Driver behaviour characterization using artificial intelligence techniques in level 3 automated vehicle.
| dc.contributor.advisor | Zhao, Yifan | |
| dc.contributor.advisor | Brighton, James L. | |
| dc.contributor.author | Yang, Lichao | |
| dc.date.accessioned | 2024-04-03T12:15:55Z | |
| dc.date.available | 2024-04-03T12:15:55Z | |
| dc.date.issued | 2021-09 | |
| dc.description | Brighton, James L. - Associate Supervisor | en_UK |
| dc.description.abstract | Autonomous vehicles free drivers from driving and allow them to engage in some non-driving related activities. However, the engagement in such activities could reduce their awareness of the driving environment, which could bring a potential risk for the takeover process in the current automation level of the intelligent vehicle. Therefore, it is of great importance to monitor the driver's behaviour when the vehicle is in automated driving mode. This research aims to develop a computer vision-based driver monitoring system for autonomous vehicles, which characterises driver behaviour inside the vehicle cabin by their visual attention and hand movement and proves the feasibility of using such features to identify the driver's non-driving related activities. This research further proposes a system, which employs both information to identify driving related activities and non-driving related activities. A novel deep learning- based model has been developed for the classification of such activities. A lightweight model has also been developed for the edge computing device, which compromises the recognition accuracy but is more suitable for further in-vehicle applications. The developed models outperform the state-of-the-art methods in terms of classification accuracy. This research also investigates the impact of the engagement in non-driving related activities on the takeover process and proposes a category method to group the activities to improve the extendibility of the driving monitoring system for unevaluated activities. The finding of this research is important for the design of the takeover strategy to improve driving safety during the control transition in Level 3 automated vehicles. | en_UK |
| dc.description.coursename | PhD in Manufacturing | en_UK |
| dc.identifier.uri | https://dspace.lib.cranfield.ac.uk/handle/1826/21127 | |
| dc.language.iso | en_UK | en_UK |
| dc.publisher | Cranfield University | en_UK |
| dc.publisher.department | SATM | en_UK |
| dc.rights | © Cranfield University, 2021. All rights reserved. No part of this publication may be reproduced without the written permission of the copyright holder. | en_UK |
| dc.subject | Action recognition | en_UK |
| dc.subject | non-driving related task | en_UK |
| dc.subject | deep learning | en_UK |
| dc.subject | level 3 automation | en_UK |
| dc.subject | computer vision-based driver monitoring | en_UK |
| dc.subject | driving safety | en_UK |
| dc.title | Driver behaviour characterization using artificial intelligence techniques in level 3 automated vehicle. | en_UK |
| dc.type | Thesis or dissertation | en_UK |
| dc.type.qualificationlevel | Doctoral | en_UK |
| dc.type.qualificationname | PhD | en_UK |