System Diagram
A high-precision automatic deceleration system utilizing advanced microsleep-detection technology, powered by RaspberryPi, Python, OpenCV, and Arduino. Designed to intelligently recognize early signs of driver drowsiness and autonomously reduce vehicle speed, this system aims to significantly minimize traffic accidents caused by human fatigue.
SMYDEKOS (Smart System for Automatic Speed Deceleration) is a vehicle safety system designed
to detect microsleep conditions in drivers and perform automatic preventive actions, such as
reducing vehicle speed and activating visual-audio alerts.
This innovation aims to reduce traffic accidents caused by human error, especially drowsy
driving. By integrating computer-vision-based Camera, Raspberry Pi, and the Arduino,
the system provides an additional safety layer for both drivers and surrounding road users.
Utilizes CVZone, OpenCV, and Mediapipe to detect drowsiness indicators such as eye-aspect ratio, blink duration, and head-pose estimation.
When microsleep is detected, the Raspberry Pi sends commands via PyFirmata to the Arduino Mega to trigger the automatic deceleration mechanism.
The system activates audio alarms, voice warnings, and LED indicators to alert the driver before the situation becomes dangerous.
Designed using accessible and low-cost components—such as a standard webcam, Raspberry Pi, and Arduino Mega—making it suitable for general vehicle integration.
The prototype demonstrated reliable microsleep detection with low-latency response, delivering deceleration commands to the Arduino within milliseconds. This system validates the feasibility of integrating real-time vision-based drowsiness monitoring with autonomous speed control for future automotive safety applications.
Key challenges included minimizing communication latency between the Raspberry Pi and Arduino, and fine-tuning the detection algorithm to reduce false triggers. Through this project, I gained deeper experience in computer vision pipelines, hardware-software synchronization, and real-time embedded control design.