Beyond the Current Curve: A Novel Curve Warning System Considering Subsequent Curve Speed Limits

Abstract

This work presents a novel curve warning system that addresses the existing Advanced Driver-Assistance Systems (ADAS) limitations. While current ADAS offer curve warnings, they often lack the consideration of subsequent curve’s speed limit and distance, hindering optimal warning location calculations. Our proposed method leverages Google Maps path data and an existing curve detection system to extract curvature information. This information is then utilized to warn drivers about approaching curves at the safest possible location, taking into account the speed limit of the adjacent curve. Real-time experiments showcase the effectiveness of this system in providing timely and accurate warnings. Furthermore, we conduct a comparative analysis within the Carla simulation environment. We evaluate the existing path-tracking algorithms across various speed scenarios (50, 70, and 80 km/h) and different speed considerations: curve-aware speed and our proposed adjacent curve-speed-limit-aware speed. Metrics such as Root Mean Square Error (RMSE), navigation time, steering angle, speed variation, and throttle usage are employed for evaluation. Our proposed method, incorporating adjacent curve information, demonstrates significant safety improvements, particularly in reducing the RMSE error when navigating the curves on simulated autonomous vehicles.

Keywords

• Advanced driver assistance system (ADAS)
• curve detection
• accident-prevention
• path tracking algorithm
• Carla simulation