Use Cases of the Near-Miss Database
As one of the world's largest traffic accident databases, we have utilized this big data through analysis and application to enhance traffic safety education and advance preventive safety technologies. To date, the following examples of utilization have been implemented.
【Application #1】Analysis of Accident Causes
The causes of accidents are analyzed using time-series data from 15 seconds before and after the dashboard camera's trigger (thresholds based on acceleration magnitude). A distinctive feature of this database is its ontological structure, which facilitates contextual understanding and makes time-series analysis straightforward, while also allowing for comprehensive searches of large data. This structure is designed with consideration of ITARDA's database framework.
【Application #2】Traffic Safety Education
We provide near-miss video data upon request from various entities, including the National Police Agency, the Ministry of Land, Infrastructure, Transport and Tourism (MLIT), traffic safety education content creators, and TV programs. For example, the Society of Automotive Engineers of Japan uses this data for DVDs designed for hazard prediction training. The MLIT uses it for educational DVDs to prevent aggressive driving, and the National Police Agency utilizes it during lectures for driver’s license renewal courses. Additionally, it is used as reference footage in news program explanations.
【Application #3】Research on Preventive Safety and Autonomous Driving
The data is utilized for evaluating the Advanced Emergency Braking System (AEBS) and the safety of autonomous driving. By analyzing the speed and timing of pedestrians darting out as captured by dashboard camera images, accident scenarios can be reconstructed and used in driving simulators to assess the effectiveness of AEBS in avoiding accidents. Additionally, creating near-miss incident maps for urban areas enables the prediction of accident trends based on road environments and intersection configurations.
