Identificación de Daños en la Superficie de Carreteras Mediante Patrones de Aceleración (Identifying Roadway Surface Disruptions Based on Accelerometer Patterns)

Fernando Martinez (fmartinez2004@gmail.com), Luis Carlos Gonzalez (gonzalezgurrola@gmail.com), Manuel Ricardo Carlos (p168786@uach.mx)


Universidad Autonoma de Chihuahua
This paper appears in: Revista IEEE América Latina

Publication Date: May 2014
Volume: 12,   Issue: 3 
ISSN: 1548-0992


Abstract:
In developed countries large-scale technology is used for the monitoring and management of road infrastructure. Sensor networks together with surveillance cameras help identify elements disturbing cars' mobility. On the contrary, in developing countries, the lack of this level of infrastructure makes the task of roads' maintenance and traffic control challenging. The presence of roadway surface disruptions (RSDs), in particular, affects the economy and reputation of individuals and companies. Urban computing strategies can be adopted to collect data about the presence of RSDs and be applied to monitor traffic related issues. The implementation of such level of navigation assistance could help save freight and money to companies and authorities, and even reduce cars' accidents. In this direction, current mobile technologies can help with the identification and location of road imperfections alerting drivers about alternative routes. In this paper we present our work that seeks to enable citizens' cars as road watchers. By means of the mobile phones' acceleration sensing capabilities we are identifying and tagging the presence of RSDs. Using Android-based devices situated on the copilot floor side of a car, 5 Mbytes of road information has been collected. We run a series of experiments aiming to differentiate acceleration patterns associated to potholes, speed bumps, metal humps and rough roads. Currently, the classification of disruptions is being experimented with techniques from the field of Machine Learning (ML) such as artificial neuronal networks and logistic regression. Classification of individual events is over 86% of accuracy that is competitive with those reported in the literature. In this work we provide the first public dataset that could be used by other researchers to offer more insight in this problem.

Index Terms:
road surface disruptions, RSD, mobile computing, mobile sensing, machine learning algorithms   


Documents that cite this document
This function is not implemented yet.


[PDF Full-Text (482)]