Automatic Detection of Driving Distractions

Project Overview

Developed as part of the "Embedded System Modelling" course, this project aims to design a safety-critical system capable of detecting driver fatigue and distraction in real-time. The goal was to follow a rigorous engineering workflow (V-Model), from requirements analysis to architectural design using SysML and implementation in C.

Architecture & Hardware

The system follows a classic embedded architecture: Data Acquisition -> Processing -> Actuation.

1. Sensing Layer (Input)

• IMU (Gyroscope): Real-time monitoring of head inclination (X/Y axes) to detect nodding off.
• Steering Sensor: Measures angular velocity to detect erratic driving patterns.
• Telemetry: Speedometer and distance sensors (LIDAR/Ultrasonic) for context-aware analysis.

2. Embedded Controller

A Raspberry Pi running a custom C program acts as the ECU. It processes sensor streams via GPIO/I2C and executes the state machine logic.

3. Actuation Layer (Output)

• Visual Interface: LED array (Warning/Critical) and OLED Display.
• Safety Interlock: Automatic braking signal triggering based on risk level.

Modeling & Methodology

Before implementation, the system was modeled using SysML (Systems Modeling Language) on Enterprise Architect. We defined Block Definition Diagrams (BDD) for hardware structure and State Machine Diagrams to handle transitions between "Normal", "Drowsy", and "Critical" states.

Real-Time Detection Logic

The system operates on a 200ms polling cycle to ensure immediate response.

Symptom	Sensor Data	Threshold Condition
Micro-sleep	Gyroscope (IMU)	Tilt > ±30° for > 400ms
Tailgating	Distance + Speed	Distance < (Speed / 2) (Safety Gap violation)
Erratic Steering	Rotary Encoder	Delta > 20°/s at Speed > 40 km/h

Response Matrix

A prioritized interrupt system determines the safety action.

Risk Level	Condition	System Reaction
Caution	Single symptom (Tilt OR Steering)	Visual Alert (Yellow LED)
Warning	Multiple symptoms	Visual + Audio Alert (Red LED)
Critical	Symptoms + Safety Distance Breach	Full Alert + Autonomous Braking (Level 1)
Danger	Imminent Collision Risk	Emergency Braking (Max)

Conclusion

This project demonstrated the practical application of Model-Based Systems Engineering (MBSE). By combining rigorous SysML modeling with low-level C implementation, we created a robust prototype capable of saving lives. The source code is available here: GitHub Repository.