Most road crashes occur in black spots. Black spots may have different crash probabilities, in accordance with the demographic factors of drivers, the environment, the weather, traffic conditions of the roads, etc., at different times and situations. In this research, we introduce a system that automatically and intelligently analyzes road segments based on occurred crashes, and in situations of high crash probability in the target spots, it provides the driver with visual and audio warnings before reaching the spot. The data on injury and fatal crashes of Tabriz-Ahar highway, in the northwest of Iran, from 2012 to 2016 were chosen for the feasibility study of the system. Eight one-kilometer black spot segments were introduced in this highway. Binary logit model was employed to obtain the crash probability function in black spots, and the variables of driver’s age, familiarity with the route, day and night, rainy or snowy weather, workday, autumn, winter and hourly traffic volume were significant with 95% confidence level. The identified black spot locations were loaded on the system software in addition to the prediction model, and the system was installed on cars. To evaluate the system, the changes in driving speed were analyzed in two groups, control and intervention. The evaluation results showed that the smart warning is effective in reducing the number of male drivers exceeding the speed limit. Also, the differences in mean speed and speed variation in black spots with and without the smart warning system are significant.
