As the drivers approach the signalized intersection, the drivers will face the dilemma of choosing to stop or continue ahead at the beginning of the yellow signal. Since the yellow signal lasts only 3 s, an improper decision will result in major accidents before it turns red. Back-end crashes and crossing of intersections result from sudden stops, resulting in a red-light violation, and/or it might end with a right-angle crash. Understanding the driver’s driving behaviour when it is in the dilemma zone approaching the signalized intersection and then continuing by classifying the outcome of the driving behaviour (safe stopping and unsafe stopping) is this study's main motive. The car's parameters and the driver's driving behaviour parameters are captured using a driving simulator. The simulated test environment is designed to span rural, suburban, and urban roads to mimic the real-life scenarios and understand the driver’s driving behaviour in different environments. The driver’s behaviour is classified using different machine learning approaches, SVM and KNN. The categorization of 60 instances is recognized by the 70%–30% rule used for training and testing data, resulting in the inaccuracy of 88.1% for SVM and 72.1%, and KNN is 83.3% and 61.1%, respectively. The classification performance gave 89.6% (SVM-training), 91.6% (SVM-testing), 85.4% (KNN-training), and 66.6% (KNN-testing) for five cross-validations and 30 iterations using random sampling. The outcome of this analysis indicates that SVM gave better results compared to KNN.
Classification of Driving Behaviour Using Machine Learning Methods at Signalized Intersections in Urban and Suburban Roads
Algorithms for Intelligent Systems
Proceedings of 2nd International Conference on Artificial Intelligence: Advances and Applications ; Chapter : 36 ; 415-426
2022-02-14
12 pages
Article/Chapter (Book)
Electronic Resource
English
Predicting Pedestrian Safety at Suburban Signalized Intersections
British Library Conference Proceedings | 1996
|Safety Impacts of Photo-Red Enforcement at Suburban Signalized Intersections
Transportation Research Record | 2006
|Learning V2V interactive driving patterns at signalized intersections
Elsevier | 2019
|The Behaviour of Pedestrians at Signalized and Non-signalized Intersections
British Library Conference Proceedings | 1993
|