All about the Road: Detecting Road Type, Road Damage, and Road Conditions - Robotics Institute Carnegie Mellon University

All about the Road: Detecting Road Type, Road Damage, and Road Conditions

Nikitha S. Poddatur and Christoph Mertz
Conference Paper, Proceedings of ITS World Congress, November, 2017

Abstract

In this paper we describe a cost-effective system that can continuously monitor the status of the road infrastructure. It can measure the amount of cracking in the pavement, determine the type of road, and can measure the amount of snow or slush on the road. The image and sensor data is collected by smart-phones mounted on vehicles that travel the roads on a regular basis. The images are analyzed by a machine vision algorithm that uses bag-of-words and other visual features within superpixels and a SVM as the machine learning classifier. We evaluate the accuracy of the system.

Notes
Associated Lab - NavLab

BibTeX

@conference{Mertz-2017-103577,
author = {Nikitha S. Poddatur, Christoph Mertz},
title = {All about the Road: Detecting Road Type, Road Damage, and Road Conditions},
booktitle = {Proceedings of ITS World Congress},
year = {2017},
month = {November},
}