1. Acceleration Reading
The MPU6050 reads acceleration values on the X, Y, and Z axes.
Senior Project
AI-Based Smart Accident Detection and Emergency Alert System
A Raspberry Pi and IoT-based safety system that monitors vehicle acceleration in real time, uses a Decision Tree AI model to detect possible accidents, attempts to retrieve GPS coordinates, and sends emergency notifications through the Blynk mobile platform.
Real Time
Acceleration monitoring
Python AI
Decision Tree model
Blynk IoT
Mobile emergency alerts
About The Project
Smarter accident detection when every second matters.
This project uses AI and IoT technologies to detect possible vehicle accidents. The MPU6050 sensor collects acceleration data, the Raspberry Pi 3 processes the readings, the Decision Tree model classifies the movement, and Blynk sends emergency alerts. GPS coordinates are included only if available.
cpu
Raspberry Pi Processing
Python reads acceleration values, calculates total_accel, and runs the model.
pin
GPS + Blynk Alerts
Confirmed alerts include a Google Maps link only when GPS data is available.
System Workflow
From motion sensing to emergency notification in real time.
2. total_accel Calculation
Raspberry Pi 3 calculates total_accel from the live acceleration readings.
3. Threshold + AI
If total_accel is 1.8g or higher, the data is passed to the Decision Tree model.
4. Accident Confirmation
An accident is confirmed only after two consecutive accident predictions.
5. GPS Attempt
After confirmation, the GPS module attempts to retrieve latitude and longitude.
6. Blynk Alert
Blynk sends an emergency alert with a Google Maps link if coordinates are available.
Hardware Components
Core devices used to sense, process, locate, and notify.
Raspberry Pi 3
Main processing unit running Python and the AI model.
MPU6050 Sensor
Measures acceleration on X, Y, and Z axes.
GPS Module
Attempts to retrieve latitude and longitude after accident confirmation.
Blynk IoT Platform
Sends emergency alerts to the user.
Decision Tree Classifier
Classifies movement as Accident or No Accident.
AI System
Hybrid detection reduces false positives.
The Decision Tree Classifier was trained using a CSV dataset containing acceleration readings and labels. The selected features are accel_x, accel_y, accel_z, and total_accel. The system uses threshold filtering first, then AI classification. To reduce false positives, two consecutive accident predictions are required before sending an emergency alert.
01
total_accel below 1.8g is considered normal movement
02
Features: accel_x, accel_y, accel_z, and total_accel
03
Values at 1.8g or higher are passed to the AI model
04
Decision Tree classifies movement as Accident or No Accident
05
Two consecutive accident predictions confirm the alert
Blynk Notifications
Real-time IoT alerts keep the user informed immediately.
Mobile Emergency Alerts
Blynk sends an emergency notification when the system confirms a possible accident.
Google Maps Link
If GPS coordinates are available, the alert includes a Google Maps location link.
GPS Fallback
If GPS is unavailable, the emergency alert is still sent without location details.
Blynk IoT Platform
The Raspberry Pi connects the physical prototype to Blynk for user notifications.
Features
Built for fast detection and clear emergency action.
AI Accident Detection
Real-Time Acceleration Monitoring
Total Acceleration Calculation
Two-Step Accident Confirmation
GPS Location If Available
Blynk Emergency Alerts
Raspberry Pi 3 Integration
Future Enhancements
Next steps for expanding the project beyond the prototype.
V2V communication
Cloud database
Smartphone application
Advanced AI models such as CNN/LSTM
Improved GPS module or active GPS antenna
Testing on an actual vehicle