Real-Time FPGA-Based ADAS Solution for Driver Drowsiness Detection and Autonomous Stopping

This study addresses driver drowsiness, a leading cause of traffic accidents, by developing a real-time Advanced Driver Assistance System that integrates biometric detection and autonomous vehicle control. The objective of this study is to enhance road safety through the early detection of drowsiness and automated intervention. The proposed system detects signs of drowsiness by monitoring facial and ocular features using a real-time video stream. Once a predefined threshold is exceeded, an audible alert is triggered. If the driver remains unresponsive, the system gradually reduces the vehicle’s speed and initiates an automated stop procedure. Methodologically, the system employs OpenCV for image processing and a convolutional neural network for lane detection and vehicle control. It is implemented on a high-performance hardware platform using field-programmable gate arrays programmed via Vivado High-Level Synthesis to ensure low-latency operation. The results confirm the system’s real-time capability, accuracy in drowsiness detection, and effective vehicle control under drowsy driving conditions. The system’s novelty lies in its combination of biometric monitoring, deep learning, and hardware acceleration to provide faster and more reliable intervention than existing Advanced Driver Assistance System technologies. This integration sets a new benchmark for proactive road safety measures.