Existing methods for road quality analysis using acceleration patterns, cover different kinds of Road Surface Disruptions (RSD)  and report a wide variety of performance metrics. Our work describes the underlying limitations in the approach of using acceleration patterns for predicting road quality and uses various machine learning algorithms for classifying RSDs. We have shared a large amount of heterogeneous data set, collected over a continuous drive, using fourteen vehicles, in the city of Lahore, Pakistan. The design and deployment of a custom-made data logger with accelerometer and GPS sensor has also been introduced by the authors. Four commonly occurring RSDs have been studied, namely: 1) Cat eyes 2) Manholes 3) Potholes 4) Speed Bumps. Finally, the best classifiers and feature sets were used to develop annotated maps for road repair authorities and regular drivers for timely maintenance and intelligent navigation.
 González, Luis C., et al. "Learning roadway surface disruption patterns using the bag of words representation." IEEE Transactions on Intelligent Transportation Systems 18.11 (2017): 2916-2928.