Smart AgroSensor for Irrigation Applications

This repository contents a decision support system for intelligent irrigation in crops from a low-cost Agroclimatic station, based on Arduino, Raspberry and Machine Learning.

• Watch the video on Youtube

Repository folders:

/yourRoot    -path
  |--  Readme.md
  |-- src # source codes
	|-- Arduino     # slave source codes
	|-- Graphics    # Graphics code
    |-- Raspberry   # master source codes
    |-- Training    # Training code
  |-- Data
    |-- ciclo_1.csv # dataBase file for cicle 1
    |-- ciclo_2.csv # dataBase file for cicle 2
    |-- ciclo_3.csv # dataBase file for cicle 3
    |-- Validacion_1.csv # dataBase file validation for cicle 1
    |-- Validacion_2.csv # dataBase file validation for cicle 2
    |-- Validacion_3.csv # dataBase file validation for cicle 3
    |-- Learning_models # trained clasifiers in .sav

Hardware requirements

• Agro-sensor system, developed at the University of Ibagué.
• Arduino Mega 2560
• Raspberry Pi 3

Software requirements

• Arduino IDE
• Libraries for arduino:
  • TSL2561
  • OneWire
  • Wire
  • DallasTemperature
• Raspbian OS
• Python 2.7 in this case I used Spyder
  • Flask
  • Sklearn
  • Pandas
  • Serial
  • smtplib

How to run

webControl mode:

• Load on arduino mega 2560 the file /yourRoot/src/arduino/AgroSensor_Code.ino
• Run the script: /yourRoot/src/raspberry/web.py by using sudo python web.py
• To access the interface:
  1. It must be connected to the same ethernet network of the raspberry Pi
  2. In your browser enter the IP address of your raspberry Pi, for example: http://172.17.100.26:8000 or the default IP of the server http://0.0.0.0:8000
  3. In the interface you will find the agroclimatic variables and the ON / OFF buttons to control the irrigation system

Learning mode:

• If you want to start a data collection of irrigation cycles:
  1. Load on arduino mega 2560 the file /yourRoot/src/arduino/AgroSensor_Code.ino
  2. Run the script /yourRoot/src/raspberry/Agrosensor.py by using sudo python Agrosensor.py
• If you want to train supervised classifiers with the obtained data:
  1. Open the python editor and run the script /yourRoot/src/training/Train.py
  2. The data file is loaded, for example, Ciclo1.csv, this must be in the /yourRoot/src/data/ folder
  3. Once the code is executed, it will give you a file.sav of the classifier that was trained, this is generated in the folder /yourRoot/src/data/learning_models

Automatic mode:

• Load on Arduino mega 2560 the file /yourRoot/src/arduino/AgroSensor_Smart.ino
• Before running, the code should load the file.sav of the classifier
• Run the script /yourRoot/src/raspberry/AgrosensorS.py by using sudo python AgrosensorS.py

Authors:

Universidad de Ibagué - Ingeniería Electrónica Proyecto de Grado 2019/A

• Harold F. Murcia
• Daniel J. Jimenez