FECT

Abstract: Road safety plays a vital role in modern transportation systems due to the rising risk of accidents from increased traffic. Speed bumps are commonly installed to regulate vehicle speed and protect pedestrians. However, manual monitoring of these safety elements is resource-intensive. This research aims to develop an adaptive system for detecting speed bumps using signal processing and sensor data (accelerometer, gyroscope, GPS, compass, and timestamp). Machine learning algorithms, including k-nearest neighbors (KNN), dynamic time warping (KNN-DTW), and support vector machine (SVM), were tested for speed bump detection and classification. Experimental results show that SVM outperforms the others, achieving 92% accuracy, compared to KNN’s 89% and KNN-DTW’s 87%. SVM is purposed to detect speed bumps and visualize them on a map. The study demonstrates that employing signal processing and machine learning techniques can effectively detect speed bumps, enhancing road safety monitoring.
Keywords: Road Safety, Speed Bump Detection, Signal Processing, Machine Learning