In recent years, there has been an extremely worrying rise in motor vehicle crashes, particularly in developing countries. This has posed a substantial risk to the lives of adolescents and young adults. This research addresses severe road safety issues by applying machine-learning techniques to data collected from in-vehicle sensors. In this research, features used are engine speed, engine torque, throttle position, vehicle speed, steering wheel angle, and more in identifying unsafe driving behaviors, which are significant contributors to road accidents. A range of boosting algorithms are proposed to classify between safe and unsafe driving behaviors, such as CatBoost, AdaBoost, LightGBM, GradientBoost, and Extreme Gradient Boosting, and LightGBM emerging as the strong classifier with the highest accuracy of 99.68%. The goal of examining vast sets of informative data collected from actual driving situations is to improve road security and lessen the financial load connected to road mishaps. By employing machine learning algorithms and incorporating cutting-edge car detectors, we establish a more protected and reliable road ambience for everyone.
Machine Learning for Road Safety Enhancement Through In-Vehicle Sensor Analysis
2024-07-24
684477 byte
Conference paper
Electronic Resource
English
Enhancement of road traffic safety and efficiency through mobile sensor networks
Tema Archive | 2007
|