Can John Deere become one of the leading AI and robotics companies in the world alongside Tesla and Silicon Valley technology giants over the next decade?
That notion may seem incongruous with the general perception of the 185-year-old company as a heavy-metal manufacturer of tractors, bulldozers and lawnmowers painted in the signature green and yellow colors.
But that is what the company sees in its future, according to Jorge Heraud, vice president of automation and autonomy for Moline, Illinois-based Deere, a glimpse of which was showcased at last January's Consumer Electronics Show in Las Vegas, where Deere unveiled its fully autonomous 8R farm tractor, driven by artificial intelligence rather than a farmer behind the wheel.
The autonomous 8R is the culmination of Deere's nearly two decades of strategic planning and investment in automation, data analytics, GPS guidance, internet-of-things connectivity and software engineering. While a good deal of that R&D has been homegrown, the company also has been on a spree of acquisitions and partnerships with agtech startups, harvesting know-how as well as talent.
"This comes from our realization that technology is going to drive value creation and increase productivity, profitability and sustainability for farmers," Heraud said.
While Deere made a big splash at CES and intrigued the investment community, Stephen Volkmann, equity research analyst at Jefferies, said, "We are very, very, very early in this process."
"The total global fleet of autonomous Deere tractors is less than 50 today," he added. And even though Deere's goal is to have a fully autonomous farming system for row crops in place by 2030, Volkmann said, "in Wall Street time, that's an eternity."
For the time being, Deere is creating value and profits with well-established automated systems that can be retrofitted to its existing tractors, such as GPS-based self-steering and precision seeding that measures how deep and far apart to plant. Those steps have to be in place, Volkmann said, before you can put full autonomy around them.
The autonomous 8R represents a giant leap in current agtech, not to mention the marketing benefit. "Prior to its introduction at CES, everybody thought [full autonomy] was pie in the sky," said Scott Shearer, chair of the department of food, agricultural and biological engineering at Ohio State University.
Around the world, Shearer said, there are probably 30 different autonomous tractor projects in the works, though none are commercially available. "But when Deere, with 60% of the tractor market share in North America, comes out with one, that's when reality sets in," Shearer said.
That reality reflects Deere's autonomy strategy. "The AI we use involves computer vision and machine learning," Heraud said, science that was well underway at Silicon Valley startup Blue River Technology, which Deere bought in 2017 for $305 million — a deal that also brought on Blue River co-founder and CEO Heraud. Blue River's "see and spray" robotics platform utilizes dozens of sophisticated cameras and processors to distinguish weeds from crop plants when applying herbicides.
Attached to the autonomous tractor are six pairs of stereo cameras that can "see" an obstacle in the field — whether it's a rock, a log or a person — and determine its size and relative distance. Images captured by the cameras are passed through a deep neural network that classifies each pixel in approximately 100 milliseconds and decides whether the tractor should keep moving or stop.
"We've curated hundreds of thousands of images from different farm locations and under various weather and lighting conditions," Heraud said, "so that with machine learning, the tractor can understand what it's seeing and react accordingly. This capability also allows the farmer, instead of being in the tractor, to operate it remotely while doing something else."
Heraud was referring to autonomous driving, another piece of Deere's agtech puzzle that came together when it purchased Bear Flag Robotics last year for $250 million. Also a Silicon Valley startup, launched in 2017, Bear Flag's autonomous navigation system can be retrofitted onto existing tractors. In the case of Deere's 8R model, a tractor which first went on the market in 2020, the latest version with autonomous capabilities uses technology from Blue River.
Since the CES rollout, Deere has acquired AI assets from two other agtech pioneers. In April, Deere formed a joint venture with GUSS Automation, which has devised semi-autonomous orchard and vineyard sprayers. Using AI and IoT, multiple GUSS (Global Unmanned Spray System) sprayers can be remotely controlled by a single operator, running up to eight sprayers simultaneously from a laptop. GUSS can detect trees and determine how much to spray on each one, regardless of height or canopy size.
A month later, Deere announced the acquisition of numerous patents and other intellectual property from AI startup Light, according to The Robot Report. Light's depth-perception platform improves upon existing stereo-vision systems by using additional cameras, mimicking the structure of a human eye to enable more accurate 3D vision. Deere plans to integrate Light's platform into future versions of its autonomous farm equipment.
To keep a close eye on other agtech R&D, Deere has established a Startup Collaborator program to test innovative technologies with customers and dealers without a more formal business relationship. "The hope is that they find the diamonds before they become obvious to [competitors] and keep them in the fold," Volkmann said. Among the current crop are Four Growers, a Pittsburgh-based startup providing robotic harvesting and analytics for high-value crops, starting with greenhouse tomatoes, and Philadelphia-based Burro, which is producing small, autonomous robots that can assist farm workers with various conveyance tasks.
Not surprisingly, Deere's biggest competitors have been developing automation and autonomy for its farm machinery, too. AGCO, whose brands include Massey Ferguson and Fendt, "has been automating farming operations since the mid-1990s," said Seth Crawford, senior vice president and general manager of the Duluth, Georgia-based company's precision agriculture and digital division. "We're at a stage we call supervised autonomy, where we still have someone in the cab of the machine," he said. "The buzz is around fully autonomous operations, but where farmers are willing to pay for automation is feature by feature."
Whereas Deere is focused on adding full autonomy to its own farm equipment, AGCO is eying the wider retrofit market, Crawford said. "In summer 2023, we'll have a performance-enhancing retrofit kit available for multiple brands of machines," he said. "Where others say we bring you autonomy with a half-million-dollar tractor," he said, alluding to the price tag of Deere's 8R, "we have kits that allow you to do that with your existing fleet. We see a huge opportunity with the installed base, where farmers want to adopt technology to enhance their outcomes, and yet don't want to flip their entire fleet and make that massive investment."
In 2016, Case IH, a subsidiary of CNH Industrial, headquartered in London, rolled up to the Farm Progress Show with what it called the Autonomous Concept Vehicle. The sleek prototype tractor, minus a driver's cab, hinted at the view of autonomy at the time. Fast forward six years, to September's Farm Progress Show, where Case IH unveiled its Trident 5550 autonomous applicator.
Released in 2017, the Trident 5550 — with a cab — is designed for spreading dry and liquid materials in farm fields. The model at the farm show was retrofitted with autonomous technology developed by Raven Industries, which CNH acquired for $2.1 billion in June 2021. Similar to Deere's autonomous 8R, the enhanced Trident employs self-driving capability, advanced cameras and AI to interpret a continuous stream of images to detect obstacles.
The company plans to have a limited number of the machines ready for farmers to test before going to market perhaps next year, said Chris Dempsey, global director at Case IH Precision Technology, though the exact release date is to be determined. "We want to get customer feedback and understand their confidence level [in autonomy] before we go commercial," he said.