backend

🔌 NeuroWarn BCI Backend

This directory contains the Python code that powers the brain-computer interface and hardware control system.

🧩 Key Components

• 🧠 EEG Processing - Processes brain signals from the Emotiv headset
• 🤖 RNN Warning System - Predicts potential hazards using neural networks
• 🔄 Arduino Communication - Controls the wheelchair motors and reads sensors
• 📡 WebSocket Server - Sends real-time data to the frontend dashboard

📁 File Structure

• wheelchair_drive.py - MAIN PROGRAM that coordinates all system components
• cortex.py - Interface with Emotiv Insight headset
• sub_data.py - Data subscription and processing
• send_websocket.py - Communication with web application
• control.py - Basic wheelchair control logic
• ai_copypaste.py - RNN model implementation

🚀 Setup & Installation

# Clone the repository
git clone https://github.com/mattenarle10/neurowarn.git
cd neurowarn/src/backend

# Windows
python -m venv .venv
.venv\Scripts\activate

# macOS/Linux
python3 -m venv .venv
source .venv/bin/activate

# Install required Python packages
pip install -r requirements.txt

# Make sure your virtual environment is activated
# Windows: .venv\Scripts\activate
# macOS/Linux: source .venv/bin/activate

# Go to the backend's neurowarn directory
cd neurowarn/src/backend/neurowarn

# Start the main program
python wheelchair_drive.py

🧠 Model Integration

The backend integrates with the LSTM neural network models to process EEG data and predict user intentions:

1. Data Collection: The system connects to the Emotiv Insight headset via the Cortex API
2. Data Processing: EEG data is collected in real-time and processed through the following steps:
   • Raw EEG data is collected from 5 channels (AF3, T7, PZ, T8, AF4)
   • Data is normalized using a pre-trained StandardScaler
   • Processed data is organized into a sliding window of 5 time steps
3. Prediction: The LSTM model predicts the user's intended movement direction
4. Command Execution: Predicted commands are sent to the Arduino to control the wheelchair

🔄 Data Flow

Emotiv Headset → Cortex API → Backend → LSTM Model → Command Prediction → Arduino → Wheelchair
                                  ↓
                              WebSocket
                                  ↓
                              Frontend
                          (Visualization)

🛠️ Hardware Communication

The backend communicates with the hardware components through:

1. Serial Communication: Commands are sent to the Arduino via serial connection
2. Sensor Data: LiDAR sensor data is received from the Arduino for obstacle detection
3. WebSocket: Real-time data is sent to the frontend for visualization

⚙️ Configuration

Before running the system, you need to:

1. Connect the Emotiv Insight headset and ensure it's paired
2. Connect the Arduino to the computer via USB
3. Configure the serial port in the wheelchair_drive.py script
4. Create a profile in the Emotiv app for consistent EEG readings

🧪 Testing

To test the system without a physical headset:

1. Use the test data provided in the /src/models/testset.csv file
2. Run the test script: python test_prediction.py
3. Monitor the console output for predicted commands