Research on Safe Driving Diagnosis Algorithms for Driving in Poor Visibility Traffic Environment Context

In recent years, the pace of decline in traffic accidents in Japan has slowed, and accidents involving vulnerable road users remain a serious problem. Many of these accidents stem from psychological factors among drivers, such as an inability to anticipate dangers lurking in environments with poor visibility, like when passing parked or stopped vehicles. This study proposes a new safe driving diagnostic system to address this issue by focusing on the driver's “awareness.” This system does not aim to intervene while driving but seeks to encourage voluntary behavioral improvement by providing feedback after driving. We developed a diagnostic algorithm that quantifies potential risk by calculating a “virtual collision speed.” This metric is derived from continuous simulations simulating pedestrians suddenly emerging from behind obstacles. The algorithm calculates the speed at which the driver's vehicle would collide with a virtual pedestrian after applying emergency brakes. This value serves as a quantitative indicator of driving risk, considering both vehicle speed and lateral distance from obstacles. To validate the algorithm's effectiveness, simulations were conducted using real-world near miss driving data. Results showed a strong correlation between the calculated virtual collision speed and the danger level of officially classified events. Higher-risk events consistently exhibited higher maximum virtual collision speeds and broader potential collision zones. This confirms that our algorithm effectively visualizes and quantifies potential risks often unrecognized by drivers, serving as a valuable tool for providing objective feedback and fostering safer driving habits.