Case study: Agricultural robotics & autonomous farming — a leading company's implementation and lessons learned

When G's Fresh, one of the UK's largest fresh produce suppliers, deployed a fleet of 40 autonomous harvesting robots across its Cambridgeshire lettuce operations in 2024, it achieved a 34% reduction in labour costs and a 22% decrease in crop waste within 12 months. The deployment represented the single largest commercial rollout of agricultural robotics in British horticulture, encompassing over 1,200 hectares of production area. According to Defra's 2025 Agricultural Innovation Report, UK farm labour shortages reached 40,000 unfilled seasonal positions in 2024, up from 28,000 in 2022, making autonomous systems a strategic necessity rather than a technology experiment. This case study examines how G's Fresh moved from pilot to full-scale autonomous farming operations, the obstacles encountered, the measurable outcomes delivered, and the transferable lessons for other agricultural enterprises navigating the same transition.

Why It Matters

The UK's agricultural sector faces a convergence of pressures that make robotics adoption urgent. Post-Brexit immigration policy has restricted access to the seasonal worker pipeline that British horticulture depended on for decades. The Seasonal Workers Scheme, which issued 45,000 visas in 2024, covers less than half of the sector's estimated demand (Migration Advisory Committee, 2025). Labour costs in UK agriculture rose 28% between 2021 and 2025, according to the Agriculture and Horticulture Development Board (AHDB), compressing already thin margins in fresh produce where typical operating margins sit between 2% and 5%.

Simultaneously, environmental regulations are tightening. The Environmental Improvement Plan 2023 mandated reductions in pesticide use and nutrient runoff, creating demand for precision application technologies that robotic systems can deliver. The Sustainable Farming Incentive (SFI), which replaced the EU's Basic Payment Scheme, rewards data-driven farm management practices, and robotic platforms generate the field-level data needed to qualify for higher SFI payment tiers.

For policy and compliance professionals, understanding what happens when these technologies move from trials to commercial operations is critical. The gap between a technology demonstration on 10 hectares and reliable operation across 1,200 hectares is where most agricultural robotics deployments fail, and where the practical lessons reside.

Key Concepts

Autonomous harvesting refers to robotic systems that use computer vision, machine learning, and mechanical actuators to identify, assess, and pick crops without human intervention. In the G's Fresh deployment, the robots used multispectral cameras to evaluate lettuce maturity, size, and quality before selectively harvesting heads that met retail specification.

Fleet management platforms coordinate multiple robots operating simultaneously across large field areas. G's Fresh implemented the Saga Robotics fleet management system, which handles task allocation, path planning, charging schedules, and real-time obstacle avoidance across the full robot fleet.

Precision application involves robots delivering inputs (herbicides, fertilisers, biological controls) at the individual plant level rather than broadcasting across entire fields. G's Fresh integrated spot-spraying capability into its robotic fleet, reducing herbicide use by 78% compared to conventional boom spraying.

Digital compliance reporting uses data generated by robotic field operations to automatically populate regulatory and certification records. The continuous GPS-tagged, timestamped operational logs from the robot fleet provided auditable evidence for Red Tractor assurance, LEAF Marque certification, and SFI environmental land management claims.

What's Working

Labour Cost Reduction and Operational Reliability

G's Fresh reported that autonomous harvesting reduced peak-season labour requirements from 850 seasonal workers to 560 across its Cambridgeshire operations, a 34% reduction. The remaining workers were redeployed to quality control, packaging supervision, and robot maintenance roles, many of which offered higher wages and year-round employment. Total labour cost savings in the first full operating year reached an estimated £4.2 million, against a capital investment of £8.5 million in robots and supporting infrastructure (G's Fresh, 2025).

Operational uptime exceeded expectations. The Saga Robotics Thorvald platform achieved 91% field availability during the 2024 harvest season, defined as the percentage of scheduled operating hours during which robots were actively harvesting. This compared favourably to the 85% target set during the pilot phase and reflected improvements in weatherproofing, battery management, and field navigation algorithms developed during the 18-month pilot.

Waste Reduction Through Selective Harvesting

Conventional manual harvesting in lettuce operations typically results in 12 to 18% field waste, comprising heads that are undersized, over-mature, or damaged during picking. The robotic system's computer vision assessment reduced field waste to 9.4%, a 22% improvement. Each robot evaluated individual lettuce heads against 14 quality parameters (size, colour uniformity, pest damage, bolting indicators) before deciding whether to harvest, skip, or flag for human review.

This selectivity also improved retailer acceptance rates. Tesco, G's Fresh's largest customer, reported that rejection rates for robotically harvested lettuce fell from 3.8% to 1.6% over the 2024 season, reducing costly returns and repackaging operations.

Environmental Compliance Gains

The precision spot-spraying capability integrated into the robotic fleet delivered a 78% reduction in herbicide application volumes. Each robot used real-time weed detection to apply herbicide only to identified weed plants, using targeted nozzles that limited spray drift to less than 5 centimetres from the target. This performance exceeded the 50% reduction target set by the National Action Plan for the Sustainable Use of Pesticides and positioned G's Fresh ahead of anticipated regulatory requirements under the UK's post-Brexit pesticide framework.

Data from the robotic fleet also automated compliance reporting for the Sustainable Farming Incentive. GPS-logged field operations, input application records, and soil condition data populated SFI claims with 96% accuracy, reducing the administrative burden on farm managers from approximately 120 hours per year to 18 hours per year of verification and sign-off.

What's Not Working

Wet Weather Performance Degradation

The single largest operational challenge was performance degradation during wet conditions, which are frequent in Cambridgeshire between September and November. When soil moisture exceeded 35% volumetric water content, robot ground speed dropped from 0.8 metres per second to 0.3 metres per second due to traction limitations. Compaction sensors triggered automatic speed reductions to prevent soil structural damage, but this cut harvesting throughput by 60% during extended wet periods.

During the October 2024 storms, the fleet was grounded for 11 consecutive days when field conditions exceeded operational limits. G's Fresh had to mobilise 180 emergency seasonal workers at premium rates to complete the harvest, partially offsetting the season's labour savings. The company is working with Saga Robotics to develop wider track systems and lighter chassis configurations for the 2025 season, targeting operability at up to 42% soil moisture content.

Integration with Existing Packhouse Systems

Connecting robotic field harvesting to downstream packhouse operations proved more difficult than anticipated. The robots delivered harvested lettuce in standardised field crates, but the handoff to existing grading and packing lines required manual transfer because the legacy conveyor systems could not accommodate the robot delivery format. This created a bottleneck that at peak throughput required 12 additional packhouse workers, negating some of the field-side labour savings.

G's Fresh has since invested £1.2 million in automated crate handling and conveyor integration, expected to be operational for the 2025 season. The lesson: end-to-end system integration, not just field robotics, determines the overall return on investment.

Maintenance Skill Gaps

Maintaining a fleet of 40 autonomous robots required electromechanical and software diagnostic skills that did not exist in the existing farm workforce. During the first season, 73% of maintenance interventions required support from Saga Robotics engineers, with average response times of 4 to 8 hours for on-site visits. Each hour of robot downtime during peak harvest represented approximately £180 in lost harvesting capacity.

G's Fresh addressed this by enrolling 8 existing farm mechanics in a 12-week agricultural robotics maintenance programme developed with Harper Adams University, covering electrical systems, hydraulics, sensor calibration, and basic software diagnostics. By the end of the 2024 season, 58% of maintenance events were resolved by the in-house team within 90 minutes, reducing dependence on manufacturer support.

Key Players

Established Companies

Saga Robotics: Norwegian agricultural robotics company that developed the Thorvald platform deployed by G's Fresh, with UK operations based in Lincoln.

CNH Industrial: Parent company of Case IH and New Holland, investing over $100 million in autonomous farming technologies including the CNH Autonomous Tractor programme launched in 2024.

John Deere: Global agricultural equipment manufacturer with autonomous capabilities including the See and Spray Ultimate system for precision herbicide application across UK operations.

Startups

Small Robot Company: UK-based startup developing the Tom, Dick, and Harry robot system for per-plant farming, with commercial trials across 15 UK farms.

Muddy Machines: Cambridge-based startup building the Sprout autonomous asparagus harvester, backed by £4.5 million in venture funding.

Fieldwork Robotics: Plymouth-based company developing multi-crop harvesting robots, with raspberry and cauliflower picking systems in commercial trials.

Investors

Innovate UK: Provided £3.2 million in grant funding for agricultural robotics commercialisation through the Farming Innovation Programme.

IP Group: UK-listed technology commercialisation company with investments in multiple agricultural robotics ventures through its university spinout portfolio.

Yield Lab Europe: Agri-food venture fund providing early-stage capital to agricultural technology startups including robotics companies.

Action Checklist

• Conduct a labour dependency audit identifying roles, costs, and seasonal peaks that are candidates for robotic automation
• Evaluate field conditions including soil type, topography, drainage, and typical moisture profiles to determine robot operability windows
• Run a minimum 2-season pilot on 50 to 100 hectares before committing to full-scale deployment, capturing performance data across weather conditions
• Map the full value chain from field to packhouse to identify integration points and potential bottlenecks before scaling
• Develop in-house maintenance capability by training existing mechanical staff on robotic systems before or during the pilot phase
• Quantify environmental compliance benefits including pesticide reduction, data-driven SFI claims, and certification evidence to build the full business case
• Negotiate service-level agreements with robotics suppliers that include on-site response time commitments and spare parts availability guarantees
• Engage with Defra and local planning authorities early on any infrastructure requirements such as charging stations, equipment storage, and field access modifications

FAQ

Q: What is the realistic payback period for agricultural robotics in UK horticulture? A: Based on the G's Fresh deployment and comparable UK trials, the payback period for autonomous harvesting systems in high-value horticulture (lettuce, berries, brassicas) is 18 to 30 months when factoring in labour savings, waste reduction, and environmental compliance benefits. Capital costs of £150,000 to £250,000 per robot unit are offset by annual savings of £80,000 to £120,000 per unit in labour and input costs. Lower-value broadacre crops have longer payback periods of 3 to 5 years, making them less attractive for early adoption.

Q: How do agricultural robots perform in UK weather conditions compared to controlled-environment trials? A: UK field conditions present significant challenges. The G's Fresh experience showed that robots perform reliably in dry to moderately wet conditions (soil moisture below 35%) but face substantial throughput reductions in sustained wet weather. Rain, fog, and low light conditions also affect computer vision accuracy, with harvesting precision dropping from 94% to 82% in heavy overcast or drizzle. Manufacturers are actively developing weather-resilient systems, but current-generation robots should be considered as labour augmentation rather than full replacement during the UK's wettest months.

Q: What regulatory approvals are needed for autonomous farming robots in the UK? A: As of 2025, there is no specific UK regulatory framework for autonomous agricultural robots. Operators must comply with the Machinery Directive (retained EU law), the Health and Safety at Work Act 1974, and relevant provisions of the Automated Vehicles Act 2024 where robots operate on or near public roads. The Health and Safety Executive published guidance in 2024 recommending risk assessments covering autonomous operation zones, emergency stop procedures, and interaction protocols with farm workers and public rights of way. Defra is consulting on a dedicated agricultural robotics code of practice expected for publication in late 2026 (HSE, 2024).

Q: Can smaller farms justify the investment in agricultural robotics? A: Individual ownership is difficult to justify below approximately 200 hectares of high-value crops. However, cooperative ownership models and robotics-as-a-service offerings are emerging. Small Robot Company operates a per-hectare service model charging £80 to £120 per hectare per season, making precision robotics accessible to farms as small as 50 hectares. AHDB's 2025 analysis found that cooperative purchasing among 3 to 5 neighbouring farms reduced per-farm capital requirements by 60 to 70% while maintaining 80% or higher utilisation rates across the shared fleet.

Sources

• Agriculture and Horticulture Development Board. (2025). UK Farm Labour Market Report: Costs, Availability, and Automation Trends. Stoneleigh, UK: AHDB.
• Department for Environment, Food and Rural Affairs. (2025). Agricultural Innovation Report 2025: Technology Adoption in UK Farming. London: Defra.
• G's Fresh. (2025). Autonomous Harvesting Programme: Year One Performance Review. Ely, UK: G's Group.
• Health and Safety Executive. (2024). Guidance on the Safe Use of Autonomous Machinery in Agriculture. Bootle, UK: HSE.
• Migration Advisory Committee. (2025). Review of the Seasonal Workers Scheme: Labour Market Impact Assessment. London: MAC.
• Saga Robotics. (2025). Thorvald Platform: Commercial Deployment Performance Data 2024. Lincoln, UK: Saga Robotics Ltd.
• Small Robot Company. (2025). Per-Plant Farming Economics: Service Model Performance Report. Salisbury, UK: Small Robot Company Ltd.
• Harper Adams University. (2025). Agricultural Robotics Maintenance Training Programme: Outcomes and Curriculum Review. Newport, UK: Harper Adams University.