Title

Kalman Filtering to Track Changes in Pupil Size for Automated Driving Systems

Document Type

Article

Publication Date

2021

Publication Title

2021 IEEE 94th Vehicular Technology Conference (VTC2021-Fall)

Keywords

Filtering; Signal processing; Supervisory control; Size measurement; Data models; Noise measurement ; Pupils

Abstract

Automation has become indispensable in all walks of everyday life. In driving environments, Automated Driving Systems (ADS) aid the driver by reducing the required workload and by improving road safety. These systems require human drivers to remain vigilant and maintain supervisory control over ADS. Therefore, the cognitive attention of the drivers must be estimated accurately for the safe adoption of ADS. Because of the non-invasive recording setup used in low-cost infrared eye-trackers, pupil size measurements are increasingly becoming applicable in the estimation of cognitive load. However, pupil size measurements are highly noisy, resulting in the poor classification of cognitive load levels. In this paper, we propose a methodology for improved classification of changes in cognitive load through the introduction of a state-space model-based approach to filter the pupil size data. The proposed approach was demonstrated on data collected from 16 participants while they performed driving task and several secondary tasks that are designed to emulate three different levels of driving distraction.

DOI

10.1109/VTC2021-Fall52928.2021.9625045

Link to Full Text
COinS