Construction of drones for crop monitoring

Building a drone for use in field research considerably reduces costs, allowing countless possibilities for adaptation and programming according to the user's needs.

The use of Unmanned Aerial Vehicles (UAVs), popularly called drones, in agriculture has been attracting the attention of farmers, field technicians and researchers due to the countless possibilities of use on rural properties. By collecting images, simple visual assessments can be carried out, thus allowing the identification of planting flaws, fertilization problems, stains within plots, as well as errors in the application of phytosanitary products. At a more careful level of analysis, drones can also act in obtaining georeferenced images of rural properties, allowing the management and sizing of cultivation areas, areas of permanent preservation, which previously could only be done through the use of images obtained with assistance from planes and satellites at a considerably high cost.

Through the images collected with the aid of the drone and the use of geoprocessing techniques, topographic maps can be generated to create irrigation and drainage models, leaf area index, level of pest infestation, as well as nutritional deficiencies. Due to the diversity of situations in agriculture, images collected in a relatively short period of time allow for faster processing, anticipating diagnosis, decision-making and interventions in crop management.

There are different models of drones on the market and the most varied prices, and the differentiating factor will be the type of use and monitoring to be carried out. One of the obstacles encountered by users who intend to collect data in the field is the high cost of this equipment due to the particularities and degree of demand, both of the equipment and the level of precision of the images to be collected.

Accuracy in image collection is directly related to the quality presented by the drone, mainly due to its flight stability and autonomy, as well as the system used to capture images. Building a drone from existing components on the market is a low-cost alternative, especially when carrying out research work, which allows the equipment to be readjusted and programmed as needed.

To carry out more complex analyzes and diagnoses, simply collecting images of cultivation areas does not allow us to make more precise inferences, and for this, images in the visible and infrared spectrum are necessary. After collecting several images, it is necessary to generate a mosaic, obtaining an image of the entire area considered the object of the study, which is intended to be analyzed in a later stage. By using different spectra and simple calculations, more reliable parameters such as NDVI (Normalized Difference Vegetation Index) and NDWI (Normalized Difference Water Index) can be obtained. NDVI makes it possible to differentiate more clearly the chlorophyll level of plants, so that, for example, areas with insect attacks or diseases will be easily highlighted in the images.

On the other hand, NDWI is an interesting resource for irrigated rice cultivation areas, allowing for the analysis of irrigation failures through the absence of water depth in crops. Much work at an advanced level has been carried out seeking to correlate the NDVI related to the chlorophyll content, diseases and nutritional deficiencies of cultivated plants, however, further studies are still needed to determine the ideal flight altitude, as well as calibration and model validation.

In this context, the Laboratory of Intelligent Systems and Modeling (LabSIM), belonging to the Federal University of Pampa – Campus Itaqui (RS), has been carrying out several researches focused on the agricultural area regarding the acquisition of data and the development of low-cost alternative technologies. The construction of a quadcopter-type drone, the target of research carried out in the laboratory, basically consists of hardware, firmware and software.

The hardware is the physical structure of the set, consisting of a microcontroller capable of following commands, as well as receiving readings from sensors and sending commands to the propulsion engines. Among the sensors used, there is an inertial sensor composed of an accelerometer and gyroscope, both responsible for the stability and balance of the set, a barometric pressure sensor and a sonar, used to control altitude. The GPS and magnetrometer (compass) are responsible for the positioning and orientation of the drone, and a transmitter and receiver module is also required for the telemetry system, which allows communication with the control station, and a radio control for manual operation. The microcontroller used is an Arduino Mega 2560 equipped with a board containing all the sensors, with the (quadcopter structure) was built with aluminum tubes and sheets. The propulsion system consists of electronic speed controllers (ESCs), motors brushless engine and propellants (propellers).

The structure allows the coupling of a camera for image collection and a gimbal, a mechanism that allows the camera to always be kept at a vertical angle during flight.

The firmware consists of a set of programmed instructions loaded into the microcontroller, responsible for managing all internal operations, such as reading sensors and controlling motor power. The MegaPirateNG firmware is used, which is open source and compatible with all sensors in the set.

The software is installed in the control station (notebook or tablet), where it receives flight information, as well as the definition of parameters, calibration and flight plan, if the system is autonomous. With the autonomous system, it is possible to draw up the flight plan using georeferenced maps using Google Maps. Furthermore, the actions that will be performed, such as image capture, speed and altitude changes, are configured according to the user's needs.

When the user only chooses to build a drone to carry out visual analysis, based on images of the cultivation areas, using a compact camera for this purpose, the cost of the equipment with the added components is approximately R$2. However, when the research to be carried out requires images that allow for detailed analysis and with a more accurate level of precision, it becomes necessary to use special cameras (visible and infrared spectrum), whose prices can easily exceed R$ 10 thousand.

FINAL CONSIDERATIONS

Using common electronic components on the market, it makes it possible to build a drone according to need, being a viable alternative for carrying out research due to its versatility and ability to adapt to different equipment and sensors. In addition to these advantages, the low cost of the components is also mentioned, which is an extremely important factor, especially in periods of scarce resource availability, thus requiring the development of new technologies and alternative systems.

 

Carlos Alexandre Romani, Paulo Fernando Escobar Paim, Alexandre Russini, Rogério de Vargas Rodrigues, Cristiano Galafassi, LabSIM, Federal University of Pampa - Campus Itaqui

Article published in issue 163 of Cultivar Máquinas.