On-road Stress Analysis for In-car Interventions During the Commute

Abstract

This paper focuses on the larger question of when to administer in-car just-in-time stress management interventions. We look at the influence of driving-related stress to find the right time to provide personalized and contextually-aware interventions. We address this challenge with a data driven approach that takes into consideration driving-induced stress, driver (cognitive) availability, and indicators of risky driving behavior such as lane departures and high steering reversal rates. We ran a study with sixteen commuters during morning and evening traffic while applying an in-situ experience sampling. During 45 minutes of driving through various scenarios including city, highway, and neighborhood roads we captured physiological measurements, video of participants and surroundings, and CAN bus driving data. Initial review of the data shows that stress levels changed greatly between 2 and 9 (out of a 0-min to 10-max scale). We conclude with a discussion on how to prepare the data to train supervised algorithms to find the right time to intervene stress while driving.