A handful of researchers keep a watchful eye as a pair of farm robots toil dutifully in local fields.
Based out of the Ontario Crops Research Centre — just south of the Holland River border between Bradford West Gwillimbury and the Township of King — University of Guelph professor and professional agrologist, Mary Ruth McDonald, along with her team are closing in on the end of the second of three growing seasons researching the capabilities of two self-driving robots that use geolocation to navigate fields, remembering where crops have been planted.
Since April 2023, they’ve been putting FD20, from Danish company FarmDroid, and Orio, from French company Naïo Technologies, through their paces seeding, weeding and spraying on both the muck soil of the Holland Marsh and the mineral soils of a nearby highland field.
Unlike some other projects involving farm robots, the goal for McDonald and her team isn’t just to see if they work, but to create a series of trials to analyze and compare the robots side-by-side with typical tractor equipment.
“We’re just in the process of analyzing that data, but right now it looks to be very close to equivalent,” she said.
That could lead to cost savings for growers, since the robots don’t use fuel like a tractor, and more importantly, don’t require a person to operate them, allowing that person to take on other tasks.
“We’re trying to get a sort of hour value on that, but we’re not there yet,” said Kevin Vander Kooi, a research technician on the project.
That’s partly because the team isn’t just researching what the European robots can do, but also how to make them more effective here in Ontario, especially in helping to grow the onions, carrots and beets for which the marsh is famous.
“Weeds grow thicker and faster on muck soil than they do on mineral soil, so that’s one of the things we’re really focused on,” McDonald said, explaining that herbicide-resistant weeds are becoming more common across the province, while some herbicides are no longer available.
As a result, cultivating the weeds by machine is one of the only methods to control them, unless crews hand pick them, but that labour could be in short supply.
Many growers already rely on the Seasonal Agricultural Worker Program in order to meet the existing demands of farming, and the Canadian Agricultural Human Resource Council predicts there will be more than 100,000 job vacancies in agriculture by 2030.
“One of the underlying things about the robots is that technology is going to be cheaper and easier to deal with over time, and of course wages keep going up,” McDonald said.
Currently, the team estimated the Canadian retail price of FD20 to be about $200,000, with Orio closer to $300,000.
Overcoming challenges
Despite being autonomous, the team discovered FD20 was “very, very slow” at seeding, which can pose challenges for growers who already have short planting windows.
Even with a solar panel allowing the robot to work during the day while simultaneously charging its battery to work all night, it still wasn’t planting enough.
That’s when Vander Kooi came up with the idea of seeding onions in spaced clusters of three, which has almost tripled the speed, thanks to the ingenuity of fellow research technician Geoff Farintosh, who was able to 3D print new plastic components for the seeder and “trick” the machine into using them correctly.
“We’ll probably start changing more components now too,” Farintosh said.
They also upgraded the FD20 from three wheels to four, and are considering simplifying the dual seeding/weeding setup to reduce complications.
“Sometimes these robots are a little over-engineered and we kind of dumb them down a little once they get here,” Farintosh said.
Less specialized, Orio is essentially an autonomous electric tractor, to which the team have been able to fit equipment for spraying and weeding, with plans to try seeding with it next year.
Other challenges aren’t so easily overcome, according to Vander Kooi, who explained error codes and malfunctions from electronic sensors can require more oversight of the robots than desired.
As part of the project they’ve been providing data to the manufacturers who have also been helping to find solutions.
Orio also needs a charging station instead of using solar, and can’t be left out in wet weather, whereas FD20 will automatically shut down in the rain.
While Orio has spent almost all of its time at the research station, McDonald estimates FD20 has spent about 95 per cent of its time performing literal field research at a grower’s property, to ensure the robots can handle real-world large-scale sites.
So far, the researchers said that grower is “pretty happy” with the robot, which is lighter than a typical tractor, allowing it to go out into the field even when a tractor can’t.
More work remaining
Determining if the robots are ready for more wide-spread adoption remains a focal point, and McDonald understands that growers may be hesitant to risk moving to a new system if the one they have is already working.
“We’re trying to do that ground work to help mitigate the risk to them,” Vander Kooi said.
The three-year project still has the 2025 growing season to go, before another off-season of preparing reports, and McDonald is hoping the project will also result in several scientific papers.
“It’s been changing so rapidly, just within the first two years, that it’s hard to anticipate where we’re going to be at the end of the three years,” she said.
The project has been possible thanks to funding from three separate sources, including about $180,000 from the Ontario Agri-Food Innovation Alliance (a collaboration between the Ontario government and the university), $30,000 from the Fresh Vegetable growers of Ontario, and $45,000 from Ontario-based Haggerty Agrobotics, as well as support from Picketa Systems and the AgRobotics Working Group.
For more information, including annual reports with summaries of the ongoing work, visit the research centre’s website.
