Localized Intelligent Traffic Monitoring System through Ontological Modelling

For decades, one of the Philippines' major problems has been traffic congestion. However, only a handful of proposed solutions focused on intelligent transportation systems (ITS). This paper covers the development of an ITS that employed localized qualitative traffic information and formulated an ontological model for road traffic information. The proposed system utilized sensor data from heterogeneous sources to provide qualitative traffic information such as route recommendation, accident detection, and route passability during flood and weather disturbances. The study utilized a Haar cascade deep learning algorithm to extract traffic volume information and accident information in video streams; while the integrated sensor system was used to obtain temperature, relative humidity, rainfall intensity, and flood level information of the deployment site. These data were used in the formulated ontological model that served as a mindmap that can be navigated by an AI model autonomously. Simulation and validation of the model showed up to 91% accuracy, a 15% improvement when compared to a standard traffic monitoring system without the ontological model using random forest machine learning model in different test scenarios.