Strawberry farming innovations: robots and decision support tools
Econome à LégumesProfessional strawberry growing is going through a period of profound transformation. On one side, economic pressure shows no signs of easing: harvesting accounts for 60 to 70% of variable production costs, skilled labour is increasingly scarce, and margins on long distribution circuits continue to shrink. Losses from poorly managed irrigation or fertilisation can exceed 15% of marketable yield in a single season. Disease pressure — Botrytis, powdery mildew, Drosophila suzukii — demands ever-finer responsiveness, even as the availability of field advisors declines and their costs rise. On the other side, a wave of technological innovations in strawberry farming is reshaping what is possible: AI-vision harvesting robots, real-time connected decision support tools, IoT sensor networks in the field, and early experiments with vertical indoor growing.
These innovations are not trade show gadgets. They address concrete operational challenges that every strawberry production manager knows: how to harvest faster without sacrificing fruit quality? How to anticipate a Botrytis outbreak before it takes hold? How to turn a stream of sensor data into an actionable irrigation decision?
But technology does not mechanically simplify crop management. It shifts the need for agronomic expertise — towards interpretation, contextualisation, and situation-specific decision-making. This article takes stock of the real state of these innovations, what they can deliver today, and what they still cannot do alone.
Fraisibot supports you on the most complex technical decisions in your production
Before going further, three questions regularly asked by strawberry growers who have adopted digital tools:
- My irrigation DSS is flagging a water deficit at D+2, but my Clery plants are at the end of their second flush with a maximum fruit load. Should I manage irrigation the same way as at the start of the season?
- My substrate sensor reads 22% VWC this morning. The forecast shows three overcast days ahead. Should I irrigate or wait?
- My Botrytis risk model has just gone to orange alert. I applied a biocontrol product five days ago, and the tunnel was ventilated this morning. What is my real intervention priority tonight?
None of these questions have a generic answer. They depend on your variety, your substrate, your phenological stage, and your treatment history. Fraisibot, your specialist AI agronomist for strawberry growing, integrates these parameters to help you decide — not with a standardised answer, but with reasoning adapted to your actual situation.
Harvesting robotics in strawberry growing: where do things really stand?
Autonomous systems — vision, grasping, throughput
Strawberry harvesting has always been the operation that concentrates the most labour and the most economic pressure. In table-top soilless growing, an experienced picker harvests 20 to 30 kg per hour in good conditions. At ground level, this figure drops to 7 to 15 kg per hour depending on access conditions and fruit load.
Current strawberry harvesting robots — including the most advanced prototypes such as Tortuga AgTech and Octinion — share a common architecture: computer vision to identify ripe fruit by colour and shape analysis, articulated arm or pneumatic gripper for picking, and autonomous navigation on standardised raised rows. The Tortuga robot, awarded in 2024, uses AI to recognise ripe strawberries with accuracy documented as superior to human picking, and can operate 24 hours a day without interruption. In testing, documented throughput rates range between 15 and 20 fruits per minute.
The economic model is structurally significant: a single supervisor can manage up to 15 units simultaneously. Compared to the hourly cost of manual harvest labour, the return on investment projection becomes theoretically attractive beyond certain area thresholds. In practice, however, this calculation remains theoretical for the vast majority of growing operations today.
Adoption conditions in professional French and European growing operations
Strawberry harvesting robotics is not yet a mainstream technology for small and medium-sized French horticultural businesses. Several structural constraints explain this.
Growing system compatibility. Current robots are designed for standardised, raised rows with clear inter-row spacing and a regular canopy height. Table-top soilless growing is the only format that is truly compatible. Open-field growing on mulched ridges remains out of reach for fresh market strawberries — the full mechanisation of ground-level harvesting, practised in the United States for processing strawberries, involves losses and quality levels incompatible with European fresh market standards. In France, approximately 60% of volumes are still produced in soil; robotic harvesting would therefore concern, in the short term, only a fraction of the national growing area.
Cost and return on investment. These high-technology systems remain expensive and are considered uneconomical at scale except for highly specialised operations. The break-even point depends on local labour costs, growing area, and production model. For a mid-sized operation (less than 5 ha of soilless strawberries), the financial equation is generally unfavourable today. The calculation shifts in areas where labour costs are structurally high and seasonal worker shortages are recurring — precisely the context where the first commercial adoptions are taking shape, notably in Belgium and the Netherlands, where intensive soilless greenhouse production is already highly developed.
Real technical limitations. Current robots struggle with trailing varieties or irregularly shaped fruits, rows with uneven ripeness, and variable lighting conditions. Bruising remains an unresolved issue for some thin-skinned varieties. Harvest selectivity — the ability to pick only fruit at the correct ripeness stage — is well handled by vision systems, but requires growing uniformity that is difficult to achieve without significant standardisation of plant material and practices. This standardisation has a cost: it can reduce varietal flexibility and limit the possibilities for extending the season through complementary cultivar combinations.
Robots for other tasks — weeding, crop protection, surveillance
Harvesting robotics captures most of the attention, but it is not the only application under development. Several complementary systems are emerging in strawberry growing, covering time-consuming operations that also weigh heavily on labour per hectare.
Autonomous weeding robots. Small autonomous machines capable of navigating between rows and mechanically cultivating the alleys are under development and testing. In strawberry growing under plastic mulch, weed pressure in the alleys remains a real constraint, particularly in organic production where synthetic herbicides are excluded. These machines are more advanced today on vegetable crops with wider inter-rows — fennel, leeks, cabbages — but trials on strawberries are beginning, driven by specialist manufacturers. The specific challenge for strawberries: avoiding damage to the fragile stolons and petioles during inter-row passes.
Rail-mounted spray systems. Automated spray booms on overhead rails, integrated into strawberry greenhouses, can apply biological night-time treatments with precision on every row, without an operator present. The benefit is threefold: time savings on applications, perfectly consistent dosing, and elimination of operator exposure to products — an increasingly important argument as phytosanitary regulations tighten. These systems are operational in some large Belgian and Dutch greenhouse operations. Their adaptation to smaller structures under movable plastic tunnels remains a future development.
Surveillance drones. Drones equipped with multispectral cameras can detect zones of water stress or pest attack through leaf canopy colour analysis, before symptoms are visible to the naked eye. In low-tunnel or open-field strawberry growing, the application remains experimental: the low plant canopy, high planting density, and canopy structure complicate image interpretation. Results are more reliable in higher-canopy soft fruit orchards. However, for monitoring large open-field plots — everbearing strawberries in open fields in south-west France, for example — drone surveys for early detection of Botrytis outbreaks or chlorosis zones linked to EC drift could become a relevant support tool in the medium term.
Decision support tools in strawberry growing: a new approach to irrigation and nutrition management
Connected tensiometry and water management
The strawberry plant has a shallow root system and is extremely sensitive to water fluctuations. Frequent, small applications are structurally preferable to infrequent, high-volume irrigations. This is the primary argument for connected irrigation management tools: they enable frequent decisions based on continuous measurement, rather than a single empirical daily assessment.
Tensiometric probes (such as Watermark sensors) and capacitive probes are the two reference technologies for water management in soilless strawberry growing. They measure respectively the matric tension of the substrate (in kPa) and volumetric water content (% VWC). Documented reference thresholds by phenological stage are as follows:
- Post-planting establishment: maintain between 5 and 15 kPa, with high moisture levels for the first 15 days to secure root establishment
- Vegetative growth: 10 to 20 kPa, adjusted according to local evapotranspiration and weather conditions
- Flowering: 10 to 15 kPa — consistency is absolute at this stage; any water stress directly compromises fruit set
- Fruit development: 10 to 15 kPa — the most water-demanding stage, accounting for approximately 50% of seasonal water volume
In practice, a tensiometry-based irrigation DSS typically uses a trigger threshold of 15 kPa and a stop threshold of 8–10 kPa depending on the stage. These values provide a solid baseline — but they do not replace stage-specific, variety-specific, and growing-system-specific agronomic reasoning.
Nutrition management is integrated into this same framework: electrical conductivity (EC) of the nutrient solution is a parameter to monitor continuously in soilless production. Strawberry plants are highly sensitive to salinity. Irrigation water EC should ideally remain below 1.2 mS/cm. In soilless substrate monitoring, drainage EC is generally maintained between 1.4 and 2.2 mS/cm — beyond this, toxicity symptoms appear (leaf necrosis) depending on the variety.
Epidemiological models — Botrytis, powdery mildew and crop protection decision support
Crop health risk management DSS represent the other major axis of digitalisation in strawberry growing. Their principle: modelling fungal risk from continuous climate parameters, to advise growers on the optimal time to intervene.
For Botrytis cinerea, models such as MS-BOT and BoMa use temperature (maximum risk between 15 and 25°C) and relative humidity (triggered above 75% RH) as input variables. The phenological stage is decisive: risk is highest during flowering and fruiting. It is during flowering that the fungus preferentially colonises petals and stigmas before progressing to the calyx and fruit. During harvest, it spreads exponentially on ripe or damaged fruit. The accuracy of the DSS depends on the quality of the climate data fed into it: a weather station positioned outside the tunnel does not faithfully reflect actual conditions inside, where temperature and RH can differ significantly depending on the time of day and ventilation status.
For powdery mildew (Podosphaera aphanis), the SPAW model uses a different logic: risk intensifies during alternating warm days and cool nights, which generate morning dew that extends the duration of leaf wetness. This climatic pattern, common in spring under plastic tunnels, is precisely the context in which powdery mildew can progress rapidly on susceptible varieties. Excess nitrogen — common early in the season when pushing vegetative growth — amplifies powdery mildew susceptibility by promoting soft, dense foliage: climate DSS and nutrition management are therefore closely linked.
The real contribution of these DSS tools is not to replace field observation, but to rationalise it. A model may flag that risk has reached orange alert on a Wednesday evening when you had not planned a field visit until Friday. Without a modelling tool, this critical intervention window goes unnoticed. With one, you can decide whether to bring forward your visit — provided you combine the model's information with your real-world knowledge of the situation: tunnel condition, date of last treatment, sensitivity of the variety in place. French technical institutes (CTIFL) and their Belgian (PCfruit) and Swiss (Agroscope) counterparts develop and refine these models each season through comparative trials, calibrating alert thresholds to regional conditions.
Sensors and IoT in strawberry growing — building your plot-level dashboard
Field sensors — what can be measured today
A sensor network in a professional strawberry operation rests on several complementary measurement layers:
Climate parameters (via weather station or micro-station inside the tunnel): air temperature, relative humidity, wind, rainfall and global radiation (PAR/ETP). This data feeds both epidemiological models and the calculation of potential evapotranspiration for irrigation management.
Substrate parameters: tensiometric probes for matric tension, capacitive probes for volumetric moisture, EC and pH probes for monitoring nutrient solution in soilless production. Optimal solution pH sits between 5.8 and 6.5 for maximum nutrient uptake efficiency.
Canopy parameters: infrared thermometers measuring temperature directly at bud and leaf level — particularly useful for frost risk detection. The critical alert threshold, at the open flower stage, is -0.5°C: from -1 to -2°C, the pistil is destroyed. This parameter is decisive in spring for crops under unheated tunnels.
Reference values for optimal growth temperatures are well documented: 18 to 22°C during the day, 10 to 13°C at night. Stress alerts are triggered below 5°C (vegetation stops) and above 30°C (heat stress, fruiting inhibited). Below 6 hours of PAR radiation per day, floral initiation can be compromised in photoperiod-sensitive varieties.
The architecture of an IoT network in a strawberry operation must account for practical constraints: for an area of 0.5 to 3 ha under tunnel, a LoRaWAN or 4G gateway centralising 3 to 5 measurement points per growing block is a coherent setup. A 15 to 30-minute acquisition frequency captures critical dynamics without generating an unmanageable data volume.
Integration into daily crop management decisions
The ideal dashboard for a strawberry crop manager combines three real-time data streams: field measurements (substrate, climate, canopy), the 48–72h climate forecast, and the current phenological stage of the crop. It is the articulation of these three levels that produces relevant decision support — not the raw measurement taken in isolation.
A concrete example: a capacitive probe reads 28% VWC in the substrate at 8am. Is this a signal to trigger irrigation? It depends on the substrate being used (rockwool and coir have different water retention curves), the phenological stage (active fruit swelling or end of maturation?), the fruit load on the plant, and the 48-hour weather forecast. The raw measurement is universal. The decision is always contextual.
Another example: drainage EC reads 2.8 mS/cm this evening. Should the nutrient solution concentration be reduced? Or is this an afternoon evapotranspiration spike that will normalise by tomorrow morning? The answer depends on the 48-hour trend, the solution temperature, and the phenological stage — a plant at the end of fruit swelling tolerates EC drift less well than a plant in active vegetative growth.
This is where a real risk emerges: over-instrumentation without an interpretive framework. A grower who receives real-time drainage EC, substrate moisture, canopy temperature, PAR and Botrytis alerts — without an agronomic reading framework to prioritise these signals — does not make better decisions. They make more stressful ones. Data is not advice. It is its raw material. The challenge is not having more measurements, but having someone or something capable of contextualising them to your specific situation.
Fraisibot answers your agronomic questions about your strawberry crop, integrating your actual situation parameters. Not a generic threshold from a technical datasheet — reasoning adapted to your stage, your substrate, your history.
Towards indoor strawberries and vertical farming — realities and perspectives over the next 5–10 years
The state of vertical strawberry growing experiments
Vertical strawberry growing — hydroponic towers, strawberry walls, multi-tier growing systems under artificial LED lighting — is an experimental reality in several countries. Operators such as Dyson Farms in England have presented high-tech vertical strawberry farms capable of producing for urban markets out of season. Some Scandinavian and Japanese supermarkets already sell strawberries from these systems.
The documented advantages of these configurations are real: complete climate independence, near-total suppression of foliar diseases (Botrytis, powdery mildew) through ambient humidity control, higher productivity per square metre of floor space than conventional growing, and off-season production capability.
The main barrier remains the energy cost of artificial lighting. Unlike other crops (lettuces, herbs, microgreens) with modest light requirements, strawberry plants demand high illumination levels to perform well in terms of yield and organoleptic quality. Trials conducted in Belgium with nocturnal LED lighting on everbearing varieties show interesting results for boosting photosynthesis and autumn yields under glass — but costs remain prohibitive at scale. For small and medium-sized French growing businesses, vertical farming for strawberries is not a near-term option in a standard profitability calculation.
Reference yields in conventional soilless growing (gutter systems under tunnel, without tiering) sit between 35 and 60 t/ha for well-managed everbearing varieties, equivalent to 3.5 to 6 kg/m². Vertical growing systems have not yet demonstrated a meaningful improvement in this ratio when energy costs are factored into the equation.
What agronomic AI brings to a controlled growing environment
The value of agronomic decision AI is not limited to open-field or tunnel crops. In a controlled environment, it takes on an additional dimension: closed-loop management, where every measured parameter continuously feeds an optimisation model that adjusts setpoints in real time.
Startup Koidra documented a 30% yield increase in strawberry greenhouse production by optimising climate regulation through AI — refining light recipes (spectrum, photoperiodism), closed-loop irrigation and nutrition sequences (continuous EC/pH adjustment against growth curves), and thermal and hydric ambient parameters. This type of result illustrates what AI contributes that conventional automation cannot: progressive learning of the parameter combinations that maximise yield and quality in a given context, without the operator having to manually configure each setting.
CTIFL in France is testing misting systems coupled with humidity management algorithms in strawberry greenhouses, with the goal of maintaining ideal growing conditions while reducing powdery mildew risk without excess water. In Belgium, trials with nocturnal LED lighting on everbearing varieties are exploring light spectrum optimisation to boost photosynthesis and autumn greenhouse yields. Both approaches converge on the same vision: a greenhouse where all ambient parameters are managed autonomously and adaptively, with humans as supervisors and decision-makers for exception cases.
The 5–10 year horizon for mid-sized French growing operations is not the urban vertical farm, but the smart greenhouse: a soilless system under tunnel, managed by a connected sensor network, with a decision-support DSS capable of interpreting continuous data streams and recommending real-time adjustments. In this model, the crop manager's role evolves — not towards machine dependency, but towards increased expertise in reading agronomic data and adjudicating situations that algorithms cannot yet resolve independently.
What digital tools cannot decide for you
DSS tools, sensors, and epidemiological models generate signals. They do not make decisions. This distinction is fundamental, and it is frequently underestimated when these technologies are presented as turnkey solutions.
Here are three real-world situations that tools cannot resolve without you:
Situation 1 — Botrytis alert. Your model goes to orange alert on Thursday evening. You applied a biocontrol product five days ago, your tunnel was ventilated throughout the morning, and your plants are of a variety with partial grey mould tolerance. Your model does not know your last treatment date, the actual state of your ventilation, or the susceptibility of your cultivar. You do. The decision of whether to intervene tonight rests on this combination of information — not on the model alone.
Situation 2 — Irrigation decision. Your substrate sensor reads 22% VWC this morning. You are at the end of flush 2, your plants have a maximum fruit load, and the forecast shows three overcast days with a drop of 8°C. Do you reduce irrigation frequency to avoid excess water under low evapotranspiration conditions? Or do you maintain it to support the last fruits sizing up? Generic thresholds do not make this distinction. Your agronomic reasoning, combined with your knowledge of the variety in the ground, does.
Situation 3 — Investment decision. You are evaluating the installation of a network of 8 capacitive probes on your 1.8 ha tunnel plot. The supplier presents generic results showing 15% water savings. Were those results obtained on your substrate type, your variety, your climate and soil conditions? Any ROI projection for a technology investment depends on the specific conditions of your operation — not on averages from technical literature.
Field variability is irreducible. It comes down to your soil and climate, your variety choice, your phenological stage at the moment of decision, your plot history, and your current constraints. No digital tool, however powerful, can aggregate these parameters without you providing them. This is precisely what Fraisibot enables: personalised advice on your strawberry crop, reasoning from your actual situation, not a standard profile.
Digital strawberry growing — the tool measures, the professional decides
The technological transformation of strawberry growing is under way, but it does not follow a linear path. Harvesting robotics structurally reduces labour dependency in standardised soilless operations — but it is only economically viable today for highly specialised businesses. Irrigation DSS and epidemiological models provide a rationalised management baseline — but their generic thresholds must be recontextualised for each specific plot situation. IoT sensor networks continuously produce data — but raw data is not a decision. Indoor vertical growing opens promising perspectives for urban niche markets — but its economic equation remains out of reach for small and medium-sized French growing businesses in the coming decade.
The common thread running through all these innovations is the same: the more data tools generate, the more agronomic interpretation expertise becomes the true limiting factor. An operation that instruments itself without the means to interpret its measurements does not improve its results — it adds complexity to its decision-making. The question is no longer "do I have the measurements?" — it is "who helps me turn them into the right decision for my crop, at this stage, in these conditions?"
That is precisely the answer Agronomia provides. Secure your crop management decisions today with Fraisibot, your specialist AI agronomist for strawberry growing — available 24/7, without appointments or travel, at a monthly subscription cost lower than a single field visit from a traditional advisor. To discover all our crop-specific specialist agents, visit all our AI agronomist agents on the Agronomia platform.
To go deeper into the topics covered in this article, explore our other technical guides in the Strawberry Technical Guides blog: strawberry irrigation management, heat stress and irrigation adjustment, protecting strawberry flowering from spring frost, and choosing the right tunnel system for strawberries.