A personal view.
One of the main precision agriculture tools has been the UAV (Unmanned Aerial Vehicle), more commonly known as the drone. Indeed, its ability to zip around on data collection missions has come to symbolize hi-tech efforts to find better, more sustainable solutions to agriculture.
The principle on which drones are deployed is sound enough. The remote monitoring and mapping of crop condition so that farmers can make more informed decisions is exactly what precision agriculture is all about. When such data is used to adjust fertilizer prescriptions, for instance, inputs (and therefore costs) can be saved and many more environmentally friendly smiles can be confidently shared by all those concerned.
I’ve always had a passion for aviation and agriculture, and thus was very excited to be amongst the earliest innovators that used UAVs in farming. Starting my work prior to commercial drone regulations and even prior to commercially available drones, I have watched the industry go from every single UAV having to be painstakingly hand-built to the modern purpose-made agricultural models available today.
Yet, even with the massive strides in UAV quality and performance, we still do not see the widespread adoption of this technology in a commercial ag setting. Look around and you'll soon realize that the same can easily be said for satellites and ground-based active sensors. The sad fact is that of the millions of acres tilled in the various countries around the world, precision techniques are only employed on a percentage that ranges somewhere between 'minute' and 'minuscule'.
This is not to detract from the value offered by drones. Certainly, they play an important role in agronomic research and agricultural science. On a research plot requiring hundreds of measurements that could never all be done manually, drones are ideal. Moreover, drones have proven to be indispensable to many farmers, not least because they offer access to places restricted to conventional machinery, perhaps because of tight spaces or because a particular area is too wet and muddy.
However, the question which must be asked is: 'What would make a remote sensing system more useful to production agriculturalists?' To my mind, the criteria must be: 'A system which provides timely, accurate and actionable results at a price point that offers a solid Return on Investment (ROI).'
Until now, this has been an order too tall for drones, satellites and active sensors. The main reason being that while each precision tool may be good or even excel at one or more criteria, they have all failed to offer an effective all-round package which is easy-to-use, logistics-free and cheap.
But I am getting a little ahead of myself. A closer look at the fundamentals is first called for. For remote sensing to work its wonders and save the farmer and/or agronomist the drudgery of physical inspection, the sensor - in whatever form or shape it may be - has to offer an appropriate balance of four types of resolution:
Let's deal with them one at a time.
The spatiality of precision tools and their ability to detect detail varies greatly. While offering far higher resolutions than spaceborne satellites, drone resolution is cost dependent, with better resolutions coming at a premium. This means, for example, that no satellite tasked to agriculture can detect weeds, but nor can many of the drones in actual operation. Ground-based active sensors tend to have very high resolution when compared to satellites, but this comes at a cost, as we shall later see.
The spectral abilities of a sensor are important in making meaningful vegetative indices, such as NDVI (Normalized Difference Vegetation Index). These are determined by measuring the relative amount of light absorbed by photosynthesis and the reflection of NIR (Near-Infrared) light by leaf biomass. They indicate plant health and can be used to help determine such things as nitrogen deficiency, senescence (aging), spatial patterns of disease, pest infestation and so on. Accuracy is key in the spectral stakes, with equipment cost normally dictating whether NIR can be correctly distinguished between RedEdge and red wavelengths in the spectrum, for instance.
Radiometrically, a sensor has to be able to detect subtle differences in reflectance. Apart from improving NDVI accuracy, it will also allow it to discern if a sub-nitrogenous plant has enough nitrogen to reach its maximum yield potential. Again, sensor capability is determined by cost. Typically, only drones costing $15,000 USD or more have this capability.
Temporal resolution is the revisit time. How often does a sensor have to be deployed in order to achieve a useful dataset? How easily can a sensor be deployed, especially seeing that many agricultural practices are so time sensitive and circumstance dependent? As satellites and drones largely rely on weather, they perform poorly here. A week or just a few days of cloudy or windy conditions, for example, could very easily mean a farmer missing an application window and compromising an entire season. Moreover, drones require a particular set of logistics, which further limit their temporal abilities.
The problem with these four competing parameters boils down to the fact that due to technological and financial constraints, as a sensor is made better at one resolution type, it tends to lose ground on one or more of the others. To exemplify, a sensor with very high spatial resolution will usually have a very narrow field of view, which is the electronic equivalent of having tunnel vision. This is a particular foible of ground-based sensors which may have a very high spatial resolution of less than 1 cm per pixel, but only have a field of view of 2 m or less. This severely compromises temporal resolution because it increases the required data collection time.
If only there was a way to reconcile these conflicting demands - to have resolution enough to resolve all the problems.
Which brings us to the effective all-round package I hinted at earlier. It is the world's first camera-based, real-time VRA (Variable Rate Application) control device known as the Augmenta System. It's an unobtrusive plug n' play ground-based retrofit mounted on the cabin roof of an ordinary tractor (with a hitched/trailed VRA implement) or self-propelled VRA spreader/sprayer (there is also a provision to actuate non-VRA spreaders).
An array of 4K multispectral cameras have a 40 meter (130 ft) 'scanning' width while achieving a spatial resolution of 12 pixel/cm as they look down on the crops ahead while the farmer drives along. Powerful machine vision techniques and AI algorithms create a vegetative index to create an on-going prescription map. This, in turn, is used to control the VRA equipment to intelligently optimize the application of nitrogen fertilizer, for example, to make crops more uniform, increase productivity and avoid waste.
Being camera-based, it really does tick all the resolution boxes. It has the spatial resolution to address all relevant field problems, including weed detection (a Green on Brown selective spraying service is to be offered by Q4, 2021). Moreover, it can accurately distinguish no-growth areas such as water-logged or rocky field patches and not apply inputs. Active sensors controlling the VRA application of nitrogen would detect little to no chlorophyll reflectance in such areas, and therefore erroneously (and wastefully) apply the maximum amount of nitrogen. Furthermore, it has both the spectral and radiometric resolutions to make its own highly accurate vegetative index (the AUG-Index) and determine nitrogen levels in sub-nitrogenous plants so they can reach their full yield potential. Importantly, being ground-based and logistics free, it is characterized by high temporal resolution and offers short and reliable revisit periods.
Nonetheless, all the resolution in the world is utterly useless, unless the data can be accurately interpreted to produce an actionable result. This is precisely where the Augmenta System flies in the face of drones and satellites. It has robust AI self-learning algorithms and the considerable computing power necessary to automatically turn raw data into an intelligent varied rate of input. There is no field collection wait, no data processing time, no need for RX file or shapefile generation (although these are also catered for), no delay whatsoever. It is all instantly actionable. Almost incredibly, the entire Augmenta Farming process is fully-automatic, done in real-time and in one hassle-free pass.
Irrespective of the size of their operations, this AI algorithmic approach makes far more financial and practical sense for farmers than hiring remote sensing and agronomy experts. The Augmenta System provides ease-of-use and up-to-the-minute accuracy – remember that the prescription maps generated by drones and satellites are old data by the time they are applied, whereas the Augmenta System operates in real-time. Additionally, because Augmented Farming makes substantial cost savings on inputs as well as increasing yield, it offers an enviable ROI.
A simple comparison of typical workflows makes its own point in terms of workload and time efficiency:
Drone workflow for a single person
Day 1
Scout the field to ensure that the flight can be done in accordance with all local aviation (e.g. FAA) and any other governmental regulations.
Time spent: 10 minutes.
Charge batteries.
Time spent: 1 hour per battery
Fly drone over field.
Time spent: Most multispectral flights will take 1 hour to collect multispectral data for 160 acres.
Transfer data from sensor to PC.
Time spent: 15 minutes.
Day 2
Process the drone data.
Time spent: 30 minutes to several hours.
Upload processed data to GIS (Geographic Information System) and add other data layers.
Time spent: 30 minutes to several hours.
Construct and export the prescription map (as an RX file/shapefile).
Time spent: 10 minutes.
Day 3
Import RX file/shapefile to VT/Monitor in tractor or self-propelled spreader/sprayer.
Time spent: 5 minutes
Setup VRA spreader/sprayer.
Apply VRA to the field.
Satellite workflow for a single person
Day 1
Find satellite data for your field and wait for processing and download.
Time spent: 1 hour or more if not automated.
Upload satellite data to GIS and add other data layers.
Time spent: 30 minutes to several hours.
Construct and export prescription map (as an RX file/shapefile).
Time spent: 10 minutes.
Day 2
Import RX file/shapefile to VT/Monitor in tractor.
Time spent: 5 minutes
Setup VRA spreader/sprayer
Apply VRA to the field.
Augmenta workflow for a single person
Day 1
Setup VRA spreader/sprayer
Select application type, crop type, fertilizer and maximum rate (/recommended dose) using the in-cabin Augmenta Tablet.
Time taken: 30-45 seconds.
Apply VRA to the field.
Go home early!
As can be seen, the Augmenta System literally reduces what takes other systems multiple days to accomplish into mere seconds.
There are other considerations which cause serious problems for drones in commercial ag settings. Currently drones add multiple layers of complexity to an operation, regardless of whether a farmer hires out the flights or does them in-house.
Hiring a third party entails phone calls, scheduling and cost (drone companies are also businesses which operate on margins) each and every time a flight is required. In most countries, in-house flights require the farmer to have a UAV pilot license. This requires studying, a formal examination and, at least in the USA, retesting every two years. Then there is the hassle of taking time out of an already busy schedule to fly the fields, remembering that they have to be done on days with suitable weather conditions (i.e. less windy days, especially when using a spray drone), as scheduled by weather forecasts which are not exactly famed for their accuracy. Windows of opportunity are often so narrow that they do not allow enough time to collect and process drone data prior to a flight.
The drone pilot is responsible for the UAV at all times. In the USA, it can't be flown over 120 meters (400 ft), leave the pilot's line of sight, nor is it legal to fly in all airspaces. Approval to fly in the airspace above your farm must be granted by the Federal Aviation Administration first. The airspace must also be physically checked pre-flight. Aside from avoiding all general aviation aircraft, pilots must also avoid crop dusters and pipeline/powerline inspection planes that often fly in the same area and at similar heights to the drone. This increased interaction with low flying aircraft greatly increases the odds of a midair collision. Moreover, maintenance standards are high to avoid malfunctions and crashes. Even a light crash can render a drone useless for days or even weeks - spelling disaster for time-sensitive farming operations which rely on drone data collection.
In comparison, Augmenta Farming is entirely headache free. No prior planning is necessary other than the normal set up and loading of the tractor/implement or self-propelled rig. Because it is permanently on site, you can collect data every time you take your tractor into the field, for whatever reason that might be. There are no extra regulations, no governmental bureaucracy (in fact you can get rebates for sustainability), no licenses or exams. There is no maintenance to speak of, other than wiping the camera array window clean now and again, and an occasional firmware update (which is done automatically anyway) via a built-in 3G/4G mobile/cellular connection. And unless you are particularly fast behind the tractor wheel, absolutely no chance of a midair collision!
With this groundbreaking innovation, I believe that practical precision agriculture has finally arrived. All the farmer has to do is specify the operation type with a few taps on the in-cabin tablet, turn the ignition key and head for the fields. Nothing could be easier or more effective.
Check out Augmenta System performance by investigating these results from Australia.
We want to keep you informed! Read all about the Augmenta System, the company and the impact we are having on revolutionizing farming.