Why Lincolnshire's fields need more than seasonal workers
Every summer, the fields around Boston, Spalding and the Fens fill with workers who have travelled thousands of miles to pick vegetables that will appear on supermarket shelves within days. Greater Lincolnshire produces roughly 30% of the UK's vegetables and around 12.5% of total UK food output — drawn from approximately 25% of England's Grade 1 arable land. The agrifood chain here employs more than 75,000 people and generates around £2.5 billion a year. At peak harvest, South Lincolnshire alone dispatches between 1,500 and 3,000 lorry-loads of food daily.
Almost none of that is possible without seasonal migrant labour. The Seasonal Worker Visa route is the primary mechanism keeping harvests viable, and it exists precisely because domestic supply cannot fill the gap. A joint industry report estimated 500,000 vacancies across UK food and farming nationally. Locally, the workforce is both shrinking and ageing, and post-Brexit immigration constraints have made an already tight situation structurally precarious.
This is not a crisis waiting to happen at some point in the future. It is a pressure that farm operators in Lincolnshire manage every season, and one that a declining supply of willing workers makes harder each year. The question of how to maintain food output from this region without assuming an inexhaustible pool of seasonal labour is, increasingly, a practical engineering problem — not a political one. That is precisely why the robotics research taking shape at the University of Lincoln is less about novelty than necessity.
What JABAS.AI actually does — and why GPS was the problem
Picture a robot that navigates open fields with ease, following a GPS-defined path between rows of crop. Now wheel it into a polytunnel — a low-roofed structure packed with strawberry plants, metal supports and irrigation lines — and the satellite signal that guided it disappears almost entirely. The same failure happens under dense canopy or close to farm buildings. For conventional agricultural robots, which depend on GPS waypoint navigation to know where they are and where to go, these environments are effectively impassable. Yet polytunnels and covered structures are precisely where much of Lincolnshire's intensive soft fruit production takes place.
That is the practical bottleneck that JABAS.AI — Just a Better Autonomy Solution — was built to address. Launched on 13 May 2026 as the fifth spin-out from Ceres Agri-Tech, the University of Lincoln–Cambridge–UEA partnership funded by Research England and EPSRC, it replaces satellite dependency with sensors that read the physical world directly. The platform combines LiDAR — which bounces laser pulses off surrounding surfaces to build a spatial map — with computer vision and real-time localisation algorithms. Rather than asking 'where am I on a pre-loaded map?', it continuously constructs and refines an understanding of its environment as it moves. Unpredictable layouts, cluttered aisles and shifting obstacles are conditions it is designed to handle rather than avoid.
Equally significant is how it is offered. JABAS.AI operates as 'autonomy-as-a-service', meaning the platform integrates with robots a farm already owns. There is no requirement to replace existing machinery — the software layer bolts on, giving a fleet of disparate machines shared navigation capability and the ability to coordinate tasks such as locating workers or transporting harvested produce.
The project's founder and CTO is Professor Marc Hanheide, the University of Lincoln's Professor of Intelligent Robotics, whose laboratory work underpins the localisation approach.
The fleet shift — from a single machine to coordinated autonomy
The problem with the early wave of agricultural robots was not the hardware — it was the paradigm. A single machine, dispatched to complete a defined task in a controlled patch of field, can deliver results in a trial. Deploying that same machine commercially, across a real farm's shifting layouts, variable lighting, leaf canopy and human foot traffic, is where adoption stalled. GPS-dependent systems could not cope reliably enough with that variability to justify the investment at scale.
What JABAS.AI proposes is a different unit of deployment altogether. Rather than one robot doing one job, a coordinated fleet operates as a distributed logistics system: machines locating workers in the field, navigating around each other and around people, and moving harvested produce from picker to collection point continuously. On soft fruit farms, an estimated 15–20% of all working time is spent simply moving trays of picked fruit — a physically repetitive task that adds no value to the crop but consumes a significant share of every working day. That is precisely the kind of task a coordinated fleet can absorb, freeing human labour for work that requires judgement.
Commercial operators in Lincolnshire are already beginning to use digital-twin simulations and predictive swarm-coordination models to optimise multi-robot task allocation across crop care, phenotyping and harvesting — though these frameworks are described as entering commercial use rather than proven at scale. JABAS.AI is better understood as infrastructure on which other robots operate than as a standalone product: a shared navigation and coordination layer that makes the fleet, rather than the individual machine, the meaningful unit of farm automation.
How Lincoln built the conditions for this to happen
Spin-outs do not emerge from thin air. The University of Lincoln's Lincoln Agri-Robotics (LAR) — recognised in the UK Innovation Strategy of July 2021 as the world's first global centre of excellence in agricultural robotics and awarded the Queen's Anniversary Prize in February 2024 — has spent years running the kind of sustained, iterative research programme that makes a commercialisable platform possible. What matters practically is the infrastructure underneath: 149 projects worth over £100 million, conducted alongside more than 250 industry partners who provided the real farm conditions in which the technical problems could be properly defined.
The Ceres Agri-Tech portfolio shows what that looks like across successive spin-outs. FruitCast addressed one specific forecasting problem in strawberries; Agaricus Robotics targeted one harvesting challenge in mushrooms. Each was a bounded solution. JABAS.AI is structurally different — it is the coordination layer on which those bounded solutions can actually scale. That shift from crop-specific tools to shared fleet infrastructure reflects the developmental logic of the pipeline, not a fortunate fifth attempt.
Regulation remains a live constraint. ARRNet, led by the UK Agri-Tech Centre with the University of Lincoln as standards and testing lead, is working to clarify a fragmented regulatory environment that would otherwise slow fleet-scale deployment before it starts. The AgriFoRwArdS Robotics summit, convened in Lincoln in July 2026 within two months of JABAS.AI's launch, drew more than 100 national researchers, policymakers and industry leaders — evidence that the question of fleet autonomy has moved from research agenda to commercial timetable.
What a fleet-managed farm could look like in practice
For a farm manager who has already bought a harvesting arm or phenotyping unit, the question JABAS.AI poses is not whether to invest in more technology — it is whether the machines already on the farm can be made to work together. The 'autonomy-as-a-service' model is built for exactly that position: the coordination layer runs on existing hardware, so prior capital investment stays productive rather than becoming a liability to be written off.
That reframes the adoption decision from 'should we replace our equipment' to 'how do we get more from what we have.'
The change is just as significant for the people on the farm. For a picker, fleet coordination alters the spatial rhythm of a shift: instead of breaking from the row to carry produce to a collection point, the worker stays in place while the fleet handles movement. The physically repetitive logistics task is absorbed by the machines; human attention concentrates on the picking itself. The physical demands of the job shift alongside the task distribution.
Farm management decisions change too. Real-time localisation data across a whole site means a manager can see where workers are concentrated, where produce movement is stalling, and where a reallocation of robot resource would ease a bottleneck — operational visibility that open-field farming has historically made difficult to achieve at any useful resolution.
This is not a scenario without workers. It is one where fewer workers may cover more ground, and where the character of the work — its physical demands, its spatial rhythms — changes rather than disappears. Current testing on commercial farms has not yet produced independently verified throughput or efficiency figures, so what fleet autonomy delivers at operational scale in Lincolnshire conditions remains to be established.
What still needs to be proved — and what it means for the region
Before any fleet can operate commercially, there is a certification threshold to clear. ARRNet's 12-month, government-funded programme to develop standards and testing frameworks for multi-robot agricultural systems is a prerequisite for scale: without it, a platform that works in trials has no clear route to the legal and insurance frameworks that commercial deployment requires. The University of Lincoln leads the standards and testing work within ARRNet — so the regulatory pathway and the technology being assessed for it are on the same timetable.
That gap matters because the distance from current testing to county-wide deployment is not primarily a technical question. It is a standards and commercial-confidence question whose answer is still being built.
The workforce consequence deserves direct attention. Lincolnshire's agrifood sector employs more than 75,000 people; the Seasonal Worker Visa is currently the mechanism keeping labour supply and production together. Fleet autonomy would change what farms need from that workforce — fewer hands on logistics and movement, more on tasks machines cannot yet perform — and that shift, over years rather than months, alters the employment picture for communities across South Kesteven and the Fens in ways that are worth thinking through now.
JABAS.AI is a credible systems-level bet, backed by a serious institution and a specific regional need, and as of July 2026 it is still in testing. The useful question for people in Grantham is not abstract concern about automation — it is a practical one about what kind of food system this region is choosing to build, and who is part of making that choice.