Carnivorous Plant Greenhouse Monitoring System

A full-stack IoT solution for monitoring and maintaining the health of carnivorous plants in a greenhouse environment. This system uses a network of sensors and cameras with a dual Raspberry Pi setup to collect data, analyze plant health, and provide actionable insights.

System Architecture

Hardware Components

Raspberry Pi Setup

• plant-hub: Central server that collects, stores, and visualizes data

  • Runs Docker containers for data storage, analysis, and visualization
  • Hosts the web interface for monitoring plant health

• plant: Sensor hub that collects environmental data

  • Connected to physical sensors via I2C
  • Transmits data to plant-hub for processing and storage

Sensors

• SHT-30 Temperature & Humidity Sensor: Weather-proof sensor for accurate climate monitoring
• Ambient Light Sensor (0-200klx): Monitors light levels in the greenhouse (planned implementation)
• Web Camera: Used for visual plant health assessments via the plant-doctor AI

Optional Hardware

• Smart Plug for Humidifier: Controllable via Node-RED to maintain optimal humidity levels

Software Components

• sensor-reader: Python application running on the "plant" Raspberry Pi that reads sensor data and sends it to the plant-hub
• plant-doctor: AI-powered visual analysis tool using OpenAI's GPT-4 Vision to assess plant health from camera images
• Grafana: Data visualization and dashboard for monitoring environmental conditions
• Prometheus: Time-series database for storing metrics
• Loki: Log aggregation service
• Node-RED: Flow-based programming tool for automation and alerts
• Alloy: Observability data pipeline

Setup Instructions

Plant-Hub Setup (Raspberry Pi #1)

1. Prerequisites Install Docker and Docker Compose on the Raspberry Pi: https://docs.docker.com/engine/install/debian/

2. Clone the Repository

   git clone https://github.com/username/carnivorous-plant-health.git
   cd carnivorous-plant-health

3. Start the Services

   docker-compose up -d

4. Access the Services

   • Grafana: http://plant-hub:3000
   • Node-RED: http://plant-hub:1880
   • Prometheus: http://plant-hub:9090

Plant Sensor Hub Setup (Raspberry Pi #2)

1. Prerequisites

   sudo apt update && sudo apt upgrade -y
   sudo apt install -y python3-pip python3-venv git i2c-tools

2. Enable I2C Interface

   sudo raspi-config
   # Navigate to Interface Options > I2C > Enable

3. Clone the Repository

   git clone https://github.com/username/carnivorous-plant-health.git
   cd carnivorous-plant-health/sensor-reader

4. Install the Sensor Reader Service

   sudo ./install.sh

5. Verify Installation

   sudo systemctl status greenhouse-sensor

Sensor Wiring

SHT-30 Temperature & Humidity Sensor

• Connect VIN to 3.3V
• Connect GND to GND
• Connect SCL to SCL (GPIO 3)
• Connect SDA to SDA (GPIO 2)

Ambient Light Sensor (Optional will need to enable pins for general use)

• Connect VCC to 3.3V
• Connect GND to GND
• Connect SCL to SCL (GPIO 27)
• Connect SDA to SDA (GPIO 28)

Web Camera Setup

1. Connect the USB webcam to Plant Raspberry Pi

2. Install the Plant Doctor Service

   cd carnivorous-plant-health/plant-doctor
   pip install -r requirements.txt

3. Set up OpenAI API Key

   # Option 1: Using AI_TOKEN (preferred)
   export AI_TOKEN="your-api-key"
   
   # Option 2: Using OPENAI_API_KEY (alternative)
   export OPENAI_API_KEY="your-api-key"

4. Setup the Plant Doctor Application

   # Option 1: Preserve environment variables with sudo -E
   sudo -E ./install.sh
   
   # Option 2: Pass environment variable directly to sudo
   sudo OPENAI_API_KEY="your-api-key" ./install.sh

   Note: When using sudo, environment variables are not automatically passed to the new environment. You must either:

   • Use the -E flag with sudo to preserve all environment variables
   • Or pass the environment variable directly to the sudo command

System Features

Environmental Monitoring

• Real-time temperature and humidity tracking
• Light level monitoring (planned implementation)
• Historical data visualization and trends
• Alert thresholds for critical conditions

Plant Health Analysis

• AI-powered visual assessment of plant health
• Identification of potential issues or diseases
• Carnivorous plant species recognition
• Health status tracking over time
• Tank health analysis combining visual and sensor data
• Historical trend analysis for environmental conditions

Automation

• Smart humidifier control based on humidity levels
• Alert notifications for adverse conditions
• Scheduled image capture and health assessment
• Automatic image retention (keeps last 5 images)

API Endpoints

The plant-doctor service provides the following API endpoints:

Plant Health

• GET /: Web interface for monitoring plant health
• GET /image: Current plant image
• GET /api/health: JSON data of latest plant health analysis
• POST /api/capture: Manually trigger image capture and analysis

Sensor Data

• GET /api/metrics: Test endpoint for Prometheus queries
  • Returns 12-hour historical data for temperature and humidity
  • Includes min, max, average, current values
  • Includes rate of change analysis

Image Data

• GET /api/image/base64: Current image in base64 format
  • Useful for testing and debugging
  • Returns JSON with encoded image data

Dashboard Guide

The Grafana dashboard at http://plant-hub:3000 provides:

• Current temperature and humidity readings
• Historical trends of environmental conditions
• Alert status and history
• Plant health assessment results
• Tank health analysis and recommendations

Prometheus Metrics

The system collects and stores the following metrics:

Temperature

• temperature_celsius: Current temperature
• Historical analysis:
  • Minimum temperature over 12h
  • Maximum temperature over 12h
  • Average temperature over 12h
  • Rate of temperature change

Humidity

• humidity_percent: Current humidity
• Historical analysis:
  • Minimum humidity over 12h
  • Maximum humidity over 12h
  • Average humidity over 12h
  • Rate of humidity change

Troubleshooting

Sensor Issues

• Verify I2C connection: sudo i2cdetect -y 1
• Check system logs: journalctl -u greenhouse-sensor -f
• Ensure proper wiring and power supply
• Test Prometheus queries: curl http://plant-hub:9090/api/v1/query?query=temperature_celsius

Camera Issues

• Verify camera connection: ls -l /dev/video*
• Test camera capture: fswebcam test.jpg
• Check image quality: curl http://plant-hub:5000/api/image/base64

Connectivity Issues

• Ensure both Raspberry Pis are on the same network
• Verify network configuration: ping plant-hub from the sensor Pi
• Check service status: docker-compose ps on the hub Pi
• Test API endpoints: curl http://plant-hub:5000/api/metrics

AI Analysis Issues

• Verify OpenAI API key is set correctly
• Check image quality and lighting conditions
• Review OpenTelemetry logs for analysis results
• Test individual components via API endpoints

Contributing

Contributions to this project are welcome! Please submit pull requests or open issues for bugs, feature requests, or improvements.

License

This project is licensed under the MIT License - see the LICENSE file for details.