IMPROVING SAFETY AT SHARP CURVES: A DYNAMIC STRATEGY FOR MIXED TRAFFIC CONDITIONS

Sharp curves are particularly hazardous due to narrow, poorly maintained roads and limited visibility. Drivers in heterogeneous traffic often engage in risky behaviors, such as overtaking or speeding, which exacerbates these dangers. Traditional collision prevention methods, including convex mirrors, sensor-based systems, and headlights, offer limited effectiveness. Modern transportation systems increasingly rely on advanced in-vehicle and in-road technologies for enhanced safety. A user perception survey was conducted to evaluate the acceptance of a new driving assistance system over conventional methods. The YOLOv5 machine learning algorithm was employed to detect vehicles from video data, identifying patterns and distinguishing them from the background. The proposed curve collision warning system uses cameras and signal boards to detect when a vehicle is approaching a curve too quickly, issuing a visual warning to prompt corrective action. The system demonstrated accuracy and precision above acceptable limits, highlighting its potential to prevent accidents and save lives by providing real-time collision avoidance.