Aigen’s solar-powered autonomous AI robots called Element operates at Bowles Farm in Los Banos, California, on June 26, 2025.
JOSH EDELSON/AFP via Getty Images
British farms have been leaving food in the ground every summer since Brexit altered immigration rules and drained the seasonal workforce that kept harvests running. The UK government opened a £20 million (approximately $26.9 million) competition yesterday to change that, inviting businesses, universities, and farm consortia to build the robots that can do what seasonal workers increasingly cannot.
Round 2 of the Farming Futures Automation and Robotics competition, delivered by the Department for Environment, Food & Rural Affairs (Defra) in partnership with Innovate UK, opened August 3 and accepts applications through September 30, 2026. Collaborative projects valued between £500,000 and £2.5 million (approximately $673,000 to $3.4 million) are eligible.
Post-Brexit Fields and a £60M Food-Waste Problem
The structural problem the competition is designed to address is well-documented. The National Farmers’ Union has reported approximately £60 million of food left unharvested (approximately $80.7 million) on UK farms in a single year due to seasonal worker gaps. A 2022 UK parliamentary inquiry found that 49% of growers and food manufacturers had reduced output because of labor shortages, and that more than two-thirds of seasonal farm workers came from outside the UK through visa schemes that became administratively harder to access after Brexit ended free movement of labor.
Farm input costs have risen by an average of 44% since 2019, compounding the labor squeeze and narrowing the margins on which decisions about whether to automate are made.
Farming Minister Stephen Morgan framed the competition in precisely these terms. Farmers, he said, “routinely struggle to recruit enough seasonal workers during the busiest periods of the year.” The new round was, in his description, a “vote of confidence in British ingenuity” and a step toward modern, resilient farming.
Chris Danks, head of agrifood at Innovate UK, added that robotics and automation were “becoming increasingly important tools to help farmers improve productivity and build more resilient businesses,” and that the competition was designed to move projects “from concept to on-farm application.”
What Applicants Can Apply For
The competition is considerably broader than its 2023 predecessor, which offered £12.5 million and was limited (approximately $16.8 million) to arable and horticultural crops. This round adds ornamental horticulture, livestock management, and forestry — including tree nursery work, woodland planting, and woodland monitoring and management systems. Drones for crop monitoring, precision spraying, and livestock surveillance are in scope.
Projects must be collaborative and must run between 12 and 36 months, starting no later than March 1, 2027, and ending by February 28, 2030. A minimum of 50% of any grant money claimed by farmers, growers, or foresters must come from organizations based in England. Innovate UK has estimated that applicants face roughly a 35% chance of success based on comparable competitions.
Innovate UK will host applicant-support webinars this month: a general briefing session on August 6 at 2 p.m. British Summer Time (9 a.m. ET) via this joining link, and a consortia-building event on August 13 at 10 a.m. British Summer Time (5 a.m. ET) for organizations seeking research and farm partners — register for consortia-building event.
How the Robots Actually Work
For readers unfamiliar with what agri-robots do in the field, the two case studies from the 2023 round show the range of technology the government is looking to scale.
Agaricus Robotics, working alongside Littleport Mushrooms and the University of Lincoln, built what Defra describes as the world’s first robotic mushroom harvesting system deployed in a commercial setting, backed by £772,425 in 2023 funding (approximately $1.04 million). The machine uses computer vision to scan each mushroom bed before picking — identifying maturity by surface morphology and color — then applies a soft pneumatic gripper, a compliant end-effector that supplies controlled pressure just sufficient to detach the mushroom without bruising it. After picking, the robot trims the stalk to the farm-specified length in the same operation, combining three previously human steps into one robotic sequence.
The VISTA project, led by Outfield Technologies and awarded £786,220 in 2023 grant funding (approximately $1.06 million), demonstrates a different technology layer: data integration rather than physical manipulation. Drones equipped with multispectral cameras — capturing beyond the visible spectrum into near-infrared, which reveals plant stress and chlorophyll content invisible to the human eye — fly systematically over vineyards at JoJo’s Vineyard and elsewhere, while ground-based robots collect GPS-referenced structural data from vine rows. The system fuses both data streams to generate centimeter-accurate maps of every vine, producing yield forecasts that allow growers to predict harvest volume, time picking decisions more precisely, and negotiate better prices from buyers before a single grape is picked. GPS guidance layers then allow variable-rate application of chemicals — spraying precisely where disease or nutrient deficiency is mapped, rather than treating the entire vineyard uniformly. The full VISTA project record is available via UKRI’s Gateway.
Why Harvesting-as-a-Service Changes the Economics
The biggest barrier to farm robot adoption is not the technology — it is the price tag. High-specification harvesting robots can cost between £150,000 and £300,000 (approximately $202,000 to $403,500) per unit, placing them out of reach for most small and medium farms with thin margins and limited access to capital. Surveys of European farmers consistently identify upfront capital cost as the primary reason for not adopting robotics, and the pattern holds in the UK, where roughly 80% of farms have not yet invested in any agri-tech.
The emerging solution to this problem is harvesting-as-a-service — sometimes called robotics-as-a-service (RaaS) — in which the technology company, not the farm, owns and operates the robot, charging the farmer a per-unit or per-season fee comparable to seasonal labor costs. This converts a large capital expenditure into a predictable operating cost. The government’s competition explicitly requires funded projects to demonstrate “clear commercial potential and practical on-farm benefits,” a phrase that implicitly validates the service-delivery model over outright robot sales.
Cambridge-based Fieldwork Robotics is the most advanced example of this transition in the UK right now. The company, a spin-out from the University of Plymouth, secured £3 million in combined funding (approximately $4 million) earlier this year — a £2.2 million equity round led by Elbow Beach Capital and approximately £1.7 million in combined climate and Innovate UK grant funding. Its four-armed autonomous raspberry harvesting robot uses AI-enhanced stereoscopic cameras and machine learning classifiers to locate individual ripe fruit, then deploys soft grippers to pick at speeds comparable to human pickers.
The company is currently running two-year harvesting-as-a-service trials on commercial farms — one with food company Place UK in Norfolk, and another at Littywood Farm in Stafford — with multi-robot fleet deployment expected from 2027 if the trials meet commercial standards. In June 2026, the company secured additional backing from SEED Innovations, the AIM-quoted investment company chaired by Jim Mellon.
CEO David Fulton said: “Fieldwork is now entering its scale-up phase, moving from technology validation to full commercial adoption.”
Scope of the Wider Farming Innovation Programme
The £20 million (approximately $26.9 million) robotics competition sits within a recently announced £53 million programme expansion confirmed in June 2026. Together with earlier allocations, that brings total agricultural research and technology funding available in the current financial year to £123 million (approximately $165.5 million). Eight competitive funding rounds are open or scheduled across 2026, ranging from small on-farm ADOPT grants of £50,000 to £200,000 (approximately $67,000 to $269,000) for farm-level trials, through feasibility studies, to the larger Farming Futures R&D Fund competitions.
The robotics round is part of a broader government commitment to invest at least £200 million (approximately $269 million) in agricultural innovation by 2030, as confirmed in the competition brief. These targets are framed within the UK’s Modern Industrial Strategy, published in 2025, which identified Agri-Tech as a Frontier Industry within the Advanced Manufacturing sector plan — effectively designating farm robotics as a national industrial priority alongside semiconductor manufacturing and life sciences.
How Do the Robots Know When Fruit Is Ripe?
A brief technical aside for readers new to the computer-vision side: harvesting robots are primarily solving a detection-and-manipulation problem, not a navigation one. GPS and lidar systems handle farm navigation reliably. The unsolved engineering challenge — the reason robotic harvesting has taken two decades longer to commercialize than robotic welding — is discriminating individual ripe fruit from foliage in natural lighting, grasping it without bruising it, and releasing it cleanly at harvest speed.
Modern systems address detection with deep-learning models trained on large datasets of fruit images across ripeness stages; the Fieldwork system, for instance, uses AI-enhanced 3D cameras that generate a depth map of the raspberry canopy, allowing the robot to locate fruit in three-dimensional space even when partially occluded by leaves. The manipulation challenge is addressed with soft robotics: grippers made from compliant silicone or pneumatic membranes that conform to the fruit’s shape and exert only the force needed to detach it from the cane — typically around 0.3 newtons for a ripe raspberry. The mushroom robot’s gripper operates on the same principle applied to a different morphology.
This architecture — 3D computer vision plus soft manipulation — is modular: the same sensor suite and AI stack can be adapted to strawberries, broccoli, or courgettes by retraining the classifier and redesigning the end-effector. That modularity is part of what makes publicly funded collaborative R&D attractive to the sector; each funded project builds shared technical infrastructure as well as a specific product.
Is This Competition Right for You?
The September 30 deadline is firm. Applicants must be UK-registered businesses leading a collaboration; farmers, universities, charities, and research organizations can be consortium partners but cannot lead. Projects cannot involve aquaculture, cellular protein production, medicinal crop cultivation, wild fisheries, or equine operations. Heavy forestry harvesting machinery is also excluded.
Funding covers industrial research projects, with grant rates of 50% to 70% of eligible costs for businesses (depending on company size) and up to 80% of full economic costs for academic participants.
For organizations that missed the August 6 briefing webinar, recordings will be published on Defra’s webinar page. The August 13 consortium-building session remains open for registration.
Exchange rate conversions throughout this article are approximate, based on rates at the time of publication and subject to change.
Frequently Asked QuestionsHow does a robotic fruit picker know which fruit to pick?
Modern agricultural picking robots use a combination of AI-enhanced stereoscopic 3D cameras and machine learning image classifiers to identify individual ripe fruit. The cameras generate a spatial depth map of the plant canopy, and the AI model — trained on thousands of labeled fruit images across ripeness stages — identifies which fruit are ripe based on color, size, and shape. Once a target is located in three-dimensional space, a soft pneumatic gripper is positioned and applies just enough pressure to detach the fruit without bruising it. The engineering bottleneck is not the AI detection, which works well in controlled conditions; it is making the system reliable across natural lighting variation, irregular canopy geometry, and the speed required for commercial viability.
What is the Farming Innovation Programme, and who can apply to it?
The Farming Innovation Programme is the UK government’s umbrella R&D funding mechanism for agricultural technology, delivered by Defra in partnership with Innovate UK. It funds projects across early-stage feasibility studies, collaborative R&D, on-farm trials (the ADOPT grants), and larger Farming Futures competitions like this robotics round. Businesses of any size registered in the UK can lead applications; farmers, universities, charities, not-for-profits, and research organizations can participate as consortium partners. Projects must benefit English farmers or growers and carry out most of their work in the UK. The current financial year’s total allocation is £123 million in 2026 (approximately $165.5 million), part of a commitment to invest at least £200 million (approximately $269 million) by 2030.
Can robots realistically replace seasonal farm workers at scale?
The honest answer is: not fully, not yet, and not in all crop types — but the economics are getting close in specific niches. For soft fruit like raspberries, where labor accounts for the majority of production cost and seasonal worker availability has shrunk since Brexit, the harvesting-as-a-service model being trialed by companies like Fieldwork Robotics is approaching commercial parity. The key challenge is not the robot’s performance in isolation but its performance across a full commercial season on a real farm — which is exactly what this round of government funding is designed to test at scale. The Defra Automation in Horticulture review identified soft-fruit picking, mushroom harvesting, and brassica harvesting as the categories with the highest near-term commercial potential.
What is harvesting-as-a-service, and why does it matter for small farms?
Harvesting-as-a-service (also called robotics-as-a-service, or RaaS) is a business model in which the technology company owns the robot and charges the farm a per-unit or per-season fee for its use, rather than selling the robot outright. This solves the single biggest barrier to agri-robot adoption: the upfront capital cost, which can reach £150,000 to £300,000 (approximately $202,000 to $403,500) per unit. Most UK farms are small or medium enterprises with tight margins and limited borrowing capacity; a capital expenditure of that magnitude is simply not accessible. By converting the cost into an operating expense — comparable in structure to paying seasonal laborers — RaaS makes automation available to farms that cannot afford to buy, and gives technology companies a recurring revenue model that does not depend on one-time equipment sales. Fieldwork Robotics is currently deploying this model in Norfolk and Stafford trials in 2026.