Drones, AI and Robot Pickers: Meet the Fully Autonomous Farm | William Boston, WSJ New technologies are paving the way for farms that can run themselves, with minimal human input In the verdant hills of Washington state’s Palouse region, Andrew Nelson’s tractor hums through the wheat fields on his 7,500-acre farm. Inside the cab, he’s not gripping the steering wheel—he’s on a Zoom call or checking messages. A software engineer and fifth-generation farmer, Nelson, 41, is at the vanguard of a transformation that is changing the way we grow and harvest our food. The tractor isn’t only driving itself; its array of sensors, cameras, and analytic software is also constantly deciding where and when to spray fertilizer or whack weeds. Many modern farms already use GPS-guided tractors and digital technology such as farm-management software systems. Now, advances in artificial intelligence mean that the next step—the autonomous farm, with only minimal human tending—is finally coming into focus. Imagine a farm where fleets of autonomous tractors, drones and harvesters are guided by AI that tweaks operations minute by minute based on soil and weather data. Sensors would track plant health across thousands of acres, triggering precise sprays or irrigation exactly where needed. Farmers could swap long hours in the cab for monitoring dashboards and making high-level decisions. Every seed, drop of water and ounce of fertilizer would be optimized to boost yields and protect the land—driven by a connected system that gets smarter with each season. Much of the technology to power an autonomous revolution in agriculture already exists or is nearly ready for market launch. “We’re just getting to a turning point in the commercial viability of a lot of these technologies,” says David Fiocco, a senior partner at McKinsey & Co. who leads research on agricultural innovation. A McKinsey survey in 2022 found that around two-thirds of American farms use digital systems to manage their farm operations, but only 15% of large farms and just 4% of smaller ones have yet invested significantly in robotics or automation. Fiocco expects the use of robots to rise dramatically in the coming years. Despite the promise of digital tools and autonomous machines, cost is a big barrier. Connectivity is another hurdle. Robots need to talk to each other. Moving data to a cloud requires broadband internet, and from a remote field that likely needs to be wireless. But wireless internet and land-based broadband aren’t available everywhere in rural America. In developing countries, the digital gap is even wider. Some farmers are experimenting with edge computing, a networking design that stores data closer to where it originates. But experts say ultimately farms need to be connected to cloud-based systems. Here’s a look at some of the essential components in the vision of the autonomous farm. Autonomous tractors Tractors that can plant, till and harvest with little, or only remote, human supervision are moving from prototype to practice. Traditional manufacturers and tech startups are placing big bets. Monarch Tractor, a Livermore, Calif., firm, has rolled out an all-electric, “driver-optional” tractor now working vineyards. Its MK-V model can run up to 14 hours on a charge and be ready to roll again after six hours plugged in. Farmwise, another California company, has developed an AI-guided mechanical weeder and tiller that uses computer vision and robotics to identify and pluck weeds, running day or night, reducing the need for herbicides. In April, salad giant Taylor Farms acquired Farmwise, citing the technology’s promise to cut labor costs and support more sustainable farming. Deere & Co. is taking an incremental approach, adding layers of automation to help farmers become comfortable with the technology—and see immediate payoffs—while paving the way toward full autonomy. Some of Deere’s large sprayers use “See & Spray” technology that incorporates computer vision and machine learning to target weeds in soybean, corn and cotton crops. Trained on thousands of images to identify weeds in real time and command individual nozzles to spray only where needed, it reduces herbicide use by up to two-thirds, the company says. Thirty-six cameras mounted on a sprayer boom scan fields at 2,100 square feet per second—far beyond what the human eye can manage. Using data and AI to analyze individual plants could eventually become a mainstream practice in farming. A 5,000-acre farm can contain around 750 million plants, and the challenge is giving each one its share of tender loving care. “Sensing technology paired with models, paired with automation and eventually autonomy where it makes sense—there’s a lot of opportunity there,” says Sarah Schinckel, director of emerging technologies at the Moline, Ill.-based company. Fruit-picking robots and drones Automation, now most often used on large farms with wheat or corn laid out in neat rows, is a bigger challenge for crops like fruits and berries, which ripen at different times and grow on trees or bushes. Maintaining and harvesting these so-called specialty crops is labor-intensive. “In specialty crops, the small army of weeders and pickers could soon be replaced by just one or two people overseeing the technology. That may be a decade out, but that’s where we’re going,” says Fiocco of McKinsey. Fragile fruits like strawberries and grapes pose a huge challenge. Tortuga, an agriculture tech startup in Denver, developed a robot to do the job. Tortuga was acquired in March by vertical farming company Oishii. The robot resembles NASA’s Mars Rover with fat tires and extended arms. It rolls along a bed of strawberries or grapes and uses a long pincher arm to reach into the vine and snip off a single berry or a bunch of grapes, placing them gingerly into a basket. “Robotic harvesting can offer greater consistency and efficiency than manual labor, while reducing expenses and addressing the labor shortages affecting the industry as a whole,” Brendan Somerville, chief operating officer and co-founder of Oishii said in an email, adding that the company’s long-term vision is to fully automate its harvesting operations. Israel-based Tevel Aerobotics Technologies aims to help fruit growers reduce the need for labor with its “Flying Autonomous Robots” that can prune, thin and harvest crops. Using AI and machine vision, the robots locate the fruit, determine whether it’s ripe and then pluck it off the tree. “Growers who don’t adopt robotics won’t survive—they simply have no choice,” says Tevel Chief Executive and founder Yaniv Maor. Scaling up, however, remains a cost challenge for the company. Remote sensing, image analytics Drones and satellites, guided by artificial intelligence, are turning farms into data-driven operations. By capturing detailed images and sensor readings, they create “digital twins”—virtual replicas of fields that show exactly where crops are too dry, too wet or under attack by disease or pests. This technology lets farmers spot problems early and target interventions more precisely, cutting waste and boosting yields. While pieces of this system are already in place, the next step is a fully connected network of machines that not only detect issues but learn from them. Ranveer Chandra, a senior Microsoft executive who spearheaded agriculture technology applications, sees a future where tractors and drones work in tandem, performing tasks like planting or spraying while continuously feeding new data into AI models tailored to each farm’s conditions. “There will be more automation, more use of drones, more robotics—it won’t be farms without farmers, but AI will significantly amplify the productivity of every grower,” Chandra says. “Every time a drone flies or a tractor plants, it’s gathering data that updates the farm’s own unique AI model.” Read more: