Meet the robots that will plant and pick our food
Dwindling numbers of workers means agriculture needs to turn to tech for help – and that could spark a robotic farming revolution. Nicole Kobie reveals the future of farming
Dwindling numbers of workers means agriculture needs to turn to tech for help, reveals Nicole Kobie – and that could spark a robotic farming revolution.
The robot trundles between rows of plants, assessing the berries on the stems. Depending on the time of year, it could be spraying or applying UV light against pests, snapping images of fruit to track its growth or – one day soon – choosing which berry to delicately harvest.
Farming may be an industry with an old-fashioned vibe, but that couldn’t be further from the truth: it has long used AI to manage crops and the Internet of Things to track growth, and tractors went driverless long before cars. Planting and harvesting crops with robots has been a challenge too far, however, as it’s delicate, highly skilled work.
But the pressure is on, especially in Britain. Over the past few years, the UK saw an average of almost 70,000 seasonal workers come to the country to help with harvests, in particular to pick fruit – indeed, 12,000 people come to the UK to pick strawberries alone. The triple threat of Brexit, Covid-19 and plenty of work elsewhere meant the UK was hit by a shortage of workers in 2020, which produced big challenges: “Firms that can’t get seasonal workers will collapse,” said Professor Simon Pearson, founding director of the University of Lincoln’s Lincoln Institute of Agri-Food Technology.
That’s not an exaggeration. In 2019, farmers said they were 30% short of the workers required, leaving fruit and vegetables rotting in fields and on trees. Reports in 2020 revealed one Doncaster farm, Pollybell, had to stop producing tenderstem broccoli because it couldn’t get a few dozen people to harvest it. It’s since turned to robotics as a potential solution.
The government attempted to convince Brits to pick fruit, but it didn’t go well: the Pick for Britain campaign failed to hire a sufficient number of local workers to fill those roles. After all, it’s back-breaking, highly skilled labour.
“They call it low-skilled workers, they’re actually really highly skilled,” Pearson said.
These challenges aren’t just affecting the UK. In Japan, the average age of a farmer is 69 and in the
US, seasonal workers from Mexico continue to be political dynamite. That also means the opportunity is huge: if UK businesses can unpick this problem, they’ll have a solution the world is hungry to buy.
Robots to the rescue
That’s why farming robots are already being trialled across the world, including here in the UK. Fieldwork Robotics, a company developed by researchers at the University of Plymouth, has developed a robot that can pick 25,000 raspberries a day
– it’s not necessarily faster than human workers, who can manage 15,000 in a single, eight-hour shift, but the robot can work all day long.
Another Plymouth project is the GummiArm, which spots ready-topick vegetables such as cauliflower before gently removing them from
plants. At the Lockerley estate in Hampshire, a robot named “Tom” examines wheat for issues while “Dick” follows it to electrocute weeds. Both are made by British startup
Small Robot Company, which also makes a version called “Harry” that can plant seeds. UK startup Xihelm has built a robot harvester for fruit and vegetables grown in greenhouses, using computer vision and AI to tell a robotic arm what to pick, while also keeping an eye out for disease. The company says it believes it has built the largest, most accurate dataset of 3D images of tomatoes.
It doesn’t stop there. A fruit cooperative called Berry Gardens is working with Saga Robotics and the universities of Lincoln and Reading to develop robots at Clock House Farm in Kent. Called Robot Highways, the project aims to reduce seasonal labour by using machines to help treat crops with pesticides and pick, transport and pack fruit.
However, it’s early days, said Pearson, who worked on the project. The first piece developed by the researchers was an AI-powered robot that trundles down orchard rows to carry the fruit picked by human workers back to where it’s packed. That saves a heavy and timeconsuming trip for seasonal workers, letting them focus on the most skilled part of their jobs. This summer, robots will be set to work spraying and collecting data for crop forecasting, with pickers arriving in 2022, Pearson said.
He believes that building a platform that can navigate a farm is a key first step. Once that’s possible, it’s relatively simple to add sprayers and light machines to combat pests or cameras to assess the state of a crop. Picking is a bigger ask, as a robot needs to be able to see fruit, decide if it’s ripe, manoeuvre an arm through plants and delicately remove the berry – the latter is a real challenge. “One route is you design grasping systems, which pick the fruit without touching it,” said Pearson. Some systems use vacuums to achieve this. “The other route is you design it to pick the stem, not the fruit.”
He added that getting the arm to the right place isn’t any easier. “You need an extraordinarily high amount of computation to identify that object in a cluttered space. You’re working out the unique trajectory for every single berry in real-time.” And then doing it tens of thousands of times a day. “I’m reasonably confident we’ll get to picking [robots], but the next challenge is not picking one berry, but thousands of them,” said Pearson.
Hectare to farm
Back in 2016, Kit Franklin started to farm a small piece of land. The lecturer in agricultural engineering at Harper Adams University wasn’t seeking a quieter life, but taking decades of academic research and turning it into practical robotics for agriculture. “Academic conversations were happening for 20 years,” he said. “But nothing was happening in the field.”
Robots were being made, but they were sent to do laps around football pitches and car parks as demonstrations, rather than doing any actual farming. Franklin’s co-founder Jonathan Gill was also frustrated by the lack of progress, but spotted an opportunity with drones. The pair worked together to answer a question that Franklin boils down to: “Why is no one making a robot tractor when we can make flying machines?”
The usual excuse was cost, but the tech in drones is relatively cheap.
“We thought we’d crack on and do it, rather than just talk about it,” he said. “The key technology that unlocked it
“I’m reasonably confident we’ll get to picking robots, but the next challenge is not picking one berry, but thousands of them”
was mobile phones, and the miniaturisation of computers with cheap and accessible gyroscope and GPS systems.”
So began the Hands Free Hectare project, run by Harper Adams University in Shropshire. The goal wasn’t to build complicated robotics, but to keep systems as cheap and simple as possible, in order to keep the technology practical and useful for farmers. “We weren’t just going to make a robot tractor, we were going to farm with it,” said Franklin.
The robot itself is an off-road mobility scooter controlled by drone technology. After sending the “tractor” out to do laps of the hectare field, the team simply attached different farming implements to the scooter, letting it tow a conventional seed drill, sprayer, combine harvester and so on. “There’s nothing clever about the implements we use,” said Franklin. “A lot of other people working in the agricultural robotics space were trying to reinvent the wheel – everything has to be new, novel and robotic. We’re just making the tractor drive itself, and letting the seed drill do what it does.”
And it works. The project has now expanded from a single square hectare to an entire 35-hectare farm, letting the team better test its hardware on different shaped parcels of land with new obstacles such as trees and telegraph poles.
The motivation behind such work is different from berry picking.
Unlike fruit, crops such as wheat don’t need mass amounts of seasonal labour because that’s already been mechanised out of such agriculture by heavy machinery — and that’s the problem. Massive tractors mean fields can be drilled, sprayed and harvested faster, but the heavy weight of the machinery flattens and damages the soil. The Hands Free Hectare system is much lighter, helping to avoid such damage but also allowing more agility in farming.
Franklin’s favourite example is winter wheat crop fertiliser, which needs to go on in mid-March or harvests will suffer. Last year, heavy rains meant larger equipment couldn’t go out in the mud, so farms neighbouring the Hands Free Hectare had to wait to fertilise. “All the farms in our area missed that date by at least ten days or two weeks because it was too wet for them to go on the field whereas our light machine, plus the fact the soil was in good health from
smaller machinery, meant we could go on the field and do the job,” Franklin said. Next, the company plans to develop more tractors and teach them to work in a “swarm” together, and has partnered with tech players to ditch the drone controls for bespoke, specialised systems. He predicts the system will be viable within the next two years.
Challenges remain
The robots themselves aren’t the only hurdle. Franklin is working with the industry on non-tech barriers such as insurance and regulation to ensure they’re cleared before the robots are ready for the field. One particular challenge is public right of way, as walkers in the UK have the right to
traipse across farms – which could be dangerous with automated equipment. That’s different from Australia, where one of his partners is based, because farms are larger and more remote. “The things that worry them are wildfires whereas I’m worried about making sure dog walkers are safe,” he said.
For these ideas to become the norm for agricultural industries, farmers will need digital infrastructure such as data architecture systems to manage all the information being collected and charging systems for the robots. And they’ll need sufficient connectivity covering the entire farm. “It’s a genuine use case for 5G,” said Pearson.
While Pearson and his colleagues develop robots in farmers’ fields, the tech industry also needs to step in and develop the rest of the infrastructure to support such efforts – and it could prove quite the chance for the IT sector. Agriculture is full of ripe opportunities ready to be harvested by technology companies with practical solutions. “There are lots of people we need to get involved all over the space, whether it’s data, hardware or software companies,” Pearson said. “It’s not a solved problem. And that means it’s a great place to be.”
“There are lots of people we need to get involved all over the space, whether it’s data, hardware or software companies”