This project is to take overhead photographs of the tree plantation in order to improve tree health. Firstly, it must be established that useful information can be gathered from overhead photography. Then, the information can be used to place fertiliser, and influence tree maintainence procedures. A secondary goal of this process would be to produce accurate maps of the farm for use in other processes.
NDVI techniques have been historically used as an indicator of plant health. This, combined with RGB photography, should be able to provide some useful information. The economic argument for improving tree health is trong, since small changes in yield across the farm could result in massively increased revenues.
To achieve this, a quote should be obtained for a NDVI survey of the block, in order to determine the relative cost of a DIY system.
The first proof of concept of this project would be to perform overhead mapping with an RGB camera, to show that the system works.
The client has reafirmed the need for this problem to be solved. A quick look at the aerial photography available as a service reveals it as prohibitively expensive. There should be no need to get a quote; surveying 200 acres, 12 times a year, is clearly not a service that can be cheaply provided. 1 RGB camera and 1 NIR camera will be required. 1 Drone with a few (4?) spare batteries will also be required. The MVP would consist of just a drone with an RGB camera.
The drone should:
I also had a quick look at the software required for stitching the photos, and openDroneMap looks appropriate.
I had a chat to someone from Elk Fish Robotics, and based on my description they recommended the Phantom 4 Advanced for $1899, with batteries at the $289. The main reason for this over the Phantom 3 Advanced is future-proofing and range (up to 5 km), although reliability was cited as a potential concern, and battery life was a slight improvement (30 minutes). I found seeminly identical drones in the USA for around $1640 AUD, including shipping. I emailed the local suppliers, asking if they could negotiate on price. They came back with $1799 AUD, and pointed out that shipping would bring an additional 10% in GST, making the local offer very attractive. I'll show this offer to a local town supplier, and see if they can beat it.
A suitable NVDI attachment can be found here for $1783 AUD.
Yesterday, Jaycar in town offered $1829, including delivery. Now I need to solidify this plan so that the customer has confidence in my design.
Just for a reality check: Sentera = $4448 USD for P4 with NVDI + RGB, or around $2500 to upgrade a P4 you own. Leaves camera untouched; Dronenerds = $1999 USD for P4 with NDVI only, or around $1399 to buy NVDI camera standalone; link says use VARI as substirute for NVDI; more cameras at this shop are worth checking out.
The long story short seems to be that RGB photography will be fine to begin with. However, as was discovered painfully today, the information about tree health is useless without the ability to act on it. The spreader needs to be modified somehow to allow finely adjustable fertiliser spread. I'm not quite sure how to do this yet.
Found some good software here that I'll probably use. In the meantime, as I stated yersterday, I think it might be a long wait before this is worth spending money on.