group of drivers using both physiological domain measures as well as vehicular domain measures using statistical analysis. Second goal of this study was to find relationship between cognitive abilities and driving performance measures. To find out this difference braking patterns of drivers were analyzed just before the intersection to understand the effect of declined cognitive abilities on the effectiveness of driving. Based on the results of the experiments machine learning model was trained to classify drivers in two different classes based on their vehicular and physiological domain driving performance measures. From the experiments performed, I found out that there is some significant difference between MCI and Non-MCI group of drivers in both Physiological domain measures as well as Vehicle domain measures.