Related Experiment Video
Updated: Oct 3, 2025
07:15
Tactile Vibrating Toolkit and Driving Simulation Platform for Driving-Related Research
Published on: December 18, 2020
4.6K
Driver Behavior Profiling and Recognition Using Deep-Learning Methods: In Accordance with Traffic Regulations and
Ward Ahmed Al-Hussein1, Lip Yee Por1, Miss Laiha Mat Kiah1
1Department of Computer System and Technology, Faculty of Computer Science and Information Technology, University of Malaya, Kuala Lumpur 50603, Malaysia.
International Journal of Environmental Research and Public Health
|February 15, 2022
Summary
This study introduces a new driver behavior profiling method using timeframe data segmentation and deep learning. The Convolutional Neural Network (CNN) model achieved 96.1% accuracy, outperforming others for enhanced traffic safety.
Area of Science:
- * Computational modeling and machine learning applied to transportation safety.
- * Development of advanced driver behavior analysis techniques.
Background:
- * Traditional driver profiling categorizes drivers as 'safe' or 'aggressive,' which is impractical due to the dynamic nature of driving behavior and varying international traffic laws.
- * Existing methods fail to capture the continuous and context-dependent aspects of driver conduct.
- * There is a need for a more nuanced and adaptable driver behavior profiling system.
Purpose of the Study:
- * To propose a novel driver behavior profiling approach using timeframe data segmentation.
- * To evaluate the effectiveness of deep learning algorithms (DNN, RNN, CNN) for classifying driving data based on the new profiling method.
- * To identify the most suitable algorithm for a driver behavior recognition system aimed at improving traffic safety.
Main Methods:
- * Development of a two-part profiling procedure: row labeling (assigning safety scores per second) and segment labeling (scoring aggregated timeframe segments).
- * Criteria for scoring developed with Malaysian traffic safety experts.
- * Application and comparison of Deep Neural Network (DNN), Recurrent Neural Network (RNN), and Convolutional Neural Network (CNN) for data classification.
- * Prevention of overfitting in classification algorithms and validation using naturalistic driving data across 1-10 second segments.
Main Results:
- * The Convolutional Neural Network (CNN) demonstrated superior performance, achieving an accuracy of 96.1%.
- * CNN outperformed both Deep Neural Network (DNN) and Recurrent Neural Network (RNN) in classifying driving data according to the proposed profiling method.
- * The study validated the algorithms on various timeframe segments, confirming the robustness of the approach.
Conclusions:
- * The proposed timeframe data segmentation method offers a more practical and accurate approach to driver behavior profiling.
- * The Convolutional Neural Network (CNN) is recommended as the most suitable algorithm for a driver behavior recognition system.
- * The developed recognition system has the potential to significantly contribute to improving overall traffic safety.
