By: Nabeel Khan

Fuel and fertilizer prices are at all-time highs, as farmers the world over are scrambling to find ways to stay financially afloat. Large commercial farms have long had a leg up over smaller holdings. They already have access to artificial intelligence (AI) and cutting-edge hardware to plan their operations and automate tasks. They run broad sensor networks to monitor soil quality, operate vision-enabled drones to monitor crop health, and let autonomous tractors steer across their fields to vastly improve their productivity.

These advanced systems have been prohibitively expensive, to the point that their costs are hard to justify for smaller holdings, less than two hectares in size, which make up the vast majority of the world’s more than 570 million farms. Many of these farmers simply don’t have access to sufficient capital to invest in new equipment, and continue to rely on older machines and human labor. For them, the vast promise of the digital revolution to increase the efficiency of agricultural operations — typically lumped together as smart farming or precision agriculture — has long remained just out of reach.

But change is in the air. Over the past few years, the popularization of the internet of things (IoT) and its underlying technologies has led to the development of a new and more affordable generation of precision farming solutions. These solutions are giving smallholdings tools to increase the quantity and the quality of their production. Combining satellite-based positioning, wireless connectivity and visual sensing with advanced algorithms including machine learning (ML) and artificial intelligence, these solutions promise to help farmers produce more for less, putting them on more equal footing with their larger commercial competitors.

In this article, we take stock of key trends driving the democratization of precision agriculture. We zoom in on some of the concrete applications transforming the operations of smaller farms the world over.  We start by exploring the core enabling technologies, the applications they enable, and where they are headed.

Technological Evolution on all Fronts

At their heart, precision agriculture solutions are sensor fusion platforms, taking in data points from a variety of sensors, using algorithms to make sense of them, and extracting insights on which their users — machine or human — can act. As they mature, these sensor fusion platforms are becoming increasingly complex, crunching more and more types of sensor data with algorithms of growing sophistication to gain ever deeper and higher value insights.

These insights are often generated using AI and ML models that run at the edge of the network near the sensors — on the tractors, sprayers, or other devices themselves — rather than in the cloud.

Equipment manufacturers looking to integrate AI/ML at the edge are weighing their options in terms of adding application processors and hardware accelerators capable of running advanced ML models needed to fuse camera and sensor inputs to make real-time decisions. To simplify the adoption of AI/ML at the edge, many vendors are starting to integrate AI accelerators into their modules and systems-on-chip (SoCs), lowering the barrier to entry.

Original equipment manufacturers (OEMs), for their part, are choosing to integrate these SoCs even if their software capabilities are still behind. For them, building solutions with future-proof hardware is a potentially lucrative strategy for generating recurring revenues through firmware updates that provide advanced functions down the road.

Despite the abundance of evaluation kits from major vendors, scaling from prototypes to production with these solutions can be expensive. While integrated sensor fusion platforms with, for example, pre-loaded dynamic vehicle models for auto-steering or computer vision models for plant health can accelerate development for OEMs and reduce the need for software investment, they can be too generic and may not solve use-case-specific needs. More focused AI/ML may require more investment in data collection and training models, which, due to their proprietary nature could come at a higher price.

Camera Systems

Camera systems are already widely relied on by autonomous tractors, visually monitoring the surrounding environment and feeding their data into computer-vision solutions where it is parsed. Context-rich, vision-based data can help optimize the distribution of agricultural inputs such as water, seeds, fertilizers and pesticides using real-time variable spray-rate control based on plant size and other metrics to significantly improve efficiencies and reduce overall costs.

Camera systems come with a set of challenges that need to be carefully managed. Lenses can become contaminated with water, debris, dust, and snow. Even though these issues impact all vision-based applications, including mass-market applications such as automated and autonomous driving, there are still no reliable methods of keeping them clean, aside from water-spraying nozzles (that can cause buildup on the lenses) and old-fashioned manual maintenance.

Additionally, the steep price of high-resolution cameras can drive up the cost of end solutions, as can data storage and communication when the visual data is processed in the cloud.

Global Navigation Satellite Systems

Satellite-based positioning has also become a staple in precision agriculture solutions. Autonomous and guided tractors use the technology to drive vehicles along precise paths to increase pass-to-pass efficiency and reduce overlapping rows. Crop monitoring drones use GNSS technology to patrol predefined flight paths. And fully autonomous guided vehicles such as robotic lawnmowers use it to avoid restricted no-go zones. The latter three applications typically require centimeter-level positioning accuracies.

Centimeter-level GNSS technology has been available for well over a decade, with farmers subscribing to relatively costly GNSS correction services tailored to each user. However, it was only with the advent of affordable RTK services with availability in rural areas, as well as the dramatic decrease in cost for farmers to deploy their own RTK base stations using low-cost modules, that the price point of high precision positioning has come down far enough to make the service affordable to all but the least lucrative operations.

High precision GNSS technology brings the benefits of improved pass-to-pass efficiency – a general reduction of all agricultural inputs with all the financial, environmental, and health benefits that this entails. At the same time, it requires solutions to deal with signal delays and the resulting inaccuracies of multipath effects, caused when signals bounce off buildings, mountains, or other solid structures on their way to the GNSS receiver.

When delivered via the internet, the GNSS augmentation data stream requires an IP connection to the provider’s server. This poses challenges for farms that lack infrastructures such as Wi-Fi base stations, sub-GHz RF systems, or cellular network coverage.

Wireless Connectivity

In some way or another, all advanced precision ag use cases depend on wireless connectivity. Environmental sensors and inspection drones need it to relay data to the cloud backend. Additionally, tractors, drones, and other farm robots depend on it to upload telematics data, report their status, enable predictive maintenance tools to reduce downtime, and receive GNSS augmentation data for high precision positioning.

While cellular connectivity is the easiest to use — all it takes to upload data straight to the cloud is a mobile data subscription and a SIM card — it has two key drawbacks. The first, mentioned earlier, is that the entire farm needs robust network coverage, which is not always a given in rural areas even in the United States and other developed countries. Also, when coverage is available, the cost of data transfers can quickly add up and become prohibitive for smaller, less profitable farms, as they may not be able to negotiate affordable connectivity plans as effectively as larger operations that have much higher data usage.

One way to address the coverage issue is by choosing the right wireless communication technology. Low power wide area technologies, such as LTE-M, NB-IoT, and Cat-1 solutions offer similar coverage to traditional LTE in many countries and are available at a fraction of the cost. When higher bandwidths are required, LTE Cat 4 or higher modules offer 150 Mbps+ of throughput. 5G modules enabling gigabit connectivity are available today, but solutions may cost up to 10-times more.

As an alternative, 5G Redcap – an upcoming 3GPP technology – seeks to provide a 5G compatible, affordable solution with medium bandwidth and lower complexity, enabling affordable hardware. We will likely start to see the first RedCap offerings from mobile network operators in 2025.

Artificial intelligence and machine learning at the edge offer an additional tool to reduce bandwidth requirements, by processing sensed data near the sensors themselves. Rather than streaming vast amounts of raw sensor data to the cloud, devices leveraging edge intelligence can reduce their wireless communication bandwidth requirements and cost by uploading only relevant information.

The Trends Driving Democratization

Progress in sensing and sensor fusion platforms, camera systems, GNSS technology, and wireless connectivity is but one of the drivers democratizing precision agriculture. The other is an ongoing ecosystem-wide transformation that is breaking down many of the barriers that have kept precision agriculture solutions the preserve of large, lucrative farming operations.

Together, they are bringing down the cost of ownership of precision ag technology. Only just a few years ago, the only available solutions from major OEMs came with a hefty price tag, comprising expensive hardware and considerable recurring subscription fees, while requiring skilled labor to implement, operate, and maintain smart equipment.

Vendor lock-in enforced by locking telematics interface ports prevented farms from adopting aftermarket solutions — justified as a way to improve safety and vehicle reliability. Ultimately, however, this limited farmers’ ability to maintain their equipment themselves and to piece together cost-optimized solutions tailored to their specific needs using components from competing solution providers.

Today, all of this is changing. Hardware costs are down dramatically due, largely, to economies of scale unleashed by the Internet of Things’ explosive growth. Farmers now have access to affordable, user-friendly aftermarket solutions to upgrade tractors and other agricultural machines they already own that depend on a new generation of more cost-effective GNSS correction services. OEMs are introducing advanced functionality in their entry- to mid-range tractors in addition to their premium product line.

At the same time, the abundance of open-source projects, module-based solutions, and pre-certified radio technologies are making the development of precision ag solutions cheaper, bringing down the cost of off-the-shelf hardware. Add to that lower subscription costs for GNSS correction services thanks to improved broadcast distribution with SSR-RTK (state space representation-real-time kinematic).

The total cost of connectivity is dropping as well. Deploying connectivity infrastructure was long a non-negligible cost point, made up of wireless infrastructure and data subscription fees. Today, thanks to the increased availability of cost- and power-optimized wireless communication technologies with broader coverage (LTE-M, NB-IoT, LTE Cat 1), farmers can reap the benefits of the overall expansion of cellular network infrastructure.

Arguably one of the most promising trends driving the democratization of precision agriculture technology comes from the farming community itself in the form of do-it-yourself solutions based on open-source hardware and software designs.

Take AgOpenGPS, also referred to as AOG, an open-source auto-steering solution created by a Canadian farmer and software developer. AOG delivers all the hardware design files, the real-time microcontroller firmware, and software required to enable auto-steering on conventional tractors, regardless of their age. Thousands of hours invested by the AOG developer community have made the solution accessible to the masses, both financially — retrofitting tractors can cost less than USD $1,000 — and in terms of prerequisite knowledge.

A growing variety of precision agriculture platforms, from high-end commercial solutions that cater to the most profitable farms to low-cost do-it-yourself solutions that can pay for themselves in a year and are accessible to smaller, less profitable operations, are transforming the impact that precision agriculture can deliver. With the global food supply under tremendous pressure, this democratization of smart farming technology could come to play an important role in feeding the world.

To learn more about how u-blox enables high-precision autonomous vehicles and other smart farming applications, visit www.u-blox.com/precision-agriculture.