AI-Powered Autonomous Driving Assistance System: Lane Detection, Traffic Sign Recognition, and Real-Time Hazard Detection

ABSTRACT Road traffic accidents remain a leading cause of global fatalities, with human error contributing to over 90% of serious crashes. Traditional navigation systems like GPS provide route guidance but lack real-time awareness of immediate road conditions, lane positions, and potential hazards. Advanced Driver Assistance Systems (ADAS) have emerged as critical safety technologies, yet many existing solutions remain expensive and limited to high-end vehicles. This paper presents an AI-Powered Autonomous Driving Assistance System that combines Computer Vision (CV) and Machine Learning (ML) to provide real-time driving intelligence. The system captures live video feeds from dashboard cameras and processes them using advanced algorithms to detect lane markings, traffic signs, vehicles, and obstacles in real-time. Lane detection employs OpenCV preprocessing combined with Convolutional Neural Networks (CNNs) for robust recognition under varying lighting conditions. Traffic sign recognition utilizes CNN classifiers trained on the German Traffic Sign Recognition Benchmark (GTSRB) dataset. Vehicle and obstacle detection employs YOLOv8 object detection models with distance estimation capabilities. The system provides lane departure warnings, real-time steering suggestions, and a comprehensive dashboard interface that overlays detected information onto live video feeds. Unlike traditional GPS systems offering macro-level navigation, this solution provides micro-level driving intelligence that understands the immediate road environment, creating a comprehensive safety net that reduces human error and enhances situational awareness. Keywords: Advanced Driver Assistance Systems (ADAS), Computer Vision, Machine Learning, Lane Detection, Traffic Sign Recognition, Object Detection, YOLOv8, Convolutional Neural Networks, Real-time Processing, Autonomous Driving