Author: uuwdrmmy

Field Testing

I decided to run some field tests using hardware I already had in hand.

Parts list:

1. FPV Drone
2. Raspberry Pi Zero W
3. Raspberry Pi NoIR cam v2
4. Camera Case
5. Zip Ties
6. Anker Power Supply

Note: I tried soldering the Pi Zero W directly to the drone flight controller, but it drew too many amps and the drone kept rebooting (this started happening after my first flight). I also tried to trigger the camera using the drone radio transmitter by connecting a GPIO pin of the Pi Zero W directly to the flight controller. I will write a separate blog post if I can get this to work.

I took my Raspberry Pi Zero W, attached a NoIR camera, put it in a camera case, and zip tied it to the bottom of my 5″ FPV drone. I wrote a python script to make the zero w automatically take pictures every couple seconds.

Code with GPIO trigger

from picamera2 import Picamera2
import RPi.GPIO as GPIO
from time import sleep
from time import strftime

picam2 = Picamera2()
camera_config = picam2.create_still_configuration(main={"size": (3280, 2464)})
picam2.configure(camera_config)
picam2.start(show_preview=False)

GPIO.setmode(GPIO.BCM)
GPIO.setup(23, GPIO.IN)

def debounce():
    sleep(0.06)

#Time stamp each image
def takePicture():
    print("picture captured")
    picam2.start_and_capture_file("Plant_Test_Pics/" + strftime("%d/%m/%y at %I:%M%p:%S:%f") + ".jpg")

while True:
    time.sleep(0.01)
    if GPIO.input(23):
        takePicture()
        debounce()

Time based image trigger

from picamera2 import Picamera2
import RPi.GPIO as GPIO
from time import sleep
from time import strftime

picam2 = Picamera2()
camera_config = picam2.create_still_configuration(main={"size": (3280, 2464)})
picam2.configure(camera_config)
picam2.start(show_preview=False)

def takePicture():
    print("picture captured")
    picam2.start_and_capture_file("Plant_Test_Pics/" + strftime("%d/%m/%y at %I:%M%p:%S:%f") + ".jpg")

while True:
    time.sleep(2)
    takePicture()

All of my code is on my github repository:

https://github.com/Bobcati/BLD_NDVI_Scripts/tree/main

Last minute Anker power supply addition

I also recorded some GoPro footage to go along with the Infrared (IR) images I took. Check out this video:

Here is a slideshow with some of the infrared photos I took with NDVI processing and unfiltered IR images comparisons.

Step 0 – The Background

Beech Leaf Disease (BLD) has been infecting beech trees across the Northeast of the United States. This disease impacts the leaves of beech trees and can cause tree mortality. I have a personal connection to this because the trees where I grew up are mostly beech. I want to track this disease in order to ensure the beech forests are kept safe. This summer, it has been devastating to see how unhealthy the beech tree canopy is.

Step 1 – Questions

The first step to every research project is generating questions. I like these initial questions to be open-ended inquiries. In particular, I look back at previous projects to see how I can apply the tools/skills I learned to my current research. I wanted to contribute to research on using drones to map tree canopies.

For this project, I divided my project into two areas: researching beech leaf disease itself and researching methods to analyze vegetation using aerial photography.

Step 2 – Initial Research

I then find resources to start answering my questions. I like to put all the hyperlinks I find into a document.

For example, here are the initial hyperlinks and resources to answer the drone-mapping questions: https://docs.google.com/document/d/1lnaTr6k-asLfqFJ-mOqyBl3UcgFEhXrW8v_NtLyYdF8/edit?usp=sharing

After doing the initial research, I take a step back and isolate the major takeaways. Some of the key points I found were:

1. Normalized Difference Vegetation Index (NDVI) will be key.
2. I want to take the Open-Source route in terms of software for now
3. I need to put together a parts list for the drone I plan to build
4. I can already start doing image processing by looking at satellite imagery

I put academic research papers into their own document. I also continue to add resources and hyperlinks to the initial research document every time I come up with a new idea or learn about a new resource.

Initial resources on BLD:

• New York Department of Environmental Conservation
• Massachusetts