Agriculture is one of the last major industries to become digitized. It's not surprising, seeing as how off-road, rural environments are more challenging than roadway systems or manufacturing floors.
However, as the connectivity gap continues to close, there is tremendous opportunity to capture data that can ultimately lead to transformative technologies like artificial intelligence (AI).
"To put it as simply as possible, AI allows computer systems to complete tasks that are normally performed by humans," said Mark Kuehn, OEM sales manager of North America at Trimble.
Given that definition, AI could mean everything from cognitive tasks like data analytics and forecasting to physical tasks like spraying weeds and picking produce. As presented in AEM's whitepaper, The Future of Food Production, examples already exist that reinforce the positive impacts AI can have. For instance, robots utilizing machine learning can detect and pick harvestable fruit in a fraction of the time a human can.
It's important to remember, though, that the road to AI and complete machine autonomy is a long one.
"Completely replacing a human is pretty hard to do today," said Michael Gomes, vice president of business development of agriculture at Topcon.
Along the journey toward AI, several important steps can be taken that can have a profound effect on the way food is produced.
In the on-road world of automobiles, Gomes said that industry has outlined five levels of autonomy. Each level gains additional elements of autonomous operation until Level 5 where no human interaction is needed.
"In the off-road industry... the first step is mechanization," said Gomes. "Next is some form of automation, which much of the ag equipment industry has already been doing. Then the real opportunity emerges: an agricultural system comprised of smart, connected products."
Sensors, Connectivity, Data and Learning
"A machine becomes smart because it has sensors on it," Gomes explained. "When that machine becomes connected, machine learning can happen, which allows a machine to become much more efficient and become part of a much more efficient system."
Sensors gather data that enables machines to learn how to recognize situations and make decisions. A simple example of machine learning in everyday life is the "you may also like" recommendations various websites make based on analysis of your behavior.
"In looking at the current state in agriculture, we're still in the early stages of even understanding what the potential could be," Kuehn said. "Technologies are still being developed that enable a more complete understanding of what is happening in the field. Data not only has to be captured, but also processed into something digestible, enabling machines to ultimately carry out tasks by themselves."
Said Gomes, "Smart, connected devices with some amount of machine learning can become as smart as a dog today. Dogs can understand hand and verbal commands. Dogs can learn to go outside to go to the bathroom. Dogs know when they've done something right or wrong.
"That creates a lot of value for the agriculture industry," he continued. "If a machine is as smart as a dog, it can learn that it has to go out to X location to perform Y task."
Furthermore, it can learn how to respond to different commands it receives.
"The tractor itself is smart and connected, and it knows all kinds of things about itself like engine diagnostics and fuel consumption," Gomes said. "When you can also get data from sensors on the implement, now the implement can provide instruction to the tractor such as ‘speed up' or ‘slow down' based on the work you're trying to accomplish. When you have smart implements with a smart tractor, you have the first piece of an optimized system."
In the broader agriculture world, there are additional opportunities for sensors to observe and instruct. According to Kuehn, the industry is already in a pretty good place when it comes to soil sensors that can pick up on moisture and nutrient levels. There also is a decent amount of remote sensing technologies from satellites and drones. On-the-fly sensors such as cameras on equipment are already showing their potential.
"Sensors on a tractor driving through a field can look for consistency and certain signals," Kuehn explained. "The sensors can learn to pick up on the ripeness of a crop or greenness of a canopy. They can also look for signs of weeds or disease."
Gomes pointed to "smart sprayers" as an example.
"The sensors are taking pictures and comparing them to a reference library. The sprayer knows how to appropriately dose the right amount of material in the right place and in real time. It all starts with machine learning when there is enough data around a very defined set of problems. If I know I'm in soybeans, for instance, and I know they were planted a certain number of days ago, I know that I should be looking for these types of weeds which look like this."
Over the next 10 years, the focus will be on refining these tools even more.
"The first step is getting to a point where the machine learning has become accurate and capable enough that it can digest the information, create the prescription, and send instruction out to the tractor," Kuehn said. "That is already starting to occur today. But now it's being sent to a tractor that still requires some human interaction like hitting a button on the display. But as things progress over the next 10 years, that process can become more automated and require less human interaction to perform the task."
On that note, Gomes said another beneficial outcome of machine learning is machine coordination. The industry has grown accustomed to needing one operator per machine. Through connectivity and machine learning, the industry could move toward a more efficiency-optimized system where one human operates multiple machines working in tandem. For example, a harvester and trailer could work in coordinated fashion with varying levels of automation.
How Industry Can Prepare for the Journey
Gomes said ag equipment manufacturers must first identify where they currently are in the journey toward AI, and then decide where they ultimately want to go.
Kuehn noted that it's important to work closely with companies developing these technologies to ensure that equipment will be capable of carrying out the desired tasks.
As for the end-users, American farmers, Kuehn said a small percentage are utilizing the tech tools that are currently available. That will begin to change as rural access to high-speed internet improves and the next generation of farmers begins to establish itself.
"For those who are interested in things like machine learning and AI, it's important to stay informed about what is being developed," Kuehn said. "Talk with manufacturers and dealers about the options that are available. Precision agriculture technology providers can also help farmers understand what is available and what is coming."
Finally, Kuehn and Gomes agree that it's important for farmers to work with trusted advisors — advisors who are forward-thinkers and can help farmers navigate the fast-moving arena of technology and machine learning.
Telling the story of AI's potential in agriculture is another key element of enabling the transformation over the next decade.
"In addition to increased collaboration, the industry must continue to encourage the government to assist producers who want to acquire this kind of technology," Kuehn said. "There are already some government programs that can help, and we can learn from various programs that have been successful in other countries. This is an area where we can continue to improve. We can also continue to improve in the area of factory installation of precision agriculture technologies."
The story of AI and efficiency optimization in agriculture is a clear one.
"As the population and food demand increase, the agriculture industry must find ways to adapt," Kuhn said. "The adaptation needs to allow farmers to grow more food with fewer resources, including people. AI is one of the core pieces to grow food more sustainably. Sensors can help identify which parts of a field actually require application. Water use is another huge driver. Moisture sensors in the ground can identify when a plant is reaching a wilting point and actually needs water."
According to Gomes, technology is a lot like fashion in that it is always changing, and sometimes the change can cause a person to become a bit apprehensive.
"Nobody likes change, but most people do like progress," he continued. "The difference is when change has a purpose."
The purpose of transformational technologies like connectivity, sensors, machine learning and AI is very clear in the agriculture industry: enable farmers to fulfill higher production needs while reducing their environmental impact.
Sure, a little apprehension might come along with that. But so should a great deal of enthusiasm.
This story also appears on Agricultural Equipment Guide.
Today's top stories
