Monitoring road surface conditions has become increasingly vital for every transportation authority. Well-maintained road condition increase road user’s safety and levels of comfort. According to Toronto Star, Toronto’s potholes complains were ranked No.1 among all other complains received by City of Toronto in 2018. Potholes can lead to a considerable cost for drivers when they cause damage to their vehicles, and they can be life threatening for cyclists, leading them to fall or dangerously deviate into traffic. Moreover, potholes cost the City of Toronto approximately $4-5 million each year to fix 250,000 potholes on average each year from 2014 to 2019. Recently, the City of Toronto launched its first of many “pothole blitzes”, hoping to repair 4,000 potholes over 12 hours . Repair campaigns such as this require accurate information about the nature and location of potholes for planning purposes. Therefore, a real-time and low-cost monitoring system to actively evaluate road surface conditions by detecting road-surface anomalies such as potholes is highly desirable. The recent trend of smartphone-based sensing and big data present a great opportunity for the development of innovative road-surface condition evaluation technologies using low-cost smartphone sensor data collected by motorists, autonomous vehicles or other on-road users.
The main objective of this research is to develop a web-based Geospatial Information System (GIS) platform which facilitates the real-time monitoring of road surface anomalies and offers further improvement of road surface quality control in large cities like Toronto. The proposed research presented a low-cost, more efficient technology-based pavement evaluation approach and a centralized information system which are necessary to provide the most up-to-date information about the road status due to the dynamic changes on road surfaces. This information can assist government authorities such as Ministry of Transportation of Ontario (MTO) or municipalities to actively monitor and enhance its road surface condition with a reduced cost.
The proposed research provides a crowdsourcing- and web-based GIS platform for road surface monitoring. To this end, a hybrid-based approach, which can continually detect and classify different types of road surface anomalies from the real-time data streamed from various smartphone sensors, was developed and the operational viability was measured. The developed approach processed data from multiple smartphone-based sensors including linear accelerometers, gyroscopes, and GPS. These data can be used to infer meaningful events (i.e., road surface anomalies), according to the predefined rules. The developed approach is semi-automated with minimal user interaction and offers freedom of usage for smartphone users in terms of phone placements in their vehicles. Thus, this increases serviceability and detection rate of road surface imperfections. In addition, based on the developed detection approach, an Android-based mobile application was developed to detect road surface anomalies from smartphone sensors which can be used by the public.
The detected anomalies were then continuously accumulated, integrated and classified on the central server based on the proposed probabilistic-based integration approach to infer robust interpretation of each anomaly in terms of the level of discomfort and severity. The underlying principle of this approach is to integrate the detected events from multiple users which do not represent a binary scenario. This is primarily caused by different sensing capabilities of the sensors of different smartphones and the diversity of the mechanical properties of vehicles (e.g., suspension stiffness). The proposed approach also considered both spatial and temporal aspects of each detected anomaly for data integration and classification. The outcome from this process would be either a new potential event such as a new pothole, or continuous improvement toward the accuracy of the previously-detected events. Finally, a web-based GIS prototype was developed based on the developed Android-based mobile application and the probabilistic-based crowdsourcing technique. The developed GIS prototype aimed to assist transportation services that deal with road maintenance by actively monitoring current road surface condition for potential maintenance and rehabilitation needs. It also helps assist drivers by notifying them prior to reaching any anomalies (e.g., detected potholes) in order to avoid accidents or vehicle damage.
Currently, smartphone-based sensing is becoming widespread since the mobile devices are equipped with a variety of sensors such as cameras, accelerometers, gyroscopes, GPS, etc. Participatory sensing is anticipated to be an emerging area where smartphone-based measurements are particularly attractive since they are not only widespread but also equipped by several sensing capabilities. Detection of road anomalies from smartphones is a complex and challenging process. Different vehicles have different responses while passing over the same road anomaly due to the difference, for example, in their suspension systems. Existing studies are limited to identifying roadway anomalies mainly from a single data sensor, or lack the usage of combined and integrated multi-sensors in terms of accuracy and functionality. Moreover, previous studies attempting to classify road surface anomalies proposed hard classification approaches leading to a high rate of misclassification due to the fuzzy and unknown boundary between different anomalies sensed by smartphone sensors. The fuzzy classification approach proposed in this research study minimizes most misclassifications. In fact, the proposed probabilistic-based approach aids in the combining of data and provides more accurate inference from multiple detections. The proposed research presented a low-cost, more efficient technology-based pavement evaluation approach and a centralized information system which are necessary to provide the most up-to-date information about the road status due to the dynamic changes on road surfaces. This information can assist government authorities such as Ministry of Transportation of Ontario (MTO) or municipalities to actively monitor and enhance its road surface condition with a reduced cost.
