jpchomp / roughness

Brief project trying to find the implementability of mobile accelerometer data to measure road roughness and predict pavement condition (PSI)

Current roughness data are obtained through expensive but accurate procedures. The alternative of a cellphone accelerometer-based measurement of IRI would be of relevance as it does not require expensive equipment or highly qualified personnel. However, the question of its practical accuracy is raised and within this accuracy what its capabilities would be for pavement management and planification.

To test the accuracy and repeatability of measurements, several runs under varying conditions were performed for Windsor Road on the 2 nd November. It was found that, in accordance to other recent literature, the use of a sedan car yielded results closer to reality on all surfaces and conditions, when compared to an SUV.

After the use of the sedan, we found that results were repeatable and in accordance with profilometer roughness data. To do this we tried the software measuring roughness on Interstate Drive (PCC) and State Street (AC), both in Champaign, IL to confirm the accuracy on varying road surfaces, speed levels (15 and 30 mph) on the 7 th of November.

Finally, its possible use as a decision support tool was assessed by contrasting the results obtained with current management practices with the city of Champaign. After meeting with officials from Champaign City Public Works (see Annex 7.2), a region was analyzed for PCI contrasting on November 29 th the correlation between official measurements and IRI data with 3 different regression techniques was performed.

In all cases an adequate correlation was found, where forest models yielded the best correlation. However, these are not enough to make a highly accurate prediction of PCI values (specially in mid- range values), rather its use is recommended for pre-screening of streets requiring in-depth studying. The use of this tool could give consistent readings with a set of calibrated cars from Champaign City Public Works, yielding a reasonable degree of accuracy for road status control over time in residential areas. Also, its value could be increased by coupling recorded data with video footage, and could become a cheap, versatile tool for PMS. Final report here