A county where agriculture is not background scenery
Drive south from Lincoln on the A15 and the landscape flattens quickly into some of the most productive farmland in England. The fields here — stretching across the Fens and the Holland plain — grow a third of the nation's vegetables. Supermarket supply chains run directly through South Holland, through Boston, through the low horizons of a county that rarely appears in conversations about Britain's economic identity but sits at the centre of its food system.
The numbers make the point plainly. The food chain accounts for 24% of Greater Lincolnshire jobs — nearly double the 13% national average. In South Holland district, agricultural employment exceeds 40% of the local workforce. These are not marginal figures. When a sector this large is disrupted, the effects are felt in housing, in local shops, in school rolls, in everything.
For decades, that workforce included hundreds of thousands of seasonal workers arriving from Eastern Europe each summer to pick, pack, and tend crops. Then Brexit ended EU free movement. In 2021, the government issued 30,000 seasonal worker visas against an industry estimate of 80,000 to 90,000 needed. The gap was not theoretical: crops were left unharvested in fields. Growers did not turn to robotics because automation was fashionable. They turned to it because the alternative was watching produce rot.
That is the context in which farm robots arrived in Lincolnshire — not as a tech-sector story, but as a response to a structural labour crisis. Which makes the central question here a human one: when machines begin taking on work that people used to do, what actually changes for the workers who remain?
What Lincoln's robotics researchers set out to solve
The MeSAPro project, which ran from 2020 to 2022 as a collaboration between the University of Lincoln's Lincoln Centre for Autonomous Systems (LCAS) and Saga Robotics, was set up with a specific mandate: make working alongside robots safer, not make robots a substitute for workers.
The safety case behind that mandate is striking in its own right. Agricultural workers make up just 1% of the UK workforce, yet account for 20% of all workplace fatalities — a disproportion that reflects the physical hazards of outdoor, machinery-intensive labour. MeSAPro was designed to reduce that risk rather than sidestep it by removing workers from the equation.
In practical terms, the project delivered hazard analyses for four distinct soft-fruit deployment scenarios and formal verification methods for the robot's sensing and decision-making systems. Its most tangible output was a Human-Aware Navigation system, integrated directly into Thorvald's software stack: the robot navigates predictably around human pickers, adjusting its path in real time rather than treating workers as obstacles to route around at the last moment.
This work sits within a broader programme at Lincoln's Institute for Agri-Food Technology (LIAT), based at the 200-hectare Riseholme Campus. Active projects include the Agaricus Robotic Harvester — Innovate UK-funded and described as the world's first commercial mushroom-harvesting robot — alongside AGRI-OPENCORE, an open-architecture platform intended to let different manufacturers build compatible harvesting tools across a range of crops. The AgriFoRwArdS Centre for Doctoral Training, also at Lincoln, sustains a pipeline of PhD researchers across agri-food robotics nationally. What unites these is a consistent design logic: build the research infrastructure first, then let commercial applications follow — rather than the reverse.
What the robot is actually doing in the field tonight
Picture Thorvald at work on a summer night: a low, four-wheeled platform moving slowly between tabletop strawberry rows, emitting high-intensity UV-C light that disrupts fungal DNA at the molecular level. No chemical sprays, no human operator walking the rows. Just the robot, the light, and the elimination of powdery mildew before dawn. By the close of the 2025 UK season, more than 150 Thorvald robots had covered over 200,000 autonomous kilometres at 97% uptime — accounting for around 20% of the national tabletop strawberry market.
What is being displaced here is not the hand-picking season but the chemical application rounds and nighttime inspection shifts that bracket it. The seasonal migrant workers who arrive each summer to harvest fruit are not being replaced by Thorvald; the fungicide-spraying runs and nocturnal crop-monitoring tasks are. These are real labour reductions, but they fall on a different part of the workforce calendar.
Full picking automation is a separate ambition, and a more distant one. The Robot Highways project — £2.5m from Innovate UK, led by Saga with Berry Gardens and BT — is working toward pick-and-pack automation using 5G and edge computing at Clock House Farm in Kent. It remains a live research and development programme, not a deployed commercial system. Current Thorvald deployment and future robotic harvesting are distinct stages, with very different implications for who does what work on Lincolnshire farms.
Why picking fruit by hand is harder to automate than it looks
Soft fruit presents a genuinely awkward engineering problem. A ripe strawberry sits among leaves — sometimes hidden by them — at a stem angle that varies plant to plant. Its colour deepens unevenly across the surface, making a single colour threshold unreliable as a ripeness signal. Grip pressure must be calibrated precisely: too firm and the fruit bruises; too light and it drops. In a polytunnel with consistent overhead lighting these challenges are manageable; add variable natural light, irregular stem heights, and clusters where fruits at different ripeness stages sit millimetres apart, and the difficulty compounds significantly.
State-of-the-art systems, tested under controlled greenhouse conditions, achieve around 80% success at approximately 20 seconds per fruit — figures that represent something close to best-case performance, not field reality. Skilled human pickers work considerably faster and with higher success rates, adapting instinctively to the variability that trips up current vision and manipulation systems. The gap is not evidence of researchers missing something obvious; it reflects the genuine complexity of a task humans perform with a physical intuition that is hard to encode.
AGRI-OPENCORE, the open-architecture harvesting platform in development at Lincoln's Riseholme Campus, represents the direction of travel: a modular system that could eventually be configured for different crops and picking styles. It is a research milestone, not a commercial deployment. Robotic displacement of seasonal pickers at scale remains a technical horizon being worked toward — not something the current engineering timeline puts close at hand.
Seasonal workers, permanent staff, and where the pressure lands first
The workforce most affected by farm automation is not a single category — it divides along a fault line between seasonal and permanent employees, and the near-term pressures fall very differently on each side.
Permanent farmhands are largely insulated for now. Where change is reaching them, it tends to bring new responsibilities: supervising autonomous platforms, reading yield and crop-health data, maintaining machinery that did not exist on the farm a decade ago. The direction of travel is toward technical oversight rather than physical crop-care, and it does not, at present, threaten their employment.
Seasonal migrant pickers face a different long-term picture. As picking automation matures — still a technical horizon, not a commercial reality — harvesting roles carry the highest exposure to displacement. What is already shifting is the surrounding crop-care work: chemical spraying rounds, nighttime monitoring, and logistics tasks are the functions being absorbed first, as earlier sections have shown. The picking season itself is not yet under pressure.
No published figures break down the seasonal agricultural headcount for Lincolnshire's soft-fruit sector specifically, and the county's outsized dependence on the food chain makes that absence meaningful rather than trivial.
The larger gap is harder to quantify. The debate about automation in these fields is being conducted between researchers, technology companies, and funders. The pickers themselves are not part of that conversation — at least not in any public record. Their experience of working alongside robots, their practical knowledge of what machines cannot yet do, their anxieties or adaptations: none of it has been gathered or heard. The discussion about their industry is happening without them.
The farms that cannot afford the robots
The question running through this article — what robots mean for the people who work on Lincolnshire's farms — carries a prior question that the evidence makes uncomfortable to sidestep: which farms can afford the robots in the first place?
Thorvald is a commercial product, not public infrastructure. The large berry operations with retailer relationships, investor backing, and the volume throughput to absorb capital costs can make that calculation work. Smaller family farms — and Lincolnshire has many — face a different arithmetic entirely. For them, seasonal migrant labour, however constrained post-Brexit, remains the cheapest available option. A 2023 House of Commons Library briefing put the structural tension plainly: reliance on imported seasonal labour suppresses agricultural productivity and stalls the investment in automation that might eventually raise it. The bind is circular. Without capital, farms cannot automate; without automation, they cannot generate the productivity gains that might fund capital investment.
The risk, if that dynamic holds, is a two-tier county: large automated operations expanding market share while smaller farms remain labour-dependent and increasingly uncompetitive. That kind of divergence tends to resolve through consolidation rather than catch-up.
What Lincolnshire's agricultural landscape looks like in ten years is a genuinely open question. The honest answer is that it depends not only on what robots can do, but on who can afford to find out.
- [1] Development of an Agricultural Robot for Strawberry Fruit Harvesting and Truss Pruning. (2024). https://doi.org/10.1109/SII58957.2024.10417299 https://doi.org/10.1109/SII58957.2024.10417299
- [2] Agricultural robot — Wikipedia. https://en.wikipedia.org/?curid=11005995 https://en.wikipedia.org/?curid=11005995