While the terms 'big data' and 'artificial intelligence' often get thrown around when talking about generalized industry progress, their integration into smart agriculture over the last decade has made a tremendous impact. The shift is driven by advanced edge computing capabilities, more efficient sensors and wireless communication, and structured, labeled data for model training.
Big Data in Agriculture: A Two-Part Approach
Big data plays a variety of roles in smart agriculture, which can be split into two distinct categories — data acquisition and model training. While the applications of either methodology are nearly endless, their impact on the progress and efficiency of agriculture is dramatic.
Smart Agriculture and Big Data Acquisition
Sensors have become smaller, more beneficial, and more energy-efficient. While a single sensor may not produce a significant amount of data, an array of edge nodes containing a multitude of sensors distributed across several acres of farmland produces vast amounts of data. Big data can be used in a variety of actionable ways. It creates dashboards, conditions monitoring infrastructures, and provides farmers with intelligence that would never be available at the scale big data processing allows.
Big Data Applications in Agriculture
Let's look at a real-life example of the application of big data. Yanmar, a Japanese manufacturing company, set out to create a sustainable 'smart' greenhouse infrastructure to increase process efficiencies. The original architecture of Yanmar's smart greenhouse network is shown below. Cameras objectively track plant growth. Yanmar then utilizes Amazon Web Services to document, track, and analyze this video feed to understand the various stages of a crop's progress. Once certain growth milestones are achieved, Yanmar uses its newly found data intelligence to intuitively adjust the water and nutrient balances that the plants receive to optimize their overall growth. This pattern of growth tracking and nutrient balancing allows Yanmar to create tangible data around which method of agricultural nurturing is best and develop new methodologies for farming specific types of plants while maintaining resource-conserving practices.
All in all, Yanmar's smart greenhouse boils down to one critical technology — big data. Yanmar utilizes loads of data that require even more staggering amounts of processing to understand the data's values.
The first facet of big data involves the use of cameras, which generate massive amounts of data. Transferring this data can be costly, as it may require high bandwidth and large amounts of off-site storage. Next, processing this vast amount of big data requires substantial computing abilities, which requires a capable infrastructure.
With this infrastructure in place, the farming process can be easily optimized and controlled with minimal effort. Yanmar's solution is one of many that have built smart agriculture aimed at increasing efficiencies and (literally) growing the industry in a way that has never been done more intelligently. There is, however, another critical factor that enables Yanmar's success — the implementation of artificial intelligence in the processing of its big data.
AI in Agriculture
The classic example used to explain deep learning, a sub-technology of AI, prescribes a methodology used to get a simple camera to interpret handwritten text. This example illustrates the vast amount of data required to train a simple deep learning model to be semi-successful. While possibly an oversimplification of what AI is capable of, getting a computer to understand written numerals can be very complex. The smart agriculture industry takes this complexity in stride and utilizes AI in ways that prove extremely useful, such as identifying critical growth stages of plants, as in Yanmar's case.
Application of AI in Agriculture: Plant Identification
An even more effective application of deep learning in agriculture can be found in Blue River Technologies' use of cameras to identify different types of plants. This AI application is implemented on tractors that distribute fertilizer, pesticide, herbicide, and even water. Cameras and their associated edge computers analyze the ground as a tractor is moving, identifying and classifying plants and insects as they appear. Based on pre-trained, deep-learning neural networks, the classification of plants and animals leads a controller to distribute fertilizer, pesticide, herbicide, or, even more importantly, not distribute anything at all. This intelligent sensing method limits the number of chemicals used by accurately using them only when needed. Blue River Technologies' AI solution has achieved a 90% reduction in herbicide costs by selectively applying it when required, which could potentially lead to a global reduction of 2.5 billion pounds of herbicide.
Future of AI in Agriculture
Artificial intelligence is used far beyond plant identification in impacting smart agriculture. AI can be used in navigation, sensor analysis, asset management, threat detection, mass land analysis, and drone flight. Artificial intelligence continues to evolve, and its applications in smart agriculture are only beginning to flourish. John Deere, one of the largest agricultural equipment manufacturers on the planet, utilizes AI technology only on its highest-end equipment and navigation systems. However, as is true with most industries, the implementation of AI will continue to be seen in more entry-level models for a variety of uses.
Conclusion
Big data and artificial intelligence are in a mutually exclusive relationship — as one advances, the other benefits. As it benefits, it progresses, and the cycle continues to evolve both technologies. In more applicable terms, the use of big data collecting activities in smart agriculture will create more resilient and useful AI models, as with Blue River Technologies. The deployment of these models will create more efficient methods of smart agriculture processes, as we see with companies such as Yanmar. That, in turn, will generate more data to be understood and optimized.
The implementation of big data and AI in smart agriculture has already had profound effects on a small scale, and its global impact will assuredly be even more insightful in years to come.