Road Surface Pits and Speed Bumps Recognition Based on Acceleration Sensor
Yunfei Yin, Wanli Fu, Xianyong Ma, Junpeng Yu, Xinkai Li, Zejiao Dong
Abstract
Road potholes pose dangers to traffic safety, damaging vehicles, and increasing accident risks, especially during low visibility conditions. Traditional detection methods are costly, inefficient, and inaccurate. Leveraging advancements in smartphone sensor capabilities, this article presents a novel approach for road pothole and speed bump detection using smartphone acceleration and GPS sensors, along with extract data features filter (EDFF) technology. The method involves preprocessing acceleration data using Euler point, least squares, and wavelet techniques to enhance accuracy by reducing noise. The genetic algorithm is then employed to identify car model parameters. In addition, EDFF is established based on data features to differentiate between speed bumps and grooves and determine groove properties. Experimental results using real data from Harbin show pit recognition accuracy of over 75% and 100% accuracy for speed bumps. This demonstrates the proposal’s effectiveness in detecting pits and speed bumps accurately, making it useful for practical engineering applications.