Visual Perception in Driving

applied psychophysics | human factors | data analysis

roles: research lead | experiment programmer | vision science data analyst

Abstract

Drivers need to know when and where hazards are on the road to keep themselves and everyone else safe. However, there are cognitive and perceptual limits that can make this harder than we think it should be. The “looked but failed to see” phenomenon is a particularly dangerous problem for road safety, where drivers miss a hazard despite looking at it (Wolfe et al., 2022).

To study this, we needed to ask two key questions: Did drivers actually look at the hazard when they missed it? Do drivers need to look directly at a road hazard to respond to it? To answer these questions, we use eye-tracking to examine where drivers look while they watch short videos of real road scenes and asked them to report where the hazard was. We will examine where and when drivers look, relative to when situations become dangerous in these videos. Our MatLab Psychtoolbox analysis and results will show whether drivers need to look at dangers on the road, or if looking and failing to see is more complicated, helping us understand how to develop data-driven interventions to improve road safety for all road users, not just drivers.

As a lab effort, this work also aimed to support my professor’s Information Acquisition Theory in visual perception attention (Wolfe et al., 2020) with his collaborations from the Department of Brain and Cognitive Sciences, the Computer Science and Artificial Intelligence Laboratory (CSAIL) at MIT, and automotive industry partners.

Conference Poster