Ontology based collection and analysis of traffic event data for developing intelligent vehicles

Traffic accidents and near miss incident data are important to extract remarkable driving scenarios for the development of autonomous vehicles and improvement of transportation systems. In order to analyze accident situations, it is important to design an annotation format that can comprehensively describe the behavior of traffic participants and the traffic environments. The authors have been doing the project to collect traffic accidents and incidents to provide the miss incident database over 10 years. The annotation keywords in this database consists of compound words that are easily understood by humans, however, it is not suitable for use in automatic analysis and machine learning. In this research, we propose the annotation format based on ontology for describing traffic event data. To perform the effect of the proposed method, a driver assistance system to forecast the accident occurrence probability in an intersection base on the traffic event data.