Citation

Ahmad, Khubab and Em, Poh Ping and Ab Aziz, Nor Azlina (2024) Utilizing OBD II Time Series Data for Driver Drowsiness Detection: A One-Dimensional CNN Approach. In: 2024 Multimedia University Engineering Conference (MECON), 23-25 July 2024, Cyberjaya, Malaysia.

Text Utilizing OBD II Time Series Data for Driver Drowsiness Detection_ A One-Dimensional CNN Approach.pdf - Published Version Restricted to Repository staff only Download (557kB)

Abstract

Driver drowsiness poses a significant threat to road safety, necessitating the development of advanced detection systems to mitigate the risk of accidents and injuries. In response to this pressing issue, this research introduces a novel approach that integrates On-Board Diagnostic II (OBDII) sensor data with a state-of-the-art camera system. This innovative fusion combines crucial parameters such as speed, RPM, throttle position, and steering torque from the OBD-II sensors with the capabilities of a meticulously trained model and the camera system. By seamlessly integrating these technologies, we aim to streamline and enhance the accuracy of data labelling processes. One of the important elements of our technique is the use of time series data in the windows, These windows serve as samples that capture specific intervals of time within the overall data stream and allows us to analyse driving behaviour sequentially. Using this method, we strategically use a powerful 1D Convolutional Neural Network (1D CNN) to classification problems. Through rigorous training and validation, our integrated model achieves an impressive testing accuracy rate of 88.34% in distinguishing between drowsy and normal driving patterns. We've conducted thorough 5-fold cross-validation to evaluate our driver drowsiness detection system. The results show a mean AUC of 93% and a mean test accuracy of 84.28%, affirming our system's effectiveness in distinguishing between drowsy and normal driving patterns This achievement underscores the effectiveness of our proposed system in discerning subtle variations in driver behaviour. By providing timely warnings or interventions to drowsy drivers, our innovative solution holds promise in saving lives and preventing injuries, thereby contributing to safer road environments and the overall reduction of drowsiness-related accidents.

Item Type:	Conference or Workshop Item (Paper)
Uncontrolled Keywords:	Convolutional Neural Network, K-fold cross-validation, Driver Drowsiness Detection
Subjects:	T Technology > TK Electrical engineering. Electronics Nuclear engineering > TK5101-6720 Telecommunication. Including telegraphy, telephone, radio, radar, television
Divisions:	Faculty of Engineering and Technology (FET)
Depositing User:	Ms Nurul Iqtiani Ahmad
Date Deposited:	12 Feb 2025 06:07
Last Modified:	12 Feb 2025 06:07
URII:	http://shdl.mmu.edu.my/id/eprint/13440

Downloads

Edit (login required)