John Deere Tasks Satellite to Connect New Frontiers in Agriculture
Farming, like just about everything these days, runs on data. When a tractor is planting a field, it isn’t just laying down seeds. It is also collecting data. The precise location of seeds, the condition of the field, the efficiency of the machine — these are all data points ingested by the farm equipment. This data can be used in the moment, but also over the long term, to make farms more efficient and productive, improving fuel economy and land stewardship.
Agricultural giant John Deere has invested in connected agriculture technology over the last decade, and now boasts 500,000 farm machines connected to cellular wireless networks. The company offers a large suite of precision agriculture tools including a farm management system called the Operations Center, which pulls together data from a farm's machines, people, and technology. The amount of data collected is so great that twice a year, during planting and harvest seasons, John Deere is one of Amazon Web Services’ largest clients by data volume.
But these features only work in real-time when a farmer or machine is connected to a cellular network. And anyone who lives in or has driven through rural America knows that cellular networks leave much to be desired. When cellular networks are unavailable or lacking, the machine collects data but operates as if it is in “airplane mode” until it connects to a network. In some instances, this means farmers using flash drives to manually upload the information to a computer.
John Deere wants to change that by working with the satellite industry, in a partnership that could potentially crack open a new use case for the industry.
In September, John Deere gathered together representatives from a number of satcom companies at a test farm outside of Des Moines, Iowa, including SpaceX’s Starlink constellation, OneWeb, Hughes Network Systems, and Lockheed Martin, to announce a satcom request for proposals. The company is looking for satellite service and ruggedized satellite terminals that can connect thousands of agricultural machines and provide them with always-on connectivity, beyond the reach of cellular networks.
“Terrestrial cell tends to find the business model most lucrative when there are a lot of people and less lucrative when there isn’t,” John Deere CTO Jahmy Hindman says, explaining the problem. “Most of our customers operate in regions that are rural, that have low population densities. We have a need to create and craft a connectivity solution that would hopefully be ubiquitous across the surface of the planet, that allows all of our machines to be connected at all times.”
The cab inside the 8R autonomous tractor. The machine can operate without a driver in the cab. Photo: John Deere
The cab inside the 8R autonomous tractor. The machine can operate without a driver in the cab. Photo: John Deere
The Opportunity for Satellite
Although John Deere’s internal analysis estimates that about 75 percent to percent of the United States is covered by 3G or better terrestrial connectivity, the reality is mixed. A particular field may have good connectivity in one section, but not in another, and those gaps hold Deere back from the advanced real-time analysis, data sharing, and autonomous farming that it wants to enable.
The company has said that autonomy and artificial intelligence capabilities could unlock as much as $150 billion of incremental addressable market — but many of those use cases require connectivity.
Hindman shares the example of two sprayers working the same field at the same time, continuously sharing data about where each machine had sprayed in order to maximize efficiency. This application requires persistent connectivity.
Another application is edge computing. Today, all the computing for Deere’s machine learning algorithms is on board farm equipment, and machines have hardened graphics processing units (GPUs), but this could change if there was dependable connectivity.
“If you get to a state of ubiquitous connectivity at a latency and a bandwidth that allows us to start to move some of the compute off the machine, you free up that compute onboard the machine and actually reduce the cost of compute substantially,” Hindman says. “It opens you to additional compute capability that you never would have been able to package onboard a machine, which creates an opportunity for more complex data models and more complex machine learning models.”
There is not a solution currently on the market that is economically competitive for John Deere, one that would meet John Deere’s needs for terminal cost, and bandwidth and data fees, Hindman says.
“Terminal expense is a key part of it. Largely, it’s because solutions have been crafted around extremely small volumes or military applications,” he explains. “But it’s more than just terminal cost, it’s also data fees. We move petabytes of information and the data structure is going to be important to making the economics work.”
The CTO puts the market opportunity for the satellite industry at about 5,000 new machines annually that could be equipped with a satellite solution and about 40,000 existing machines that are good candidates for retrofit. He believes this type of volume can unlock new economics for the satellite industry. And John Deere could also bring its manufacturing expertise to the project, as Hindman noted that Deere manufactures its own electronics hardware and could manufacture a terminal.
Hindman did not provide a dollar amount Deere is hoping to hit for an affordable terminal, but says Deere needs “an order of magnitude reduction in terminal cost.”
In terms of the specific satellite solution, Hindman says he’s “orbit-agnostic,” but a solution needs to serve a multi-mode connectivity environment with both terrestrial cell and satellite connectivity.
This project will likely not result in John Deere launching its own satellites. “The geek in me really wants to launch satellites, but I’m not sure that’s in our best interest,” Hindman jokes.
The company wants to move quickly and hopes to have a solution landed by mid-2024. Deere is open to a resulting solution being sold to other customers, but not in industries that the company competes in, the CTO said.
But terminal cost is one of the ongoing hurdles in the satellite industry across many markets like consumer broadband, military, and aviation.
Whitney Lohmeyer, a satellite communications industry advisor who was the first engineer hired at OneWeb and serves on the faculty at Olin College of Engineering, is hopeful for the opportunity for the industry, but skeptical that 5,000 terminals a year is a volume that can bring the cost of a terminal down by orders of magnitude.
Constellations like Starlink and OneWeb have shown that Low-Earth Orbit (LEO) can work, but these constellations are looking at the scale of potentially putting broadband terminals on hundreds of thousands of homes and still have challenges lowering the cost, she says.
“Folks come into the industry or have been around in industry and are frustrated. They see there's a problem, and that is the low-cost phased array or the low-cost consumer terminal,” Lohmeyer says. “Players dive in and model and simulate different scenarios, and they are convinced they can do it. What inevitably happens is, as they start building out their theoretical link budget, the terminal grows in size in order to meet gain over temperature requirements. Then they find themselves in a place where they need to serve enterprise customers.”
“I am so hopeful that we can bring down the cost,” Lohmeyer said. “There's a need and a major desire to connect all sorts of different markets and applications. But it's not really the agriculture sector’s specific needs that are inhibiting a certain price point. Fundamentally, the components and the requirements on performance are preventing certain price points at this time.”
The See and Spray Ultimate is equipped with sensors for targeted, more efficient crop spraying. Photo: Via Satellite
The See and Spray Ultimate is equipped with sensors for targeted, more efficient crop spraying. Photo: Via Satellite
From Precision Agriculture to Autonomous Farming
John Deere is no stranger to satellite. The company’s precision agricultural technology is enabled by its proprietary StarFire Network, which corrects public GNSS signals to make them more accurate for agricultural use. Reprocessed signals are rebroadcast over Inmarsat’s L-band network. John Deere is one of only a handful of companies that has its own global satellite augmentation network. This network allows John Deere tractors to drive themselves, place seeds, and develop accurate geospatial maps.
Satellite imagery and satellite-enabled IoT play a role as well. Both factor into the John Deere Operations Center, which runs more than 250 software applications from partners. This can include software that uses satellite imagery, or software that incorporates data from third party in-field IoT sensors that may transmit data over satellite networks.
Every day, tens of thousands of farmers use the Operations Center to manage the farm logistics, understand what is happening, how their fleet is operating, and analyze data collected. As Deere deployed more technology, the company began adding more sensors to its equipment to collect more data, explains Lane Arthur, vice president of Data, Applications & Analytics, John Deere.
“On combines, we have a camera that has a machine learning app. As the grain is going through the chute [during harvest], we're making rapid, real-time adjustments to the quality of that stream, and changing fan speeds and other things so that we collect the most grain possible. That's the kind of technology that we've moved into over time. We're doing this on every one of our machines,” Arthur says.
This technology has revolutionized agriculture, and the next frontier is autonomous farming, which satellite can help enable, says Deanna Kovar, vice president of Production & Precision Ag Production Systems for John Deere.
“We've spent much of the last decade focused on digitizing agriculture in a way that puts the information right at the farmers fingertips in near real-time. Connectivity was a critical part of that — we've been connecting all of our production agriculture equipment since 2011,” Kovar says. “As we look forward into the 2020s, it’s all about computer vision and machine learning. These aren’t technologies that agriculture created, but they are technologies that agriculture is going to leverage. What you’ll see is how great the technology is, but also how much better we can make it when we can connect it persistently and confidently, anywhere in the world.”
Deere unveiled its autonomous 8R tractor earlier this year at the Consumer Electronics Show (CES) in Las Vegas, pitching it as a tool to help farmers feed a growing global population with less available land and skilled labor while dealing with challenges like climate and weather conditions. The company has set a bold goal to make the entire production system to grow corn and soy autonomous by 2030.
The autonomous tractor is fully loaded with tech. It has six pairs of stereo cameras to detect obstacles and calculate distance. The autonomous tractor continuously checks its position compared to a digital boundary area of where it is supposed to operate.
It can operate fully autonomously. While the StarFire Network enables the self-steering feature Auto Trac, which operates like tractor cruise control, the autonomous tractor does not require a farmer in the cabin. A farmer can drive the tractor to a field, turn it on, leave, and monitor progress through an app.
Adoption is low at this point, but the technology is nascent, Kovar said. Deere has paying customers for fully autonomous tractors this fall and expects more in the spring.
Kovar says Deere started its autonomous journey with tillage, or tractors that turn over soil to prepare for seeding, because the machine does all the work. It does not need to bring anything else to the field, or take anything away from the field. But as Deere moves into autonomous applications that are complicated, it needs persistent connectivity to manage the back and forth of data.
“At harvest, I need to take the crop from the field,” Kovar explains. “[That entails] knowing how much crop is there, how much I need to send, when I need to send it. All of that will be a part of autonomy making a difference in agriculture. But if it's not connected, if it's in airplane mode, I just wiped out the productivity value of having an autonomous machine out there in the field.”
Persistent connectivity could unlock new precision agriculture tools that have yet to be invented, because current machines cannot count on always-on connectivity.
John Deere representatives say autonomy can make a big difference in quality of life for farmers. Nancy Post, vice president of the Intelligent Solutions Group, said this is one learning so far from automation. At first, the company thought Auto Trac would mainly be beneficial because it would reduce overlap and save money.
“What we've been surprised about is how many people bought the self-steering piece simply because of the comfort,” Post says. “They are under so much pressure working 16-hour days where they have to get everything in the field or out of the field in a very short window. That comfort makes the biggest difference in the world.”
If a farmer doesn’t have to be in the cab of the tractor, the thought is that he or she can work on something else to maximize productivity, and even cut some time off a 16-hour day.
“Now I can go home and have dinner with my family,” Arthur says of the potential benefits. “I don't know how to put a value on that. But it’s pretty valuable.”
Deere is hoping that by bringing the satellite industry together, a vendor or set of vendors can come up with a solution that will meet the company’s use case and lead to a long-term collaboration, CTO Hindman says.
“We expect the group that is here from the industry to work together to form a solution that’s helpful in our agricultural use cases. We view it as an opportunity to have a partnership with whatever the solution set is that comes out of this over a long period of time,” Hindman says. “We think it’s a market-defining, significant volume relative to satellite communications. We have the opportunity to create something that doesn’t exist, with an economic value that is significant today, and going to grow into the future.” VS
Built with Shorthand