Robots Take the Wheel as Autonomous Farm Machines Hit Fields
Robots are taking over farms faster than anyone saw coming.
The first fully autonomous farm equipment is becoming commercially available, which means machines will be able to completely take over a multitude of tasks. Tractors will drive with no farmer in the cab, and specialized equipment will be able to spray, plant, plow and weed cropland. And it’s all happening well before many analysts had predicted thanks to small startups in Canada and Australia.
While industry leaders Deere & Co. and CNH Industrial NV haven’t said when they’ll release similar offerings, Saskatchewan’s Dot Technology Corp. has already sold some so-called power platforms for fully mechanized spring planting. In Australia, SwarmFarm Robotics is leasing weed-killing robots that can also do tasks like mow and spread. The companies say their machines are smaller and smarter than the gigantic machinery they aim to replace.
Sam Bradford, a farm manager at Arcturus Downs in Australia’s Queensland state, was an early adopter as part of a pilot program for SwarmFarm last year. He used four robots, each about the size of a truck, to kill weeds.
In years past, Bradford had used a 120-foot wide, 16-ton spraying machine that “looks like a massive praying mantis.” It would blanket the field in chemicals, he said.
But the robots were more precise. They distinguished the dull brown color of the farm’s paddock from green foliage, and targeted chemicals directly at the weeds. It’s a task the farm does two to three times a year over 20,000 acres. With the robots, Bradford said he can save 80% of his chemical costs.
“The savings on chemicals is huge, but there’s also savings for the environment from using less chemicals and you’re also getting a better result in the end,” said Bradford, who’s run the farm for about 10 years. Surrounding rivers run out to the Great Barrier Reef off Australia’s eastern cost, making the farm particularly sensitive over its use of chemicals, he said.
Costs savings have become especially crucial as a multi-year rout for prices depresses farm incomes and tightens margins. The Bloomberg Grains Spot Index is down more than 50% since its peak in 2012. Meanwhile, advances in seed technology, fertilizers and other crop inputs has led to soaring yields and oversupply. Producers are eager to find any edge possible at a time when the U.S.-China trade war is disrupting the usual flow of agriculture exports.
Farmers need to get to the next level of profitability and efficiency in farming, and “we’ve lost sight of that with engineering that doesn’t match the agronomy,” said SwarmFarm’s Chief Executive Officer Andrew Bate. “Robots flip that on its head. What’s driving adoption in agriculture is better farming systems and better ways to grow crops.”
In Saskatchewan, the first commercially sold autonomous tractors made by Dot are hitting fields this spring.
The Dot units won’t be completely on their own this year — farmers who bought equipment as part of a limited release are required to watch them at all times. But after this trial run, the producers may be able to let the equipment run on its own starting next year. That will open up a lot of time for the growers who will no longer need to sit behind the steering wheel.
Farmers are always managing multiple tasks, said Leah Olson-Friesen, CEO of Dot. “When you look at the amount of intelligence that’s sitting in the cab, they could be on the phone doing different things or outside of the cab — there’s some real opportunities there.”
But farmers do more than steer when they’re in the cabs of their tractors, said Alex Purdy, head of John Deere Labs and director of precision agriculture technology. Deere hasn’t yet released fully autonomous equipment because the technology that’s out there still isn’t good enough to replace people, he said.
Machinery that uses automation for tasks right now is more beneficial to farmers than autonomous equipment, Purdy said. Artificial intelligence, deep learning and advances in computer vision are going to transform agricultural machinery even further, he said.
“Automation is a never-ending journey — there’s always something that will get better over time, and there’s so much opportunity that we’re prioritizing automation over autonomy,” Purdy said.
A modern tractor does thousands of tasks, and to provide a fully autonomous solution, a deep understanding of each of those tasks is needed to automate them, said Brett McClelland, product manager of autonomous vehicles at CNH Industrial.
One of the areas that are still evolving is the ability of machines to see. “Sensing and perception is one of the most challenging overarching themes,” said McClelland.
While CNH Industrial in 2016 revealed a sleek, aggressive-looking prototype to much fanfare, the product is still in test pilots and not yet commercially available. For some tasks, current equipment is oversized, and smaller machines might be able to successfully scout a field, for instance. But they won’t be able to prepare the ground for planting carrots, where machines rip up soil 40 inches deep, McClelland said.
“Farmers have a demand for productivity, and they’ll take it in whatever way we can give it, and technology is the new way,” he said.
CNH Industrial is conducting pilots with fully autonomous equipment, working on tasks like tillage with E. & J. Gallo Winery and carrot planting with Bolthouse Farms.
Still, bigger won’t necessarily be better going forward, according to Ohio State University professor Scott Shearer. Modern tractors can weigh more than 50,000 pounds, which compresses the soil, making it less productive. Alleviating that compaction could increase crop yields by as much as 7%.
Increases in precision technology will also allow for smaller robots, as was the case for farmer Bradford’s fields where weed-killing spay was applied only where it was needed.
“To survive, farmers are always having to try to become more efficient and to keep costs down, while improving yields,” Bradford said. “The way that’s going to be achieved is in accuracy, by being timely with the operations and applying inputs directly where they are needed, rather than with a broad blanket approach to large areas. That’s where these robots will work.”