Kästle, Jan Luca; (2024) Objective Assessment of Situational Awareness in an Automated Driving Environment. Doctoral thesis (Ph.D), UCL (University College London).

Text KaestlePhDThesis.pdf - Accepted Version Access restricted to UCL open access staff until 1 October 2025. Download (31MB)

Abstract

Situational Awareness (SA), describing a person’s acquaintance with their environment, is a crucial factor in safety-critical domains, like driving, to ensure the safe execution of desired tasks. With technological advancements in driving automation and the development of automated vehicles, drivers are permitted to engage in non-driving related activities but must remain capable of taking control of the vehicle within seconds at any time. It is therefore essential to understand the factors influencing a driver’s ability to efficiently regain SA to prevent potentially hazardous situations. In literature, assessment of a driver’s SA through questionnaires may be susceptible to misrepresentation and is mainly measured post-trial, which is unsuitable for real-time applications. This thesis proposes a novel approach to detect driving SA by measuring and analysing brain and eye activity using EEG and eye-tracking glasses, ensuring a continuous and objective assessment. Within this thesis, a framework is developed in several stages. Initially, brain areas and frequencies associated with low and high SA are identified. This knowledge informs the training of a machine learning classifier to discriminate between a subject’s low and high SA in a driving-related scenario. It outperforms similar classifiers in literature, yet the distinction between driving SA and SA in other domains requires further research. To address this, eye activity is analysed, and features corresponding to SA are identified. Finally, by combining brain and eye activity data, a classifier is obtained that is able to differentiate between low and high driving SA, outperforming single-sensor models. Results show that activation of neurons in the parietal and temporal lobes of the brain in β (12-30 Hz) and γ (30-45 Hz) frequency bands corresponds with SA. Additionally, pupil diameter and the number of fixations exhibit correlations with SA. The combined classifier achieves a test accuracy of 79.3% in detecting low or high driving SA.

Download activity - last month

Export as CSVXMLJSON

Download activity - last 12 months

Export as CSVXMLJSON

Downloads by country - last 12 months

Export as JSONXMLCSV

Archive Staff Only

View Item