In recent decades various social sectors have been revolutionized by the advance of digital technology and more recently by the advent of so-called intelligent systems. Each day we benefit in a quite tangible way by the digitalization of the communications, medical, banking and entertainment sectors. Digital signal processing has several applications and possibilities, even within our daily lives already immersed in this reality. In many sectors the effects of digitalization may not be so apparent, but they are equally present. For example, what is the relationship between digital agriculture and the foods that we eat every day?
The era of digital animal raising
When we think of agricultural technology, perhaps some images that come to mind are of irrigation systems, large tractors, planters and harvesters with complex mechanical systems, and not images of computers, sensors, robots and drones. However, for some time now, a true revolution has been taking place on farms, marked initially by the adoption of information systems that modernize agribusiness management. In a second moment, the evolution of the Geographic Information Systems (GISs) and remote sensing, mainly based on satellites images and aerial surveys, allowed to consider the spatial and temporal variability of soil and weather related to harvest management, no longer treating an entire crop in a uniform manner. The objective became to apply agricultural inputs at variable rates, “in the right amount, in the right place and at the right time”.
Today, digital agriculture uses technologies such as digital signal processing, satellite positioning systems (of which the best known is GPS), intelligent sensors, computer vision and robotics, which allow improved adaptability to local conditions, even meter by meter. The process efficiency attained provides cost reductions and significant gains in productivity and quality and added value to agricultural products, as well as greater sustainability and lower environmental impact. These technologies are commonly known as “precision agriculture” or “digital agriculture”.
Two (or more) intelligent sensors “think” better than one
Much of the information that feeds digital agricultural systems comes from sensors that measure physical quantities that are important for controlling crop cultivation and animal raising, such as sensors of soil humidity and incidence of sunlight on crops, temperature sensors in poultry aviaries and many other types. In digital agriculture the measurements made by the sensors are typically not collected manually, but with telemetry, because the sensors can include embedded processors and communication systems that make them “intelligent”, allowing them to locally conduct part of the data processing and activate the controls and alarms, and also allow them to be connected to networks. These networks of intelligent sensors use principles of metrology to guarantee the reliability of the measurements, digital signal processing for data handling, technologies of the Internet of Things for connecting with the network and of Big Data for analysis of the information contained in the large volumes of data generated. As examples of projects the CERTI Foundation has executed in this field we can cite a network of intelligent sensors for environmental monitoring and a water quality monitoring buoy, which can be used in water reservoirs and in aquiculture.
From grain to grain
A type of sensor broadly used in digital agriculture is the image sensor, found on cameras installed in fields, plant nurseries and pastures, embedded in processing machines, or even in airplanes and drones, or in cell phones. The images are processed with digital techniques that usually include algorithms for filtering, segmentation, classification and identification of patterns, with multiple purposes, such as identification of pests, diseases and nutrient deficiencies, preparation of maps of productivity and tracking objects. Digital image processing is also applied in machinery for post-harvest processing, and in automatic inspection, selection and classification systems for grains, fruits and vegetables and other agricultural products, detection of defects and of chemical or biological contamination and food security in general. One type of equipment that uses image processing are grain processing machines that can examine and select or classify grains one by one at a speed of hundreds of grains per second, according to criteria of color, size, shape or texture.
What the eyes can’t see, is felt in the pocketbook!
In addition to conventional and photometric cameras, digital agriculture also uses radiometers and cameras with multi- and hyperspectral sensors capable of detecting electromagnetic radiation with greater spectral resolution and at wavelengths not visible to the human eye, such as ultraviolet and infrared light and even X-rays. This capacity is of great interest to agriculture, given that many characteristics intrinsically related to productivity, costs and production quality are imperceptible to the human eye. Characteristics of interest include the moisture content in the soil and plants, plant vigor, pasture quality, and the presence of agricultural diseases or pests.
It is in this context that the application of digital processing techniques to the multi and hyperspectral images collected by sensors on remote platforms such as satellites, airplanes and drones stands out as one of the main tools for analysis and support to digital agriculture. The development and application of techniques for processing images collected in ranges outside the visible spectrum allow the preparation of various indexes for precocious identification of anomalies, as well as taking corrective actions during the production process, thus guaranteeing better performance, reduced costs and reduced water consumption. In addition, given that the cameras and radiometers are usually embedded in remote platforms, they allow a synoptic view of large production areas, with standardized data collected systematically, generally at a lower cost than collections made on site.
In animal raising, thermal images generated by infrared sensitive cameras can be used, for example, to monitor the metabolic activity and thermal comfort of animals in nurseries, automatically controlling air conditioning systems and for precocious identification of sick animals in herds, contributing to containing the spread of infectious and contagious diseases, such as foot-and-mouth disease. Other examples of non-conventional images are those from fluorescence, used for chemical analysis, and X-rays and magnetic resonance, which can be used to identify internal infestations and other deteriorations of products not detectable by external inspection, although their use is limited to specific applications that justify the higher cost and lower velocity of imaging.
Agrobots and other surprising machines
In addition to the technologies described above, robotics and autonomous and semi-autonomous vehicles are increasingly present in digital agriculture. Agricultural robots (or “agrobots”) are being developed for planting; applying pesticides, herbicides and fertilizers; pruning; and harvesting, either in fields or in greenhouses, and some are now commercially available. Because they are smaller and more agile, the agrobots can avoid the soil compacting caused by heavy machinery and conduct operations with greater precision, such as identifying weeds and precisely applying herbicides only on them, or evaluating the degree of maturing of individual fruits, removing only those that are ready for harvest. Drones are also being increasingly used in husbandry in functions that range from capturing images described above, and to release insects used as biological agents to control pests afflicting certain crops.
Autonomous agricultural machinery typically use GPS to follow pre-programed trajectories, as well as cameras, sonar, radar, lidar (light-based imaging) and other sensors to identify and avoid obstacles found along the route. There are also semiautonomous machines, which can be used with automatic pilot, but that need supervision from an operator to deal with more problematic trajectories and unforeseen situations. Among CERTI’s fields of competence for the development of advanced agricultural machinery are metrology and instrumentation, the design and execution of mechatronics, the development of embedded systems and digital image processing, employed in computer vision systems.
