Skip to content

WeedWarden

Cara Walter edited this page Feb 24, 2022 · 23 revisions

Weed Warden | Updates | Github

Project Members: Liam Duncan

Overview

Controlling weeds during the off season is essential for farmers to protect resources and maximize crop yield. Traditionally weeds are taken care of by spraying herbicides on the field. This approach is expensive, wasteful, and harmful to the environment. Robotic weeders are a good solution to minimize environmental impact and save money on herbicides, but they are expensive (>$100,000). The Weed Warden is a low-cost (<$500) weed detection sensor that can be mounted on rovers or tractors. The Weed Warden uses a spectral triad sensor to detect weeds and can then trigger a weed removal system such as a sprayer when weeds are detected.

The system uses the enhanced normalized difference vegetation index (ENDVI) to measure the difference between near-infrared (which vegetation strongly reflects) and red light (which vegetation absorbs) to produce a value that ranges from -1 to +1 which will indicate if there is plant life under the sensor. Weed Warden is a low-cost alternative to robotic weeders that can help reduce costs and pollution in agricultural settings.

Picture of the Weed Warden:

General Description

Weed Warden consists of a spectroscopy sensor, OPEnS Lab Hypnos Board, 5V battery, and Adafruit Feather M0 with SD storage. The spectral triad sensor will take a reading of the ground and determine if weeds are present. If weeds are present this will trigger the 12V relay on the Hypnos board to dispose of the weed. The method used to determine if a weed is present is called the normalized difference vegetation index (NDVI). This algorithm will compare the near-infrared (which vegetation strongly reflects) and red light (which vegetation absorbs) to output a value between -1 and +1. By setting a benchmark NDVI value on a patch of bare dirt and comparing that value to new value the presence of weeds can be accurately determined.

A graph of typical Weed Warden operation can be found below.

The Graph above shows the NDVI value pulled from the sensor in grey, the threshold value in blue, and the presence of grass in green. When the NDVI value (grey) crosses the threshold value (blue) the 12V relay will be triggered.

Hardware Block Diagram

This block diagram gives an overview of the Weed Warden's hardware and some functionality.

Detection Algorithm

The index or indices that we will use on Weed Warden will be used to identify if vegetation is present. Once the minimal value for the lowest levels of vegetation is obtained, the calibration constant will be used to check if vegetation is present. If the value is greater than the constant, the robot will mark the area as vegetation. If the value is less than the constant, the robot will not mark it as vegetation.

Normalized Difference Vegetation Index (NDVI):

NDVI is a spectral band calculation that uses the visible (RGB) and near infrared (NIR) bands of the electromagnetic spectrum. The basic concept is that chlorophyll in plants absorb red light during photosynthesis and healthy plants reflect very strongly in the NIR band.

Enhanced Normalized Difference Vegetation Index with Blue Light (NDVIB):

This variant of the NDVI uses blue light instead of red led light. Blue light is absorbed during photosynthesis by chlorophyll, and blue light is absorbed by anthocyanins which are protective pigments that prevent the vegetation from overheating.

For this project the NDVIB algorithm is used to detect vegetation.

Objectives

  • Relatively low cost (<$500)

  • High accuracy (>98%)

Property Customer Requirement Engineering requirement Met?
Sensor Height The sensor must go on a tractor boom or a rover The sensor needs to work at up to 2 feet off the ground Yes: sensor can be used at 1 or 2 foot heights
Vegetation Size The sensor needs to detect weeds larger than a baseball The sensor needs to detect weeds that are larger than a 3" x 3" square -- Untested
Detects Weeds The system must detect weeds The sensor must recognize vegetation when under the defined height and vegetation size conditions Yes: detects weeds under the defined conditions
Accuracy The system must be accurate System must detect weeds 98% of the time for defined height and vegetation conditions Untested
Save Data Save the sensor data Save the sensor data to a micro-SD card Yes: All 18 wavelengths of sensor data are stored to micro-SD
Sensor Field of View at 2 Feet Undefined 8.88 inch radius circle NA
Sensor Field of View at 1 Foot Undefined 4.48 inch radius circle NA
Affordable The system must be affordable The system shall cost less than $500 Yes
Weatherproof enclosure The system must be weather resistant System is resilient to dust and water Yes

Future Objectives

  • Be able to support an array of sensors

  • A way to check on your sensors remotely

  • Advanced data postprocessing for user

Resource List

Tutorials

Clone this wiki locally