What happens in the fields after dark
Somewhere on a soft-fruit farm in Lincolnshire, in the small hours of a July night, a low-slung four-wheeled robot is moving between the raised tabletop rows. No picker is out here at 2am. The air smells of warm polytunnel plastic and damp compost. The only light comes from the robot itself — a measured pulse of UV-C radiation, traversing each row at a steady walking pace, treating the strawberry leaves for powdery mildew while the harvest crew sleeps.
This is Thorvald, built by the Norwegian company Saga Robotics, and it is already a commercial reality on at least five major UK strawberry operations. Lincolnshire — a county that accounts for roughly a third of England's vegetable production and a significant share of its soft fruit — has become one of the natural proving grounds for this kind of precision agricultural automation.
But Thorvald is not picking strawberries. That distinction matters. The robot has slotted into a specific gap in the farming calendar — nightly disease control — that was never filled by a human rota in the first place. So what, precisely, does its arrival change? And for the tens of thousands of seasonal workers who travel to Lincolnshire's fields each summer, does it matter at all?
How Thorvald works — UV-C light and crop intelligence
The science behind the night shift is straightforward, even if the engineering is not. Powdery mildew — a fungal disease that spreads readily in the warm, humid conditions inside polytunnels — is the single biggest biological threat to a commercial strawberry crop. Historically, controlling it meant repeated chemical applications or intensive manual scouting. Thorvald replaces both: it emits calibrated UV-C light as it moves through each row, disrupting the fungal spores without leaving chemical residue on the fruit or soil.
The platform traces its origins to a 2014 Master's project at the Norwegian University of Life Sciences, where Lars Grimstad and his supervisor Dr Pål Johan From began designing a modular outdoor robot. By 2016, Saga Robotics had spun the research into a commercial company, and from 2016 to 2020 the Thorvald II was refined alongside real UK growers in genuine field conditions — meaning it arrived on British farms already shaped by strawberry-growing practice, not laboratory assumptions. By 2023 the robot was delivering full-season protection across the first cohort of major UK operations; today more than 150 are in commercial use, covering an estimated 20–30% of the UK's strawberry tabletop market.
Disease control is the primary job, but a secondary function runs in parallel. Two onboard cameras scan every row during each nightly pass, generating around 3 million crop counts per robot per night. The result is farm-wide data on flower numbers, fruit set, and ripeness, updated every four days — intelligence that would otherwise require dedicated scouts walking the rows by hand.
The grower returns are tangible. WB Chambers Farms reports 'excellent control of powdery mildew'; LM Porter's farm manager credits the robots with a 15% rise in production, attributed directly to lower mildew losses. These are agronomic outcomes, not efficiency projections. And critically, none of them involve a harvest picker: Thorvald's two automated tasks — disease suppression and crop scouting — were never part of the manual picking rota to begin with.
Where the University of Lincoln fits in
Lincolnshire's farming density offers something that university research programmes in agricultural technology can rarely access so directly: millions of working acres, a concentration of commercial growers experimenting with new inputs, and polytunnel soft-fruit operations at a scale where trial data becomes meaningful fast. The University of Lincoln's Lincoln Institute for Agri-food Technology (LIAT) and its Robotics and Autonomous Systems group sit adjacent to that environment — a positioning that, in REF 2021 terms, corresponds to research rated over 75% internationally excellent or world-leading. That figure matters less as a boast than as a signal to industry partners that the work is serious.
Saga Robotics' own account of Thorvald's development offers a template for how the academic-to-field pipeline can operate. The company describes its journey as moving 'from research roots to real-world change' — from a university lab in Norway to UK strawberry farms over the course of a decade. LIAT and the robotics group occupy a structurally similar position: funded research depth in the systems that Lincolnshire farms are now beginning to deploy, with an immediate test environment outside the campus gates rather than at the far end of a grant application.
That proximity does not guarantee a direct pipeline from laboratory to field. What it does establish is the institutional conditions for one: a county producing roughly a third of England's vegetables, adjacent to a university with the research infrastructure to develop and iterate on the systems needed to work those fields differently.
What seasonal workers actually face right now
Thorvald does not pick strawberries. That distinction matters for anyone trying to understand what autonomous farm robots actually mean for the people working alongside them.
UK horticulture relies on an estimated 70,000 to 80,000 seasonal migrant workers each year — planting, picking, and packing crops across farms where manual labour remains essential for harvest. Lincolnshire is among the counties most dependent on that workforce, given the scale of its soft-fruit and vegetable production. When the end of EU free movement disrupted the established flow of seasonal workers from eastern Europe after 2020, farm managers across the county faced a visible gap in available labour precisely when domestic food security was becoming a political concern. That sequence — labour shortage followed by intensified interest in automation — explains in part why UK adoption of platforms like Thorvald accelerated when it did.
What Thorvald automates are tasks that were already separate from the picking rota: chemical spraying replaced by UV-C treatment, manual crop scouting replaced by camera-based counts. The harvest picker is not the worker being displaced by this particular platform.
The subtler change is in how farms use the data Thorvald generates. Ripeness information updated every four days, at field level, means that decisions about when to send picking teams into which rows can be made with a precision that was not previously available. Workforce scheduling — when workers arrive, which sections they cover, how many pickers a peak flush actually requires — can be calibrated more tightly around what the data shows. That does not necessarily reduce the total number of workers employed across a season, but it does change how labour is organised across the farm. No public data yet tracks how these shifts play out in Lincolnshire's seasonal workforce specifically, which is a genuine gap in the picture rather than a reason to assume the change is either trivial or transformative.
The longer arc — from disease control to picking
Proof of concept has its own momentum. The fact that 150-plus Thorvald robots have accumulated more than 200,000 autonomous kilometres across UK soft-fruit farms — at 97% uptime and in commercial rather than trial conditions — changes what investors and growers believe is achievable on the next step. Autonomous crop-protection was the harder problem to demonstrate in real fields. Autonomous harvesting is the more complex engineering challenge, but the ground-level question — can a robot navigate a tabletop row reliably, night after night, without breaking down? — has now been answered.
The global agricultural robot market is projected to reach US$170 billion by 2032, with harvesting named as the primary application area. That figure is a direction-of-travel marker rather than a forecast to pin to a single farm, but the direction is clear: investment in autonomous picking platforms is following the crop-protection proof of concept, not preceding it.
For Lincolnshire's seasonal workforce, the more immediate question may be practical rather than existential. On a farm running Thorvald today, some workers already deal with charging schedules, flagging faults, or interpreting data dashboards that did not exist three seasons ago. In five years, on a more automated farm, might a returning seasonal worker spend part of a shift supervising a harvesting robot's progress through a row rather than picking it themselves? That shift — from physical repetition toward oversight and maintenance — is where the transition is most likely to begin. Whether farms, further education colleges, and training providers in Lincolnshire are prepared to support workers through that change is a question the technology is beginning to make urgent.
What this means for Lincolnshire's fields and the people in them
The farms where those results were recorded are not small-scale experimenters. They are among the UK's largest soft-fruit producers, running Thorvald at full commercial scale through entire growing seasons. The pace of adoption there sets the tempo for what the broader agricultural workforce can expect to face next.
The practical question for the region is whether investment in automation is matched by any comparable investment in transition. On a farm running a robot fleet tonight, workers are already absorbing tasks — charging schedules, fault reporting, data dashboards — that did not exist in the job three seasons ago. As platforms extend toward autonomous harvesting, the skills gap will widen. Riseholme College, the land-based further-education campus north of Lincoln and the county's established provider of agricultural qualifications, is the obvious institutional home for whatever curriculum shift prepares that workforce for the next stage. Whether its programmes have begun to reflect the skill set a farm running autonomous platforms actually needs is the kind of question that does not yet have a public answer — which is itself a useful indicator of where the transition currently stands.
The technology is not approaching. It is operating commercially in Lincolnshire's fields tonight. The human side of the shift — retraining, workforce planning, the practical question of what seasonal agricultural work here looks like in a decade — is moving more slowly, and in considerably less documented ways, than the robot fleet.
- [1] Lincolnshire. https://en.wikipedia.org/?curid=53295 https://en.wikipedia.org/?curid=53295
- [2] Agricultural robot. https://en.wikipedia.org/?curid=11005995 https://en.wikipedia.org/?curid=11005995
- [3] Migrant worker. https://en.wikipedia.org/?curid=1981818 https://en.wikipedia.org/?curid=1981818