Fieldwork Robotics bets on managed fleets for autonomous raspberry harvesting

Why it matters

Fieldwork is packaging physical AI as harvesting capacity, maintenance and data. Its 2027 trials will test whether managed robot fleets can produce workable farm economics.

Fieldwork Robotics is moving Fieldworker 1, the autonomous raspberry picker shown in a video posted on X, into a two-year UK farm program designed to test whether one operator can supervise multiple robots at commercial scale. The Cambridge-based startup says the four-arm machine can match a human picker's speed and quality, a claim the deployments with Place UK and Littywood Farm will have to prove under normal farm conditions.

Founder and Chief Science Officer Dr. Martin Stoelen built Fieldwork Robotics around research he led at the University of Plymouth on soft and adaptive robotic arms. Stoelen, a Norwegian roboticist who earned his PhD from Universidad Carlos III de Madrid in 2014, now teaches at Western Norway University of Applied Sciences while remaining Fieldwork's technical founder. His work has focused on a problem that exposes the limits of conventional industrial automation: handling irregular, delicate objects hidden inside foliage without crushing them.

Fieldwork Robotics has spent nearly a decade narrowing that research into a commercial proposition. The product strategy now reaches beyond selling individual machines. Fieldwork wants to provide growers with coordinated fleets, maintenance, agricultural data services and harvesting-as-a-service, placing the cost of the robot against the quantity of marketable fruit it collects.

Selling an operating cost, rather than a machine

Fieldworker 1 has four independent arms, AI-enabled 3D vision and high-fidelity ripeness detection, according to a June 2026 investment disclosure. Each arm uses a patented inflatable membrane intended to remove ripe fruit without damaging it. Fieldwork also says the system uses spectral-frequency analysis to make ripeness decisions and can coordinate multiple robots under one operator.

That fleet architecture is the core of the business case. A machine that picks berries autonomously still requires transport, charging, cleaning, maintenance and supervision. Fieldwork's harvesting-as-a-service model lets a grower pay for access to harvesting capacity while Fieldwork retains responsibility for the hardware and its performance. Direct robot sales remain part of the plan, alongside service contracts and data products.

Fieldwork markets the result as a predictable cost per berry. Fieldwork has not disclosed pricing, current fleet size or unit-level harvesting economics, so that promise remains unverified. The farm program will need to establish picking speed, uptime, fruit quality and the number of machines a single operator can supervise. Those measurements will determine whether Fieldworker 1 reduces labor costs after maintenance and field logistics are included.

The machine itself reflects several redesigns. In 2022, the University of Plymouth reported that an earlier four-arm version was operating at two sites in Portugal. Fieldwork had redesigned its sensors and grippers to reduce slippage, and the harvested fruit had passed commercial quality controls. At that point, Fieldwork was still working toward a target of two kilograms per hour.

Fieldworker 1, announced in 2024, added faster picking and more precise ripeness detection. Fieldwork said trials with Australian grower Costa Group would begin in the first quarter of 2025. Its 2026 announcements discuss further planned trials in Australia without publishing results from that original program.

The deployment timetable has stretched

Fieldwork's latest financing is meant to move the system from validation into sustained farm use. SEED Innovations invested GBP 300,000 in June for an approximately 3.66% stake. The investment completed a GBP 2.5 million Seed+ fundraise that included a GBP 2.2 million round led by Elbow Beach Capital in April. Fieldwork also secured GBP 1.7 million in grants from Innovate UK and a climate grant.

Elbow Beach contributed GBP 1.45 million in follow-on capital after investing GBP 1.5 million in 2023. Fieldwork said the new money would fund production robots for the UK trials, with international work planned in Portugal and Australia.

The timeline shows how difficult agricultural robotics is to commercialize. In August 2023, Fieldwork said it aimed to have over 100 robots available through harvesting-as-a-service by 2025 and to triple harvesting speed. Fieldwork's current materials instead target multi-robot farm fleets from 2027, subject to successful trials.

SEED's June disclosure described Fieldwork's turnover for the year ended October 31st, 2025 as "modest." Fieldwork reported net assets of about GBP 5.54 million for that period. The disclosure did not give revenue, losses, bookings or signed fleet orders.

That gap separates technical validation from a repeatable business. Fieldwork has shown that its robots can find and remove commercially acceptable raspberries. It still has to demonstrate that machines can operate long enough, cheaply enough and with little enough human intervention to justify deployment across a farm.

Stoelen chose the crop that tests the whole system

Raspberries are a demanding starting point. The fruit bruises easily, ripeness varies across a plant and leaves can obstruct a robot's view. Earlier Fieldwork systems used cameras, sensors and machine learning to determine ripeness, then applied pressure around the stem rather than squeezing the berry. The newer inflatable membrane extends Stoelen's work on compliant mechanisms that can adapt to objects instead of forcing every object into the same rigid grip.

Success with raspberries could give Fieldwork a technical base for other crops, but each crop introduces new plant structures, harvesting motions and commercial requirements. Fieldwork has previously explored tomatoes and cauliflowers, while its current sales effort remains concentrated on fresh raspberries.

Fieldwork's decision to stay focused gives the UK spinout a clear test. The 2026 funding buys production hardware and time on farms. The 2027 fleet target requires Fieldworker 1 to become a dependable piece of agricultural equipment, supported by service operations that can keep machines running during a short and unforgiving harvest window.

For growers, the important output will be marketable fruit collected per hour at a known cost. For Fieldwork Robotics, the decisive metric will be whether each deployed robot produces recurring service revenue without requiring an uneconomic amount of human support.