Applications or the use of “drones” (more technically, UAV - Unmanned Aerial Vehicle) in agriculture have shown a strong growth trend. This trend may be much faster if we consider the countless potential applications still under development.
It is worth pointing out that when we talk about “drones” we are using a generic term that includes several components and systems, each used in one of the phases of the information generation pipeline (Figure 1).
One of the components is the UAV itself, which can be a mini-helicopter with a single propeller on a vertical rotation axis or a multirotor with 4, 6 or 8 propellers, in addition to those similar to airplanes, called fixed-wing “drones” (Figure 2). Each type and model has its advantages and disadvantages. Multirotor UAVs can generally fly at lower speeds and lower altitudes than fixed-wing ones and are more stable and safer in stronger wind conditions. Another important feature that sets UAVs apart is their “payload” capacity (or the maximum load they can carry), ranging from less than 200 grams to a few dozen kilos. In addition, there are UAVs equipped with an inertial system, GPS, one-way or multi-directional obstacle sensors, assisted landing sensors, among others.
Another component is the Control and Communication System (CCS), the so-called “ground stations” (Figure 3), which includes communication between the base device and the UAV, the software for planning and automatically controlling flights, as well as countless settings and adjustments for flight safety and energy consumption parameters. There are dozens of “ground station” solutions, but the recommendation is to adopt the one developed and supplied by the manufacturer of your UAV.
Undoubtedly, the “payload” component has become the part that gets the most attention from the market, since it is in this part and in image processing that the various applications are customized. What could go on board a UAV? The possible answers are unimaginable and diverse: from digital cameras to “sprayers”, active sensors (microwave), passive sensors (in waves visible and invisible to the human eye, including the infrared regions), a wide variety of environmental sensors, seeders, among others.
When digital cameras or sensors are used on board a UAV to collect images or videos, the Image Processing Systems (IPS) component is mandatory (Figure 3). Here, too, there is a wide range of solutions on the market. This system component allows the user to stitch hundreds or thousands of digital images into a single image of the field or experimental plot, generally called an orthorectified mosaic. Each image captured by the camera or sensor from a UAV is affected by axis rotation caused by wind or thermal currents during flight, as well as by differences in sunlight intensity throughout a mission. IPSs contain algorithms capable of correcting these distortions and generating the unified image of the target area.
Perhaps the pioneering application in agriculture was visual monitoring using images (static or dynamic) captured by digital cameras on board UAVs. Without a doubt, the speed and quality of UAV monitoring surpass that carried out on foot or with ground vehicles.
But an important step was taken when images captured at wavelengths not visible to the human eye began to be used. The use of cameras that capture images in the infrared region made it possible to generate orthomosaics showing the spatio-temporal variation of a vegetation index across a field (Figure 4). This correlates with the variation of some soil elements, but also with plant health aspects, making it possible to infer the occurrence of diseases and their location in the field. This data allows, for example, generating a spraying recommendation map. A UAV carrying a sprayer and guided by the recommendation map can be directed to apply only in the areas that need it.
The market's next step was the use of multispectral sensors that capture images in the region visible to the human eye and, at the same time, in two or three bands in the near- and mid-infrared regions. This opened up the possibility of checking variations in vegetation indices across a field related to water stress, as well as improving the quality of information related to biotic stresses. There are dozens of vegetation indices that can be calculated from images captured by multispectral sensors, some more suitable for one species than others. For each plant species, there may be more suitable vegetation indices at each crop stage.
Add to this applications for measuring plant height (Figure 5), counting plants and estimating population (Figure 6),
It is worth highlighting here the growing importance that hyperspectral sensors will assume in applications related to agronomic research and plant breeding. Hyperspectral sensors capture images across the band visible to the human eye, but also in the near- and mid-infrared bands, and can capture up to the far-infrared and microwave bands. To draw an analogy: the evolution from the first images captured in the infrared to those captured by hyperspectral sensors would be similar to the evolution from an X-ray image to those produced by color three-dimensional magnetic resonance imaging.
This class of sensors will soon be responsible for enabling large-scale phenotyping for the breeding programs of many species around the world (Figure 7), at costs thousands of times lower than current solutions.
But before flying a UAV, it is important to be aware of the related regulations. Very recently, ANAC (the Brazilian civil aviation agency) published a normative instruction. The most important restrictions include, but are not limited to: register your UAV with ANAC; never fly higher than 120 meters above the ground; flights over groups of people must be carried out at a minimum height of 30 meters and with the written authorization of those people; UAVs with a “payload” above 25 kg require a license equal to that of airplane pilots, after a course at the Department of Civil Aviation.
The map appears after you accept cookies.
Restricted Area
