Overview

A well-maintained road network is an important component of long-term urban development. Researchers have proposed smartphone-based crowdsourced applications as a low-cost effective approach to obtaining frequent road surface quality updates over the last few years. The values show large fluctuations over the conditions under which the road data was collected, which is one of the key drawbacks of these applications. Through this project, we created a road roughness monitoring platform based on passenger cars that can generate reliable findings while minimizing the impact of variables such as car type, smartphone model, or location. Among the features of the created system include automatic travel detection, the ability to use any smartphone in any location with or without an active internet connection, combining information from various sources, and viewing them on a virtual map. This study also proposes a complete system to identify and classify road anomalies from crowd-sensed smartphone data adjusting to different vehicle speeds and other characteristics. A set of field tests were carried out to evaluate the proposed system and the results show that the proposed solution is effective in predicting accurate International Roughness Index (IRI) values and road anomalies after reducing the effect of these varying factors.

Process

As the first step, we had to design a mobile application for gathering traffic data. To generate ideas for this, we examined previous map-based mobile applications. Next, we created a prototype of the mobile application, starting with some wireframes. Afterwards, we utilized Android to develop the application. We conducted testing on the mobile application and made adjustments based on the results. Due to resource constraints, we were only able to perform a few basic user tests of the application. Following that phase, we developed the web application and the machine learning model to display the results. While I primarily focused on developing the mobile application, my partner took the lead in developing the majority of the web dashboard and the machine learning model.

Challenges Faced

During the project, we encountered several challenges. The first hurdle was the insufficient amount of data available for training the machine learning model. To address this challenge, we employed appropriate pre-processing techniques and, when feasible, applied fuzzy logic inference. Another problem was that the initial version of the mobile application significantly drained the user's battery. To mitigate this, we reduced the rate of data collection through the mobile application. While there were numerous other challenges along the way, we effectively tackled them and managed to complete the project within the specified timeframe.

Accomplishments

Based on this project, we submitted two research papers. The first is to the International Journal of Intelligent Transportation System Research, with the title 'Measuring Road Roughness through Crowdsourcing while Minimizing the Conditional Effects' The other is titled 'Identification of Road Surface Anomalies Using Crowdsourced Smartphone Sensor Data' and will be presented at ICTer 2022. We also took part in the Stanford University, USA, Longevity Design Challenge.

• Measuring Road Roughness through Crowdsourcing while Minimizing the Conditional Effects - Research Paper

• Identification of Road Surface Anomalies Using Crowdsourced Smartphone Sensor Data - Research Paper

View Presentation

The presentation that follows goes into great detail regarding the project. For further information, please see the presentation.