Across the United Kingdom, the last of the spring barley has been brought in from the fields, the culmination of an agricultural calendar whose rhythm has remained unchanged for millennia. But when the nineteenth-century poet John Clare wrote, in his month-by-month description of the rural year, that in September “harvest’s busy hum declines,” it seems unlikely that he was imagining the particular buzz—akin to an amplified mosquito—of a drone.
“The drone barley snatch was actually the thing that made it for me,” Jonathan Gill, a robotics engineer at Harper Adams University, told me recently. Gill is one of three self-described “lads” behind a small, underfunded initiative called Hands Free Hectare. Earlier this month, he and his associates became the first people in the world to grow, tend, and harvest a crop without direct human intervention. The “snatch” occurred on a blustery Tuesday, when Gill piloted his heavy-duty octocopter out over the middle of a field, and, as the barley whipped from side to side in the propellers’ downdraft, used a clamshell dangling from the drone to take a grain sample, which would determine whether the crop was ready for harvesting. (It was.) “Essentially, it’s the grab-the-teddy-with-the-claw game on steroids,” Gill’s colleague, the agricultural engineer Kit Franklin, said. “But it had never been done before. And we did it.”
The idea for the project came about over a glass of barley’s best self: beer. Gill and Franklin were down the pub, lamenting the fact that, although big equipment manufacturers such as John Deere likely have all the technology they need to farm completely autonomously, none of them seem to actually be doing it. Gill knew that drones could be programmed, using open-source code, to move over a field on autopilot, changing altitude as needed. What if you could take the same software, he and Franklin wondered, and make it control off-the-shelf agricultural machinery? Together Gill, Franklin, and Martin Abell, a recent Harper Adams graduate, rustled up just over a quarter million dollars in grant money. Then they got hold of some basic equipment—a small Japanese tractor designed for use in rice paddies, a similarly undersized twenty-five-year-old combine harvester, a sprayer boom, and a seed drill—and connected the drone software to a series of motors, which, with a little tinkering, made it capable of turning the tractor’s steering wheel, switching the spray nozzles on and off, raising and lowering the drill, and choreographing the complex mechanized ballet of the combine.
“There were lots of people who thought the project wasn’t going to work,” Gill said. “Lots.” Hands Free Hectare’s budget was so small that the team had no test field or backup machinery; indeed, they didn’t secure the tractor until last December, just a few months before the barley was due to be sown. This left little time for the necessary trial and error; often, Gill and his colleagues would be midway through configuring their setup to perform one task, only to have to take it apart again to accomplish another. When they finally managed to get the crop in the ground, their rows looked wobbly. “It turns out that the autopilots in these drone systems weren’t designed to travel in a very straight line,” Gill said. “There’s no need for it—it’s just designed to get from point A to point B as efficiently as possible.” When the software hit a rock, it would navigate around the obstruction, following the path of least resistance rather than plowing through. Gill adjusted the code for straighter steering, regardless of the terrain, but not in time for the initial planting, which meant that subsequent tractor runs crushed hundreds of precious barley seedlings.
In order to live up to the name Hands Free Hectare, the team had decided that no one would set foot in the field until the harvest was brought in. This posed a problem for Kieran Walsh, its crop-health adviser, who was accustomed to collecting soil and plant samples manually and scrutinizing them for signs of infestation and illness. Walsh was aware that robotic weed detectors were already commercially available, but, to the best of his knowledge, there was nothing that could provide all the information he needed. “My initial thought was, Gosh, this is exciting,” he told me. “And then I thought, Right, actually, this is going to be quite tricky.” In the end, Gill flew drones over the field on a weekly basis, gathering spectral data that Walsh could use to measure the barley’s photosynthetic activity and assess soil moisture. With Abell and Franklin, he also built a robotic sample collector. “Ninety-five per cent of the information I wanted, they got for me,” Walsh said. “But there was that five per cent where I had to make an educated guess.”
Hands Free Hectare’s final yield was a couple of metric tons lower than the average from conventionally farmed land—and the costs in both time and money were, unsurprisingly for a pilot project, stratospherically higher. Nevertheless, the team’s experience suggests that drone agriculture offers some substantial benefits. “For starters,” Abell said, “the opportunity for doing the right thing at the right time is much higher with automated machines.” Many of a farmer’s duties are weather-dependent; an autonomous tractor could, for instance, tap into live forecast data and choose to go out and apply fungicide when conditions are ideal, even if it’s four o’clock in the morning.
More important, once the machinery no longer requires a person to sit on top of it, a farmer could deploy a fleet of small tractors to do the same work that he currently does riding one of today’s state-of-the-art, two-story-tall tractors. “That has massive, massive implications,” Walsh said. For one thing, Abell explained, it would make the application of fertilizers and herbicides far more precise. “All the boom sprayers in this country these days are at least twenty-four metres wide,” he said. “So you’ve essentially got a twenty-four-metre paintbrush to apply chemicals to a field that you’ve surveyed at two- or three- centimetre resolution.” Smaller tractors can spray crops over a smaller area, coming closer to matching the pixel-by-pixel picture of crop health provided by drone imagery. Abell, who comes from a farming family, added that transitioning away from big, heavy machinery would be better for the land, too. “All that weight is knocking the life out of the soil,” he said. “Our yields have plateaued, and that’s a big part of why.”
The Hands Free Hectare team envisions a future in which farmers are fleet managers, programming their vehicles from a central mission control and using the time saved to focus on areas that need extra attention. “The actual driving of a tractor—I didn’t miss that at all,” Abell said. “And, by not spending all your time going in a straight line on auto-steer, it gives you more time to learn about your crop and hopefully manage it better.” So far, he said, the response among farmers has been generally enthusiastic; most of the reticence appears to come from the younger, early-career crowd. “The older guys that have been sat on a tractor for the last thirty years, seeing that it’s not the best use of their time—they appreciate it more,” Abell said.
Self-driving tractors face many of the same safety issues as self-driving cars, in terms of cybersecurity and liability for accidents, so a good deal more work remains to be done before they will enter widespread use. Gill predicted that the first adopters will be in Japan, where the average farmer is seventy years old. Abell expects that commercial farmers in the U.K. will be automating at least some aspects of their operations within the next five years. The team’s focus, however, is on the even shorter term: first, a much needed vacation; then a new crop (winter wheat) in the ground by the end of October; and, finally, a special beer brewed from their hands-free harvest. “I’m hoping for a festive pint,” Gill said. “We’ll probably sell the rest to fund the project.”
