An Improved Mobile Sensing Algorithm for Potholes Detection and Analysis

Abstract

Potholes has been identified as one of the primary reasons for traffic fatalities as well as the wear and tear of vehicle tyres. Numerous efforts have been put into creating technologies that can automatically sense potholes. Notwithstanding, it is challenged with higher computational power. In this study, a mobile-based application for detecting and analyzing potholes is proposed using improved STDEV(Z) and Z-DIFF approach. In this study, an inbuilt mobile device that employs an accelerometer, GPS, and various Z-axis algorithms for potholes detection is presented. Potholes locations’ data were compiled, analysed and transferred to the traveller’s mobile software. Accuracy, precision and recall were used as the performance metrics for the algorithms. The performance evaluation results of the proposed improved algorithm for large potholes, small potholes and pothole cluster in term of sensitivity, specificity, precision and accuracy in percentage are: 81, 88, 82; 40, 58, 40; 86, 87, 94; and 76, 80, 79; respectively. Based on these results, the proposed improved algorithm outperforms the Z-DIFF and STDEV(Z) algorithms. With this approach, cost of vehicle maintenance and road accidents can be minimized as alternative routes can be used during mobility.