Agriculture is rapidly emerging as a major vertical use case for 5G and the IoT, driven by need to feed dense populations, meet goals for environmental sustainability, and improve efficiency in the face of skilled labor shortages, among other factors. The term vertical farming has emerged to describe 3D production of crops or horticultural plants stacked up indoors in greenhouses, but could apply to an industry that will increasingly employ integrated IoT across its various domains.
Although sometimes derided as primitive, agriculture has long been quick to employ mechanization, as in combine harvesting and then more recently automation. US agro and construction equipment maker John Deere prototyped its first self-driving tractor in 2003 well over a decade before first demonstration of autonomous cars on public roads. Since they operate mostly on private land without encountering other vehicles or pedestrians, tractors and other agricultural vehicles such as harvesters are ripe for autonomous operation, reducing need for semi-skilled labor often subject to seasonal peaks and troughs in demand.
More recently accessories to tractors such as seed planters and ploughs have started to be equipped with sensing devices to monitor and optimize autonomous operation in the field. At the same time, LPWAN protocols and then 5G have administered a further shot in the arm to smart agriculture. Combined with IoT advances, this is expanding the scope of field sensing, allowing greater precision and facilitating autonomous operations calling for ultra-low latency, with prospect of more widespread use of drones.
The latter will in turn increase scope of sensing further, combined with intelligent dispending of seed, fertilizer or pesticide, as well as employment of AI algorithms for monitoring and detection of impending problems, like encroachment of weeds or environmental stress caused by drought. Advances in robotics also come into play, notably through automated picking machines that are becoming increasingly capable of plucking more delicate items, such as flowers and soft fruits like strawberries. This is in addition to the lower hanging hard fruits like apples and peers more amenable to automatic picking.
Perhaps most exciting are the opportunities for more integrated smart farming to help improve productivity, sustainability, quality and cost efficiency all in one, while catering for the reality that it will be increasingly hard to find humans available or willing to perform some of the more menial tasks involved. The stage has already been set by leading edge examples and most particularly by one or two forward looking nations such as the Netherlands.
That country became a pioneer and almost evangelist for high tech indoor farming around the turn of the Millennium, since when despite being small and densely populated the country has become a significant net exporter of many agricultural and horticultural commodities, from tomatoes to tulips. Remarkably, the country has become the world’s second biggest exporter of food by value after only the United States, despite being about 240 times smaller by land area.
This was achieved through deployment of the most advanced techniques available for intensive and yet as far as possible sustainable indoor farms, often close to centers of dense population as easily seen from aircraft approaching Amsterdam’s Schiphol airport. Plants are often stacked in rows under glass illuminated by rows of LED lights and grown hydroponically without soil, using mineral nutrient solutions in an aqueous solvent. The use of hydrothermal energy in many cases has enabled production of crops or plants out of season, or even types that normally only grow in warmer climates generally.
Even so, the country would admit there is still further progress that can be made, certainly in quality if not yield, noting that Dutch tomatoes are not renowned for their flavor. This is where automation and big data can play a role, according to David Rosenberg, founder of Aerofarms in Newark, New Jersey, US, which claimed to be the largest vertical indoor farm after moving to new premises in 2016, specializing particularly in green vegetables.
Rosenberg argues that by analyzing up to 130,000 data points derived from leaf images, pertaining to height, width, length, stem ratio, spotting, tearing and color, it is possible to achieve an optimum balance between production, nutritional value and flavor. He indicates that the latter is almost the most important quality as a source of competitive advantage.
Similar approaches are being employed in the Netherlands, where the initial greater emphasis on productivity and sustainability driven by pressure to exploit a limited land area is increasingly being supplemented by quality. The latter inevitably can require compromise over yield, with selection of appropriate varieties and harvesting at slightly different times.
But it also includes sound management on the ground, where again automation can play a major role, as Dutch telco KPN has demonstrated with ecosystem partners through an IoT solution to automate and optimize elimination of unwanted potato plants from fields of sugar beet. Potatoes have an annoying habit of popping up for many years in fields where they were originally grown as crops, a problem that has long dogged farmers in the Netherlands as elsewhere where they are grown. The problem arises through practice of crop rotation to maintain soil nutrient levels and reduce need for insecticides or herbicides to control pests or parasites.
KPN has been testing robots trained through deep learning techniques to recognize and destroy rogue potato plants in fields of beet, or potentially other crops. The capacity and low latency of 5G was needed to process the images in real time at the network edge quickly enough to ensure effective operation in the fields. KPN is claiming that 95% of the rogue potatoes are removed this way with minimal damage to crops, while covering one third of a hectare per hour. This is far faster than manual workers, such that one robot would clear a typical large field in about a working week.
Agricultural drones in particular will be fecund ground for 5G because of the huge potential in spraying, surveillance and monitoring, integrating with and augmenting sensors in the field. There are already a number of pilots involving both operations and infrastructure providers. One has been rolled out recently in the Brazilian state of Goias, a savannah area dominated by large scale farming, aiming to enable fast action against disease over big areas, using 5G equipment supplied by Huawei running over a 5G network built by Claro, part of Mexico’s América Móvil.
This is being tested for soy farming, collecting data from sensors on the ground, as well as in harvesters and drones, which is then crossed with meteorological data, chiefly temperature and humidity. The aim is to identify rapidly actions that need to be taken, such as spraying sparingly with herbicide or pesticide, balancing productivity against environmental sustainability which has tended to be lacking in that region.
Not all agricultural IoT applications require full 5G performance and there will also be a major role for LPWAN protocols as part of the mix. Indeed, there already is, with NB-IoT featuring more strongly than LTE-M from the cellular stable. LoRaWAN though has made the greatest impact in the agricultural sector, largely as a result of promotional efforts, although it fits the requirements for long range and low power consumption, being suitable for many remote farms where cellular signals are unreliable or unavailable.
A recent paper published by Central Queensland University in Australia gave an upbeat assessment of LoRaWAN in farming after evaluating its performance within a smart sprinkler systems, finding that it offered cost effective IoT communications over farms up to 10 kilometers or a little more across, avoiding the need for LTE or other networks for backhaul, while keeping power consumption low enough for local solar generation.