The agriculture sector plays a strategically significant role in the economy of the European Union (EU), serving Europeans with safe and high-quality food and providing over 30 million jobs. However, the sector is under growing pressure to restructure in response to a decline in smaller farms, an ageing and shrinking agricultural workforce and low levels of professional agricultural training. Artificial intelligence (AI) is emerging as a key enabler to address these challenges. This chapter assesses AI uptake in the agricultural sector in the European Union – with a focus on AI‑powered agricultural robots, predictive analytics and crop, and soil and livestock monitoring. The chapter is based on a literature review and interviews with EU business associations and enterprises between December 2024 and May 2025.
Progress in Implementing the European Union Coordinated Plan on Artificial Intelligence (Volume 2)
Abstract
The agriculture sector plays a strategically significant role in the economy of the European Union (EU), serving Europeans with safe and high-quality food and providing over 30 million jobs. However, the sector is under growing pressure to restructure in response to a decline in smaller farms, an ageing and shrinking agricultural workforce, and low levels of professional agricultural training.
Artificial intelligence (AI) is emerging as a key enabler to address these challenges. In response to the shrinking labour force, for example, AI can reduce manual labour requirements by taking on repetitive tasks and allow farmers to shift focus and time to higher-value tasks. By increasing yields and reducing input costs, it enables data-driven decision-making processes throughout the agricultural lifecycle. It also boosts predictive analytics and monitoring tools by aiding farmers to optimise pesticide and herbicide usage. This, in turn, improves soil health and reduces environmental impact by enabling targeted interventions in the field. Finally, tailored AI solutions can tackle low farm literacy and lack of formal education by helping smallholders overcome resource and knowledge constraints.
This chapter provides an overview of how AI is being used across the EU agricultural sector and identifies what is needed to scale these technologies more widely and responsibly. It begins with a broad mapping of application areas in which AI is already being deployed or holds strong potential across the agricultural lifecycle – from cultivation and monitoring to harvesting. It has two parts:
Overview of the EU agricultural sector and the strategic role of AI: a profile of the agriculture sector analyses its structural and economic make-up. It examines the role of AI as a key enabler of precision agriculture, which uses digital techniques to assist in decision making. It explores the role that AI techniques play within the agricultural lifecycle, painting a comprehensive picture of AI uses. It then dives into the state of play of AI adoption in the European Union along with key EU policies that support its adoption in agriculture. This section ends with the main barriers that hinder faster and wider AI uptake in the agriculture sector.
Spotlight on selected AI use-cases in agriculture: the chapter provides an in-depth examination of three use-cases in the agricultural sector: i) AI-powered agricultural robots; ii) predictive analytics; and iii) crop, soil and livestock monitoring. Each section begins with a synthesis of insights from academic literature, followed by findings drawn from stakeholder interviews. For each use-case, the analysis identifies key barriers and challenges, as well as targeted policy recommendations to support the effective and scalable adoption of AI into the sector.
Methodological considerations are discussed in Chapter 1. A summary of the key recommendations is provided next.
Data availability and access
Invest in open, high-quality datasets: support public collection and dissemination of soil, weather and crop performance data to lower entry barriers and spur innovation.
Safeguard farmers’ control over agricultural data: provide sectoral-specific guidance on data sharing through advancement of responsible data-sharing frameworks.
Increase awareness of and stakeholder engagement in the Common European Agricultural Data Space (CEADS): promote and disseminate open agricultural data spaces to enhance accessibility of high-quality, field-level data to farmers and start-ups.
Promote standards to reduce fragmentation: encourage use of open data formats, application programming interfaces and protocols across platforms, equipment and systems.
Infrastructure and connectivity
Expand digital infrastructure: improve broadband connectivity, cloud access and edge- computing capacity to support real-time AI analytics, particularly in underserved rural areas.
Regulatory and policy frameworks
Adopt a comprehensive EU strategy on agricultural digitalisation: integrate funding, regulation, infrastructure and skills development.
Clarify regulatory requirements for the sector: provide specific guidance to facilitate compliance, particularly for start-ups, and small and medium-sized enterprises (SMEs).
Provide guidance on the interplay and application of the EU AI Act and Machinery Regulation: clarify how AI regulations apply to agricultural machinery.
Skills, trust and collaboration
Make AI accessible through user-centred design: develop intuitive interfaces and local language options, especially for older or less tech-savvy farmers.
Share best practices and success stories: leverage multistakeholder platforms and farmers’ associations to strengthen farm advisory services and co‑operatives.
Invest in digital skills and training: deliver hands-on capacity-building, including workshops, demonstrations and peer learning among farmers. Support “farmer ambassadors” to lead by example.
Provide grants for start-ups and SMEs: develop robotics solutions tailored to European small farms and specialty crop farms.
Prioritise development and adoption of standards: facilitate data sharing and prevent monopolisation by large equipment manufacturers.
This section outlines the structural and economic make-up of the agricultural sector, presenting information on adoption of AI. It examines the role of AI as an enabler of precision agriculture, which uses digital technologies to assist in decision making. It explores the role that AI techniques play within the agricultural lifecycle, with examples of AI uses. It then dives into the state of play of AI adoption in the European Union along with EU policies that support its adoption in agriculture. The section ends with the main barriers that hinder faster and wider AI uptake in the agricultural sector.
Key characteristics of the agricultural sector
Sectoral make-up
The agricultural sector represents a modest but strategically significant component of the European Union (EU) economy, contributing to 1.3% of EU gross domestic product (GDP) in 2024. The total value of output of the EU agriculture industry reached EUR 532.4 billion in 2024, with France contributing EUR 89.4 billion, followed by Germany (EUR 75.5 billion), Italy (EUR 75.4 billion), Spain (EUR 67.5 billion), the Netherlands (EUR 41.2 billion), Poland (EUR 37.8 billion) and Romania (EUR 20.5 billion) (Eurostat, 2025[1]). These seven EU Member States accounted for approximately three-quarters of total EU agricultural output (Figure 2.1).
Source: Eurostat (2025[2]), “Economic accounts for agriculture – values at current prices”, https://ec.europa.eu/eurostat/databrowser/view/aact_eaa01/default/bar?lang=en.
Serving 450 million Europeans, the EU agri-food trade system plays an important role in the European economy. It generated and added a value of EUR 900 billion in 2022, employing 30 million people (European Commission, 2025[3]). The European Union accounts for 47% of agriculture and fish production of the Europe and Central Asia region. It stands as one of the world’s biggest exporters of agri-food products, exporting a record EUR 235.4 billion in 2024, with cumulative export levels steadily increasing each year by 3% (OECD/FAO, 2024[4]; European Commission, 2025[5]). The United Kingdom and the United States remain the largest export partners, with agri-food trade maintaining a positive balance at EUR 63.6 billion in 2024 (European Commission, 2025[6]). The European Union is also the world’s largest producer of agricultural machinery, with an annual output of some EUR 40 billion, and the leading exporter by value. Notably, EU exports of agricultural machinery to the United States grew steadily between 2019 and 2023, reaching EUR 4.7 billion in 2023. Ag-tech machinery makes up more than half of this export value with the remainder generated by tractors (CEMA, 2025[7]).
Among various economic activities, output from agriculture, forestry and fishing reported the smallest increase in 2021 and was the only economic activity to decrease in 2022 (Figure 2.2).
2005 = 100
The People’s Republic of China (hereafter “China”), India and other Asian countries are expected to drive global agricultural production, accounting for more than half of global crop output (OECD/FAO, 2024[4]). In 2022 and 2023, the European Union recorded a production value slightly lower than that of the United States, while surpassing that of Brazil (Figure 2.3). In 2023, wheat and maize were the top export and import commodities by quantity in the European region (FAOSTAT, 2023[9]). The Europe and Central Asia region is projected to experience stagnant growth, with production levels mostly driven by Eastern Europe and Central Asia. Although production growth in North America is expected to decrease, overall sectoral growth will be supported by productivity gains. The sub-Saharan Africa and Near East and North Africa region are projected to see rapid growth in livestock production, while crop production will be the main driver of growth in Latin America and the Caribbean.
The EU agricultural output is distributed across three main categories: crop production (50.3% of total output, EUR 281.9 billion); animal production and products (41.1%, EUR 218.6 billion); and a third category comprising inseparable non-agricultural secondary activities, such as processing of agricultural products and agricultural services (totalling 8.5%, EUR 45.2 billion). Crop production encompasses vegetables, horticultural plants and cereals, while dairy and pig farming dominate animal production (Figure 2.4).
Source: Based on FAOSTAT (2023[10]), FAOSTAT Value of Agricultural Production (dataset), https://www.fao.org/faostat/en/#data/QV.
Percentage of total EU output, 2024
Source: Eurostat (2025[1]), “Statistics explained: Performance of the agricultural sector”, https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Performance_of_the_agricultural_sector.
The EU agricultural sector has undergone significant re-structuring. Between 2005-2020, the number of farms declined by about 37%, a reduction of 4.6 million farms (Eurostat, 2023[11]). As of 2020, the European Union had approximately 9.1 million agricultural holdings, with pronounced concentration in certain Member States. Romania hosted 31.8% of total EU agricultural holdings (2.9 million farms), followed by Poland (14.4%, 1.3 million farms), Italy (12.5%, 1.1 million farms) and Spain (10.1%, 0.9 million farms) (Eurostat, 2022[12]).
Small-scale farming remains predominant across the region, with approximately two-thirds of all agricultural holdings operating on less than 5 ha. This characteristic is particularly evident in Romania (where more than 90% of farms are smaller than 5 ha), Malta, Cyprus, Greece, Portugal, Croatia, Hungary, Bulgaria and specific regions of Poland, Spain and Italy. These smaller holdings typically focus on specialty crops such as olive groves and vineyards, often operating within geological and topographical constraints (Eurostat, 2022[12]). Around one-tenth of farms (11.4%) had at least 30 ha (Eurostat, 2023[13]). While farms over 100 ha accounted for only 3.6% of the total number, these large farms used more than half (51.8%) of the total land share available for agricultural production in the European Union (Eurostat, 2023[13]).
The share of land used for agriculture in the European Union is less than in several of its trading partners. In 2020, France and Spain had 27.4 million ha and 23.9 million ha of used agricultural areas, the two largest land-use share of the 157 million ha for agricultural production in the European Union (Eurostat, 2023[13]). This falls short of other non-EU countries with a strong agricultural sector, such as China, the United States, Brazil or India, with agricultural land of 520.7 million ha, 405.8 million ha, 239.4 million ha and 178.5 million ha, respectively (World Bank, 2021[14]). Ukraine, for instance, has 41.3 million ha of agricultural land, covering 68.5% of its total area and ranking among the world’s most fertile regions (State Statistics Service of Ukraine, 2023[15]). Most crop production is handled by medium-size enterprises, while a small number of large agribusiness companies – rather than individual farmers – operate land holdings as large as 500 000 ha (Román, 2024[16]).
The 9.1 million EU farms have diverse specialities. Most farms specialise in crops (58.3%), followed by livestock (21.6%) with the reminder a mix of the two (19.3%) (Eurostat, 2023[11]).1 Specialised crop farming is more prevalent in Eastern European countries (Bulgaria, Hungary, Romania) and Mediterranean regions (Greece, Malta, Croatia). Conversely, livestock specialisation characterises northwestern European agriculture (Luxembourg, Ireland, the Netherlands) (Eurostat, 2023[11]).
Not only do farm types vary, but EU farms remain diverse in what they cultivate (Eurostat, 2022[12]) (Figure 2.5). Among crop specialists, root crops (e.g. potatoes, sugar beet) in field vegetables and field crops accounted for 18.3% of the total. These were followed by cereals, oilseeds, protein crops, olives, fruits, vineyards and horticulture. Among livestock specialists, dairy farms were the most numerous, taking up 5.1% from the total share of EU farms, followed by cattle rearing and fattening. This diversity in farm specialisation underscores the challenge of one-size-fits-all policy solutions for the EU agricultural sector, as technological solutions and farmer needs vary substantially across farm types and specialisations.
The EU agricultural workforce has experienced a sustained decline, with an average annual reduction of 2.6% between 2008 and 2023. This downward trajectory continued in 2023, with the sector employing 7.6 million full-time equivalent workers – a substantial decrease from 8.7 million in 2020 (Eurostat, 2022[17]; Eurostat, 2025[1]). This contraction in agricultural labour has been particularly pronounced in Denmark, Finland, Latvia and Bulgaria, with only Romania, Cyprus and Malta exhibiting increases in labour volume. The lack of attractiveness of the sector to potential workers combined with profit volatility are the primary drivers of this labour outflow (European Commission, 2024[18]), particularly to the younger generations (Campi et al., 2024[19]). Agricultural employment was projected to decrease by 28% between 2017 and 2030 (European Commission, 2017[20]). As of 2020, young farmers under the age of 35 represented only 6.5% of the agricultural workforce (European Commission, 2024[18]). Younger farmers typically manage medium to large-sized farms compared to farmers aged 65 years or older (Eurostat, 2022[17]).
Percentage of EU farms, 2020
Source: Eurostat (2022[12]), “Statistics explained: Farms and farmland in the European Union”, https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Farms_and_farmland_in_the_European_Union_-_statistics&oldid=656883.
The age distribution of farm managers varies significantly across Member States. Austria (23.4% of all farm managers) and Poland (21%) have relatively high proportions of farmers under 40. Conversely, Cyprus (5.1%), Portugal (6.4%), Greece (7.2%) and Spain (7.7%) have significantly lower proportions and correspondingly higher percentages of farmers aged 65 and above.
Professional agricultural training among farmers is limited to specific Member States. The Netherlands, Luxembourg, France and Czechia report a high proportion of farm managers with full agricultural training.2 Most farmers, around 70%, gained skills through practical experience alone. Romania and Greece had an overwhelming majority of farm managers with practical experience, with only 0.7% receiving full agricultural training (Eurostat, 2022[17]).
The EU agricultural sector faces pressures from climate change and structural shifts (European Commission, 2024[18]). Exacerbating climate conditions are projected to reduce production volumes and quality levels for various crops, including apple, peaches, nectarine and tomatoes, while milk productivity is expected to increase but at a slower pace (European Commission, 2024[18]). Structural shifts in land use, ongoing biodiversity loss and persistent greenhouse gas emissions compound these challenges. Water scarcity, land degradation and extreme weather events are becoming more frequent, posing additional risks to farm viability and rural livelihoods (OECD, 2023[21]).
Innovation is central to building a resilient, productive and sustainable agricultural sector in the European Union (OECD, 2023[22]). Technological advances such as precision agriculture, smart irrigation systems, climate-resilient crop varieties and digital farm management platforms are helping to increase input efficiency, reduce environmental impacts and enhance profitability. These solutions not only support environmental sustainability but also address labour shortages by automating time-consuming and repetitive tasks. Furthermore, farmer-led innovation, when coupled with robust advisory services and peer-learning networks, can significantly improve adoption of new technologies and ensure that solutions are adapted to local conditions and challenges (OECD, 2023[23]). Among these innovations, AI is emerging as a transformative force in precision agriculture. The following sections explore the specific applications and potential of AI in driving sustainable productivity in the EU farming sector.
AI as a key enabler of precision agriculture
AI could transform the agricultural sector by increasing productivity, improving sustainability and enabling more efficient use of resources (Rejeb et al., 2022[24]). AI techniques such as deep learning (DL), machine learning (ML) and artificial neural networks hold great potential in enhancing farming practices and contributing to the sustainability of the sector through data-driven approaches (Rejeb et al., 2022[24]). For instance, computer vision solutions and AI algorithms can bring forth benefits such as task automation along with enhanced food quality and safety (Patricio and Rieder, 2018[25]). In particular, AI systems can analyse large volumes of data collected through Internet of Things (IoT) devices and sensors, supporting more accurate diagnostics for soil fertility, irrigation needs, pest control and crop disease (Lin et al., 2019[26]). These systems contribute to better resource management and facilitate real-time responses to environmental and agronomic variables. Moreover, AI enables automation of labour-intensive tasks via robotics, helping to optimise resource efficiency and generate higher agricultural output (Rejeb et al., 2022[24]).
Likewise, AI is being integrated into traditional agricultural methods and IoT technologies to support monitoring and optimisation of crop growth using the principles of precision farming3,4 (Elbasi and Mostafa, 2023[27]). Precision agriculture applies data-driven techniques across farming, livestock, aquaculture and agroforestry, encompassing both automation and digitalisation (Ceccarelli et al., 2022[28]). AI plays a key role in precision agriculture, enabling more insights from large quantities of data and informing decisions related to field crops and animal production (EPRS, 2023[29]).
Trend topic analysis indicates a shift from earlier AI applications. From its earlier focus on AI-powered robots, applications are moving towards a wider range of AI techniques, including Big Data, IoT, convolutional neural networks (CNNs), DL and ML. Among these, the latter three stand as the most common approaches in accelerating the transition to precision agriculture (Rejeb et al., 2022[24]). Meanwhile, robots, IoT, wireless sensor networks and unmanned aerial vehicles (UAVs), or drones, are prevalent technologies powered by AI to modernise farming practices (Rejeb et al., 2022[24]).
An AI pipeline orchestrates the flow of data into an algorithm, which enables the training of the model, and generates predictions and decisions (Zhou and Chen, 2023[30]). Data remain at the core of every step in the AI pipeline. Typical data gathered in the AI pipeline in agriculture include information on weather information, soil, seed, irrigation, fertiliser use and historical yield, as well as held management data, satellite images, and data collated from robots and drones (Figure 2.6). These data are fed into the AI algorithm to train the model and make predictions (Figure 2.7) that can then empower farmers in making optimal decisions throughout the agricultural cycle.
While AI-based predictions can help farmers make more informed decisions, the increasing complexity of AI models do present challenges. As several studies have noted (Wadoux et al., 2020[31]; McBride, 2022[32]), current ML methods often struggle with causal inference as opposed to identifying patterns or correlations in data. The increasing complexity of AI models may pose difficulty for farmers to interpret the rationale behind the predictions or to trust their recommendations.
Source: Zhou and Chen (2023[30]), Artificial Intelligence in Agriculture, https://doi.org/10.1007/978-3-030-89123-7_183-2.
Source: Zhou and Chen (2023[30]), Artificial Intelligence in Agriculture, https://doi.org/10.1007/978-3-030-89123-7_183-2.
Agricultural cycle and AI techniques
The agricultural cycle can be divided into three phases and substages: cultivation (preparation of soil, sowing of seeds), monitoring (adding fertilisers, irrigation, weed protection) and harvesting (harvesting, storage) (Figure 2.8).
Source: Zhou and Chen (2023[30]), Artificial Intelligence in Agriculture, https://doi.org/10.1007/978-3-030-89123-7_183-2.
With its capability in prediction, automation and navigation, AI is emerging as a key enabler across all three stages of the agricultural lifecycle (Wakchaure, Patle and Mahindrakar, 2023[33]). During the cultivation phase, AI supports land preparation, irrigation planning and seed sowing. In the monitoring phase, AI is used to collect data, identify disease, control weeds and apply pesticides. In the harvesting phase, AI applications include picking, cutting and storage. For instance, AI-powered robotic harvesters use computer vision to identify and selectively pick ripe fruits. For post-harvest handling, AI systems optimise storage by monitoring temperature and humidity, and by directing automated sorting and stacking based on product quality and shelf life. AI contributes most significantly to monitoring compared to its use in harvesting (Wakchaure, Patle and Mahindrakar, 2023[33]). AI is applied to solving the path-planning problem, or navigation, of agricultural robots in the field, in addition to its role in core farming operations (Wakchaure, Patle and Mahindrakar, 2023[33]).
AI use-cases
AI supports the automation and digitalisation of agriculture, transforming business models and practices across the agricultural value chain (Ceccarelli et al., 2022[28]). It addresses knowledge needs of farmers and adds to their ability to identify diseases, monitor irrigation, reduce human labour and maximise yield production. AI adoption increases efficiency and productivity in agricultural labour to better meet food demand, process larger amounts of agricultural data for actionable insights, and improve sustainability and reduce environmental impact (Rejeb et al., 2022[24]).
AI can contribute to increased crop yields, pest control, soil monitoring, workload management and optimisation of sowing and harvesting times, while improving agricultural tasks throughout the food supply chain (Zhou and Chen, 2023[30]). Zhou and Chen (2023[30]) identify eight applications of AI in agriculture (Figure 2.9).
Source: Zhou and Chen (2023[30]), Artificial Intelligence in Agriculture, https://doi.org/10.1007/978-3-030-89123-7_183-2.
Crop and soil monitoring
Use of AI can introduce new ways to map key soil and chemical properties to ensure agriculture remains sustainable. With use of chemicals and fertilisers on the rise, maintaining crop and soil health is vital for sustainable agriculture (Zhou and Chen, 2023[30]). Currently, 60-70% of European soils are classified as unhealthy, threatening food security, environmental resilience and economic sustainability for the agricultural sector (Joint Research Centre, 2023[34]; Chowdhury et al., 2024[35]; Forrester, 2025[36]). While the mapping of the physical and chemical properties of soil – such as texture, structure, pH, nutrient levels and moisture content – has been well established for decades, AI introduces new capabilities to this domain. By processing large volumes of data from sources like satellites, drones, humidity sensors and weather stations, AI-driven remote sensing and DL algorithms enhance the accuracy of soil maps. At the same time, they account for the increased size, abundance and complexity of soil datasets (Tian et al., 2019[37]; Gill, 2025[38]; Wadoux, 2025[39]). They can also reduce the need for labour-intensive field sampling.
The European Union is actively investing in several AI initiatives for soil and crop monitoring, with the AI4SoilHealth project standing as a flagship example (AI4SoilHealth, 2025[40]). Launched in late 2022 with nearly EUR 2 million from the Digital Europe Programme, this project is developing an open-access, AI‑driven digital infrastructure that continuously assesses and monitors soil health metrics across Europe (European Commission, 2022[41]).
Crop disease diagnosis
Crop diseases can cause up to 60% of crop yield loss, with approximately USD 220 billion in economic losses annually (European Commission, 2018[42]; Ristaino et al., 2021[43]; Hossain et al., 2024[44]). As plant pests can significantly affect the quality and quantity of products, early identification of plant symptoms is crucial to mitigating their impacts. AI-driven disease diagnosis combines computer vision, sensor networks and disease prediction ML models to detect pathogens early by analysing plant discoloration, soil anomalies and microclimate patterns (Amos, 2024[45]). For instance, predictive models can forecast disease outbreaks based on environmental conditions, helping farmers to take proactive measures (Bivocom, 2024[46]). AI-powered image recognition can also analyse images of crops taken by drones or mobile devices to detect early signs of plant diseases.
Crop yield prediction
Due to climate extremes, accurate estimates of crop yield are becoming essential for planning and resource allocation in agriculture. Crop yield prediction – the practice of forecasting the amount of crop to be produced in a land area – provides food security and efficient resource optimisation such as water, fertilisers and machinery. It thus contributes to sustainable farming practices (Stelar, 2024[47]). With data from drones, remote sensing images, soil nitrogen levels, soil moisture and weather conditions, AI models can analyse diverse datasets to train prediction models to estimate yield and forecast crop output (Klompenburg, Kassahun and Catal, 2020[48]). These predictions can help farmers and agribusinesses to make better decisions regarding contracts, pricing, logistics and supply management, especially in the pre‑harvest stage (Zhou and Chen, 2023[30]). Automated counting of fruits or flowers using computer vision further streamlines yield estimation, saving time and reducing labour costs (Zhou and Chen, 2023[30]).
Predictive insights
The European Union is advancing AI yield prediction through projects like YIPEEO (Global Change Research Institute, 2021[49]). YIPEEO leverages ML to enhance field-scale crop yield forecasts across Europe, using Copernicus Sentinel data and climate models (ESA, 2019[50]). One study developed an olive grove yield predictive tool using ML and satellite imagery, proving it can accurately predict olive fruit and oil yields in Spain eight months before harvest (Ramos et al., 2025[51]).
AI could also provide large benefits to farmers with predictive insights for other stages in the agricultural cycle, including soil preparation, sowing of seeds, harvesting and storage. For instance, ML models can be deployed to predict optimal sowing dates of crops using data such as weather information, historical planting dates and annual crop yields (Zhou and Chen, 2023[30]). Inaccurate sowing dates can result in financial and labour losses, and ultimately, lower productivity rates.
AI can also manage weather data, soil information, crop management and historical harvest time to predict harvest time of crops (Boechel et al., 2022[52]). Since weather conditions such as wind and humidity can affect crop growth, it is essential to determine optimal spraying time for fertiliser or chemicals to maximise their effects (Zhou and Chen, 2023[30]). The EU-funded project AgriBIT (AgriBIT, 2020[53]) validated AI technologies for near real-time detection of pest risks and bacterial infestations in crops. It focused on fertiliser and pesticide reductions and optimising water resources in pilot farms across Greece, Portugal and Italy (European Commission, 2024[54]). For its part, Farmonaut (2025[55]), an AI-powered agricultural platform in Europe, reports that its precision spraying recommendations have enabled farmers to leverage AI-driven pest detection, crop health monitoring and real-time weather integration. In so doing, they have reduced pesticide use by up to 30% without compromising yields.
Agricultural robots
Agricultural robots are autonomous or semi-autonomous intelligent machines designed to perform tasks in the agricultural and farming space. They are taking over various jobs such as planting, weeding and harvesting in the agricultural production and management process (Kugler, 2022[56]). These machines use sensor technologies, computer vision and AI models to identify location of produce and perform tasks. Smart robots equipped with computer vision and autonomous decision-making capabilities can augment human labour by executing tasks with higher precision, accuracy and efficiency (Critchley, 2025[57]). The Vine Robot, for example, navigates vineyards, collecting data on grape composition and field conditions to monitor grape growth (Pathan et al., 2020[58]; Vine Robots, 2025[59]). The European Commission invested about EUR 2 million in this project (CORDIS, 2022[60]). Robs4Crops, a Horizon 2020 project, is collaborating with commercial farms across France, Greece, Spain and the Netherlands to accelerate large-scale implementation of robotics and automation in European agriculture (Robs4Crops, n.d.[61])
Intelligent spraying
Traditional pest and weed control methods often involve indiscriminate chemical spraying, which can be costly and environmentally damaging (Zhou and Chen, 2023[30]). As outlined in its “Farm to Fork” strategy, the European Union plans to halve use of pesticides by 2030 (European Commission, 2025[62]). AI applications allow for intelligent spraying for pesticides and herbicides by processing data from various sensors that monitor crop and environmental conditions. The collected data are used to train ML models to identify problem areas, weeds, pests and diseases (Ngo et al., 2019[63]). Results are fed into AI-driven autonomous precision spraying equipment to apply pesticides and herbicides only where needed, reducing chemical usage, lowering production costs and minimising environmental impact (Zhou and Chen, 2023[30]). The EU-funded Life Smart Sprayer project aims to reduce herbicides by 40% across six regions covering 10 200 ha in France, Germany, Hungary and Romania (European Commission, 2025[62]).
Irrigation optimisation
AI-driven irrigation optimisation systems are helping address the crucial need for efficient water management need by automating water delivery (Zhou and Chen, 2023[30]). These smart systems use remote sensors Arduino5 and Raspberry pi36 technology to monitor soil moisture, temperature, nutrient levels and weather forecasts, enabling them to schedule irrigation only when necessary (Zhou and Chen, 2023[30]). AI models can process these data to predict evapotranspiration rates and make optimal irrigation decisions, reducing water waste and ensuring crops receive the right amount of hydration for healthy growth (Talaviya et al., 2020[64]; Abioye et al., 2022[65]).
This technology is already generating tangible results. The “Green dEal cOmpliant iRriGation Increasing Europe's Agriculture resilience to drought” (GEORGIA) project is developing an explainable AI decision support systems for irrigation management across six European countries – Greece, Cyprus, Bulgaria, Serbia, Austria and Poland (Synelixis, 2025[66]). With more than 1 200 farmers involved, the project aims to increase agricultural resilience to drought through an AI-powered water management system (European Commission, 2024[67]). In Italy, Irreo provides a dynamic irrigation system that combines satellite data, AI and automation to increase crop yield through improved water efficiency (ESA, 2025[68]). Meanwhile, using AI, satellite imagery and IoT devices, Agrow Analytics in Spain has developed a precision irrigation system that provides customised irrigation strategies and configuration based on predicted water consumption and rainfall (Agrow, 2025[69]).
Digital twin
Digital twins are virtual models that replicate real-world crops and farming conditions, allowing farmers to simulate and predict plant responses to various scenarios. AI and ML play a vital role in driving these simulations, enabling producers to test different weather, irrigation and fertilisation strategies in a virtual environment before applying them in the field. This approach helps optimise crop management, improve yields and anticipate challenges, making it possible to tailor practices to specific environments and future-proof agricultural production (Zhou and Chen, 2023[30]).
Generative AI, particularly large language models (LLMs), is starting to transform farm advisory services and agricultural education. Generative AI enables a more scalable, personalised and context-aware support for farmers and advisers. These models can process and distil vast, unstructured datasets, making expert guidance more accessible and actionable for diverse user groups – from frontline advisers to smallholder farmers (McKinsey and Company, 2024[70]). They can also bring significant value by boosting the research and development (R&D) lifecycle – from assistance in research and delivery to development and product launch.
Through customised offers and low-cost services, LLMs plugged into agronomic advisory services can benefit farmers directly by generating price recommendations based on customer history. As an example, Digital Green, a non-profit organisation, has developed Farmer Chat, an AI-powered assistant. It leverages LLMs to shift advisory services from supply-driven to demand-driven models, delivering localised, multimodal and inclusive support (Digital Green, 2023[71]).
Generative AI may also help farmers streamline data collection and automate the preparation of sustainability reports, supporting compliance with legislation. These systems can ingest diverse data sources, including sensor readings, satellite imagery and soil analyses. In so doing, they can generate standardised reports with minimal manual effort, saving time and reducing the administrative burden on farmers, while increasing regulatory compliance (AXIOBIT, 2025[72]).
Table 2.1 presents an overview of AI use-cases in agriculture in each key agricultural phase, describing their functionality, learning methodologies and associated challenges, and aligning them with the OECD Framework for the Classification of AI systems (OECD, 2022[73]).
State of play of AI adoption in the EU agricultural sector
There is limited availability of empirical data and harmonised surveys on digitalisation of agriculture, particularly regarding their impact on farmers’ livelihoods (McFadden et al., 2022[74]). These gaps stem from the significant heterogeneity in EU farming practices, geographical conditions and the types of technologies used across Member States (Garske, Bau and Ekardt, 2021[75]; Campi et al., 2024[19]). Available evidence indicates the transition from large-scale mechanisation to AI-driven technologies is decreasing the need for labour in the European context. The current stage of AI automation will not fully replace agricultural labour. However, these AI-driven technologies are catalysing the need for a specialist workforce that can maintain robots and apply decisions based on predictive insights (Munniunker, Nel and Diederichs, 2022[76]).
In particular, there is a notable lack of research on the adoption and practical use of AI in agriculture (Wakchaure, Patle and Mahindrakar, 2023[33]). Within the EU context, literature on AI applications in agriculture remains scarce. This limits the ability to identify representative case studies or assess the broader impact of AI uptake in the sector.
Available studies point to increased adoption of automation tools (e.g. precision farming, automated feeders, drop irrigation systems, mechanised harvesting) in recent years (European Commission, 2024[18]). The digitalisation of agriculture in the European Union combines wireless technologies, IoT, AI and blockchain under the management approach of precision agriculture or “precision farming” (Kondratieva, 2021[77]).
Precision farming borrows from intelligent control systems equipped with enablers such as digital devices or software tools embodied in agricultural machinery. This makes it difficult to isolate a particular technology that provides a service primarily reliant on AI (Sparrow, Howard and Degeling, 2021[78]). However, the types of precision farming applications make it difficult to fully disentangle the contribution of AI from that of other digital technologies applied in agriculture (Sparrow, Howard and Degeling, 2021[78]).
McFadden et al. (2024[79]) present national adoption estimates of precision agriculture technologies in some EU countries. Denmark reports that 40% of its farms adopted precision agriculture technologies in 2023, a slight increase from 37% in 2022. Hungary reports 12% of farms used some form of these technologies in 2020. Hungary also reportedly saw an increase in guided/automatic steering (4% in 2020 to 5.3% in 2023) and robots (0.7% to 1.7%), while precision agriculture use decreased in plant and environmental sensors, as well as yield mapping (McFadden et al., 2024[79]). While not specific to AI applications, the 4Growth project, co‑ordinated by Wageningen University, aims to document the uptake of digital and data technologies in agriculture. It is collecting collect data across seven European observatories (4Growth, 2025[80]).
In the absence of comprehensive data on AI adoption, emerging activity in the EU start-up ecosystem offers valuable insights into the application of AI on the ground in agriculture. EU start-ups are emerging as key drivers of innovation in agricultural digitalisation, providing practical, AI-powered solutions tailored to the needs of farmers:
In the Netherlands, Source.ag leverages AI to optimise indoor farming and greenhouse management, helping growers maximise fruit and vegetable yields (Source.ag, 2025[81]). Meanwhile, Overstory B.V. uses ML to interpret satellite imagery to manage vegetation and reduce fire risk (Overstory, 2025[82]).
In France, Naïo Technologies SAS has developed autonomous field robots that assist with weeding, harvesting and material transport, supporting labour efficiency and sustainability on farms (Naïo Technologies, 2025[83]). For its part, Augmenta S.A. developed AI and precision hardware to automate real-time variable rate applications of nitrogen, plant growth regulators and harvest aids. At the same time, it offers advanced analytics through its “software as a service” mode. The company was acquired in 2023 by CNH Industrial (Farm Connexion, 2023[84]).
In Denmark, FarmDroid ApS designs autonomous robots for seeding and weeding, and reducing input costs and environmental impact (FarmDroid, 2025[85]).
Despite these examples, access to funding remains a significant barrier. Venture capital (VC) investments in AI start-ups focused on agriculture in the European Union reached only USD 57 million in 2024, down from a peak of USD 62 million in 2023. These amounts accounted for just 0.5-0.6% of the total VC investments in AI start-ups across the European Union during the years of 2023-2024, highlighting the relatively limited funding directed towards agricultural AI innovation (Figure 2.10).
Source: OECD.AI (2025[86]), Total VC Investments in AI by Country and Industry, calculations based on data from Preqin, last updated 2025-10-01, https://oecd.ai/.
Key EU policies supporting AI adoption in agriculture
The European Union employs a strategic, multi-layered approach to accelerate adoption of innovative digital technologies in the agri-food sector. This approach combines research, innovation and deployment initiatives to ensure that new solutions reach both the market and end-users effectively (European Commission, 2024[87]). Through its “Vision for Agriculture and Food” roadmap, the European Commission sets for a competitive and resilient EU agri-food system for farmers and operators. The roadmap places the agriculture sector at the centre stage of the European economy to support the shift to digital-ready farming and foster a secure agri-food supply chain (European Commission, 2025[88]).
The EU Common Agricultural Policy (CAP) has been fundamental to shaping European agriculture, providing income support to farmers, putting in place market measures and addressing challenges in rural areas (European Commission, 2023[89]). As of the 2021-2027 Multiannual Financial Framework, the CAP represents approximately one-third of the total EU budget (European Commission, 2025[90]). CAP 2023‑2027, effective from 1 January 2023, focuses on supporting farmers to ensure the level of food production and to maintain the status of the European Union as a leading producer and net exporter of agri-food products (European Commission, 2023[89]). As a key objective, the CAP fosters the digitalisation of agriculture through national digitalisation strategies (European Commission, 2025[91]). As part of their CAP Strategic Plans, Member States explore tailored digital solutions that fit country needs. These could include investment support for broadband in rural areas; eco-programmes to support precision-farming technologies; knowledge exchange through digital skills training support or demonstration farms; and farm advisory services, including the Farm Sustainability Tool for Nutrients7 (FaST) (European Commission, 2025[91]).
The CAP is complemented by several other EU-funded programmes that support the digitalisation of EU agriculture, including AI-enabled innovation. These are highlighted below.
The EU CAP Network brings together members from the European Network for Rural Development and the European Innovation Partnership for Agricultural Productivity and Sustainability (EIP-AGRI) to support the design and implementation of the CAP Strategic Plans (EIP-AGRI, 2025[92]). It provides opportunities for European peer-to-peer networking and promotes innovation and knowledge exchange through focus groups, seminars and workshops (EU CAP Network, 2025[93]).
Horizon Europe, the EU flagship research and innovation programme, supports development of advanced digital technologies for agriculture, such as smart farming systems, precision agriculture and digital platforms that enhance productivity and sustainability, as did its predecessor, Horizon 2020. Notable projects funded by Horizon Europe and Horizon 2020 include:
Agricultural Interoperability and Analysis System (ATLAS), an open interoperability network for innovative agricultural applications (ATLAS, 2025[94])
DEMETER, which established interoperable IoT-based platforms across 20 pilots in 18 countries (DEMETER, 2025[95])
SmartAgriHubs, which built a network of Digital Innovation Hubs (DIHs) to accelerate digital adoption among farmers (European Commission, 2023[96]).
Agriculture of Data Partnership, which leverages digital and data technologies, especially Earth observation data, to promote sustainable agriculture, improve policy monitoring and support the objectives of the European Green Deal and “Farm to Fork” Strategy (NCP Portal, 2025[97])
4Growth, which aims to document the state of play of digital solutions in agriculture and forestry (European Commission, 2024[98])
The Work Programme 2025-2027 of the Digital Europe Programme (DIGITAL) supports AI adoption in agriculture through various actions. It supports multi-country projects in agri-food that especially foster the access, sharing and re-use of data that reduces administrative burdens in agricultural data sharing. It also supports actions to integrate generative AI into sectoral testing and experimentation facilities in agri-food (European Commission, 2025[99]).
Since 2014, and 2021 onwards, the Connecting Europe Facility (CEF) and the Digital Europe Programme (DEP) fund the eDelivery Building Block Digital Service Infrastructure, bringing digital technology to businesses and public administrations across Member States (European Commission, 2023[100]). The CEF-Digital programme, as a second generation of CEF, aims to boost digital connectivity infrastructures across the European Union, complementing CAP investment in bolstering digital infrastructures in rural regions. As part of its “5G and Edge Cloud for Smart Communities”, CEF-Digital has awarded EUR 135 million to 41 projects. These support deployment of stand-alone 5G networks and integration with edge cloud infrastructures to enable innovation in various sectors, including agriculture (HaDEA, 2025[101]).
European DIHs (EDIHs) serve as one-stop-shops that help companies respond dynamically to digital challenges and become more competitive. In agriculture, EDIHs provide access to technical expertise and experimentation, enabling farmers to test before investing in digital solutions, including AI technologies.
European Testing and Experimentation Facilities (TEFs) for AI in Agri-Food: AgriFoodTEFs provide a network of physical and digital facilities for validating AI and robotics solutions in real-world agricultural settings, helping innovators bridge the gap between development and market deployment (European Commission, 2025[102]).
The Common European Agricultural Dataspace (CEADS) is an EU initiative to create a secure, trusted and decentralised framework for sharing, processing and analysing agricultural data across Europe (European Commission, 2024[103]). Funded by the DEP with nearly EUR 2 million and involving 15 partners representing ten countries, the AgriDataSpace project lays the groundwork for the CEADS to map the landscape of agricultural data-sharing initiatives, analyse governance and business models, assess evolving legislation and develop a comprehensive technical reference architecture (European Commission, 2024[104]).
In addition to the above, the EU Data Act (Regulation 2023/2854), applicable from 12 September 2025, is a horizontal legal framework. In the context of agriculture, it ensures that users of connected devices (including farm machinery) can access and share the data they generate. It mandates fair access conditions, supports interoperability and facilitates cross-sectoral innovation. The Data Act aims at empowering stakeholders, including farmers, to use, control and share operational data with third-party AI service providers (European Commission, 2024[105]). Complementing the Data Act, the Data Governance Act (Regulation 2022/868) establishes trusted mechanisms for data sharing, including the concept of data intermediaries and data altruism frameworks (European Commission, 2024[106]). In agriculture, the Data Governance Act can support the emergence of agri-data co‑operatives and voluntary data-sharing arrangements.
Challenges on AI uptake in EU agriculture
Broadband connectivity on agricultural fields: in-field connectivity remains uneven across and within EU countries, affecting farmers’ ability to deploy real-time sensor networks, use digital tools or access cloud-based AI services critical to precision farming (OECD, 2023[107]; European Union, 2024[108]). Spain has significantly narrowed the urban-rural broadband coverage gap. However, many rural areas across EU countries still lack reliable and high-quality broadband connectivity (OECD, 2023[107]). Given the remote locations of many agricultural sites, farmers will also need to rely on solutions such as mobile networks (4G, 5G) and satellite connectivity to address connectivity deficits and enable timely AI adoption.
Awareness of security vulnerabilities of agricultural AI systems: stakeholders have low awareness of the significant infrastructure risks from widespread adoption of AI systems (Gao et al., 2024[109]). They need to increase their awareness to ensure that AI systems in agriculture are implemented with the highest cybersecurity standards to mitigate any potential strategic threats.
Fragmented data landscape in EU agriculture: the EU landscape is characterised by lack of interoperability between systems and devices, inconsistent data formats, gaps in real-time and historical data, and limited availability of high-quality, standardised datasets across Member States. Furthermore, issues of trust and questions over ownership, privacy and liability limit data sharing, reducing the availability of data needed for effective AI solutions. To address this challenge, initiatives like the CEADS seek to establish common data-sharing specifications, governance frameworks and secure mechanisms for trusted, cross-border data exchange and integration (CEADS, 2025[110]).
Data portability: without effective mechanisms to ensure farmers can easily transfer their data between digital platforms or agricultural machinery providers, they risk being locked into proprietary ecosystems. The inability to control and re-use their own data across systems can limit their ability to adopt better or more affordable tools, reduce competition and hinder innovation. This, in turn, can prevent farmers from fully benefitting from precision agriculture and other digital services (OECD, 2021[111]; Reimsbach-Kounatze and Molnar, 2024[112]).
A persistent skills gap: many farmers lack the digital literacy to navigate and operate AI tools, interpret data outputs or integrate automated systems into daily operations (Farmtopia, 2023[113]). This disconnect is particularly acute among older farmers and smallholders, who may view AI as overly complex or irrelevant to their practices or lack the skills to understand and use these tools. Vocational training programmes, which play a key role in providing farmers with the skills and knowledge for primary agricultural production, do not sufficiently cover digital literacy. These programmes could amplify their impact by updating and adapting their content to respond to new skill demands, particularly for digitalisation (Goller, Caruso and Harteis, 2021[114]). Without such enhanced vocational programmes, alongside workshops and training programmes that combine theory and practice, farmers may miss out on the benefits of AI-driven innovations like nutrient management algorithms or disease prediction models. This critical need is reflected in CAP 2023-2027, which calls upon Member States to provide digital skills training as part of their rural development and advisory services in their CAP Strategic Plans.
More than 64% of EU farms are under 5 ha. Consequently, Member States need to appropriately account for the needs of smallholders when allocating CAP support to modernise agriculture and rural areas through fostering and sharing knowledge, innovation, and digitalisation (European Commission, 2025[91]). Such an approach is in line with the direction proposed by the European Commission for the CAP after 2027. The proposal includes more targeted and flexible income support for young farmers, small and mixed farms and those operating in rural areas, along with an improved plan to address local and sectoral-specific needs (Directorate-General for Agriculture and Rural Development, 2025[115]).
Several initiatives in other EU programmes may both inspire and create synergies with CAP support measures. For example, Farmtopia project, a Horizon Europe initiative, is helping small-scale farmers to modernise. It distributes agricultural digital solutions to 64 000 small farms and has set up 18 Sustainable Innovation Pilots to test cost-effective AI solutions tailored to smallholders (Farmtopia, 2023[116]).
Regulatory complexity and data privacy concerns: many agricultural stakeholders lack technical expertise to comply with the EU AI Act, particularly in meeting transparency and safety requirements. This may contribute to the hesitation to launching products in the European Union and delay deployment of AI systems. Farmers also express apprehension about data ownership and misuse, particularly when sharing information about business-specific knowledge with third-party platforms (Gebresenbet et al., 2023[117]). Along with the CEADS and the Data Act, edge computing and blockchain solutions could work as partial remedies, enabling localised data processing and secure record keeping. However, their adoption remains limited (Gebresenbet et al., 2023[117]).
Among the AI use-cases, AI is emerging as a key enabler in the European Union in several areas: i) AI‑powered agricultural robots; ii) predictive analytics; and iii) crop, soil and livestock monitoring – with the last two use-cases being integral components of precision farming. Part 2 of this chapter discusses these three use-cases based on literature review and stakeholder interviews. It explores their real-world applications, benefits and barriers to adoption, and ways to widen their adoption in agriculture in the European Union.
The team synthesised insights to identify key use-cases, barriers and recommendations across the three focus areas, summarising key results in Table 2.2. This approach was chosen to capture a range of perspectives while recognising the constraints of voluntary participation. The findings aim to reflect the views expressed by stakeholders engaged through these channels, while acknowledging they do not fully represent the sector.
AI-powered agricultural robots
Main use-cases reported in literature
AI plays a pivotal role in reducing manual labour and advancing sustainability through the adoption of agricultural robots. Like many smart robots, agricultural robots have advanced perception and autonomous decision-making abilities and can execute tasks with high precision, accuracy and efficiency. The core architecture of agricultural robots comprises four key components: vision systems that capture data using thermal, colour, depth and multispectral cameras; control systems that enable intelligent decision making through AI algorithms; mechanical actuators for precise operation; and mobile platforms that enable accurate navigation in agricultural environments (Cheng et al., 2023[118]). These systems work together to perform complex agricultural tasks autonomously or semi-autonomously.
There is increased interest in the use of and development of robotic devices, specifically for crop production and livestock in dairy farming such as robotic milking (McFadden et al., 2022[74]). Whereas human workers are hindered by fatigue and harsh environmental conditions, agricultural robots can withstand the demands of various agricultural settings for extended periods while maintaining optimal productivity levels. Projections show that around half of dairy in northwest Europe will be robotically milked by 2025 (McFadden et al., 2022[74]).
The European market for agricultural robots is also experiencing significant growth, estimated at EUR 6.37 billion in 2025 and expected to reach EUR 12.77 billion by 2030 (Mordor Intelligence, 2024[119]). This expansion reflects the growing role of automation to address increasing agricultural production needs and labour shortages. Germany leads the European agricultural robotics market, with the Netherlands and France following closely (Mordor Intelligence, 2024[119]). Horizon Europe has also funded and supported the development of robotics through a range of projects (Box 2.1).
The EU market is advancing through strategic partnerships among players in semi-autonomous AI technologies, including CNH Industrial, Monarch Tractor and the CLAAS Harvest Centre.
CNH Industrial has partnered with BlueWhite to integrate autonomous solutions into tractors for orchards, vineyards and specialty crops. With One Smart Spray, it delivers green-on-brown spraying technology, targeting weeds without affecting crops. It works with Advanced Farm Technologies to develop robotic harvesters for strawberries and apples. Finally, it partners with ASI to advance robotic platforms for agricultural machinery (Morrison, 2025[120]).
Smart Droplets: Smart Droplets translates real-time data collected from field demonstrators into actionable insights to optimise pesticide and fertiliser use. Aligned with the Green Deal goals, the project will demonstrate how autonomous robotic platforms, innovative spraying, digital twin and AI models can deliver environmental and economic benefits. (End date: February 2026) (CORDIS, 2025[121]).
Flexigrobots: Under Horizon 2020, the Flexigrobots project built an open AI platform to enable the use of flexible and heterogeneous multi-robot systems (unmanned aerial vehicles and various types of unmanned ground vehicles) for intelligent automation of precision agriculture operations. Three real-life pilots – grapevines in Spain, rapeseeds in Finland and blueberries in Serbia and Lithuania – demonstrated the significant economic value of these multi-robot systems in operational environments. (Closed: December 2023) (CORDIS, 2025[122]; Flexigrobots, 2025[123]).
Robs4Crops: A Horizon 2020 project, Robs4Crops tested mechanical weeding and spraying in vineyards, crop fields and apple orchards to reduce manual labour, increase safety and optimise input in large-scale pilots in the Netherlands, France, Spain and Greece. (Closed: December 2024) (CORDIS, 2025[124]).
BACCHUS: Under Horizon 2020, the BACCHUS project tested a smart robotic system built upon autonomous mobile platforms in vineyard environments. It demonstrated the system across various use-cases such as robot harvesting, vineyard inspection and yield prediction, thinning of small grapes to improve quality and vineyard health detection. (Closed: November 2023) (CORDIS, 2024[125]; BACCHUS, 2022[126]).
ROMI: Under Horizon 2020, Robotics for MIcrofarms (ROMI) developed a land-based robot, combined with a weeding app for the robot, to help microfarms (small, diversified vegetable farms) to reduce manual labour and increase productivity. (Closed: July 2022) (CORDIS, 2024[127]).
agROBOfood: Under the agROBOfood project, a network of digital innovation hubs (DIHs) accelerated the sector’s digital transformation through adoption of robotic technologies. Innovation experiments brought 92 DIHs into the network and supported 110 small and medium-sized enterprises with an open call of EUR 8 million in funding. (Closed: February 2024) (agROBOfood, 2024[128]; CORDIS, 2025[129]).
CoRoSect: Under Horizon 2020, the CoRoSect project built cognitive robotic ecosystems to replace repetitive and physically demanding tasks needed for insect farming. Large-scale pilots in five insect farms in five European countries tested its systems on handling and tending the insects, and environment sensing. (Closed: March 2024) (CoRoSect, 2021[130]; CORDIS, 2025[131]).
SOMIRO: Under Horizon 2020, the Soft Milli-robot (SOMIRO) project developed and demonstrated the world’s first energy-autonomous swimming milli-robot to reduce the environmental impact of farming, over fertilisation, pesticide use and overfeeding. (Closed: June 2024) (CORDIS, 2025[132]).
NI: Under Horizon 2020, the Natural Intelligence for Robotic Monitoring of Habitats (NI) project conceived and tested robots equipped with AI and articulated soft-robotics bodies that can move effectively in wild unstructured environments such dunes, grasslands, forests and mountains and also monitor natural habitats. (Closed: March 2024) (CORDIS, 2025[133]).
After testing its MK-V tractor in European vineyards and orchards in October 2023, Monarch Tractor expanded its operations into Europe, establishing its headquarters in Belgium (Monarch, 2024[134]).
CLAAS expanded its investment in AgXeed’s autonomous technologies to fuel its farm machinery in such countries as Germany and Switzerland (CLAAS Harvest Centre, 2025[135]).
Aerial robots
Agricultural robots come in various forms depending on farm type, crop specialisation and farm size. One category is UAVs or drones, which have many use-cases – from mapping fields to monitoring and spraying crops to managing livestock (EASA, 2025[136]). Equipped with remote sensing capabilities, these devices can help farmers identify diseases or areas with low production. They can also collect information through near-infrared, thermal spectrum cameras or laser scanners (Moysiadis et al., 2021[137]). UAVs provide an efficient bird’s eye view of fields and allow for rapid inspection with comparatively low operational costs compared to manned airborne and satellite inspection (Moysiadis et al., 2021[137]). While UAVs can be used to spray plant protection products (“aerial spraying”), Article 9 of Directive 2009/128/EC generally prohibits this practice (European Union, 2019[138]). However, in 2023, the European Union relaxed some regulations on using drones for spraying (EASA, 2023[139]).
Ground robots
Ground robots, or unmanned ground vehicles (UGVs), encompass a variety of robotic use-cases for agriculture – from field robots, fruit and vegetable robots to animal husbandry robots (Cheng et al., 2023[118]).
Field robots perform a variety of crop production tasks, either semi-automatically or automatically. Tillage robots automate soil cultivation through advanced navigation systems; seeding robots precisely sow seeds and can also fertilise and water in one pass; crop protection robots use sensors and intelligent algorithms to apply pesticides or fertilisers efficiently and safely, including both ground-based and aerial (drone) systems; information-collecting robots gather data on plant phenotypes, soil and environmental conditions to inform farm management; and crop harvesting robots use machine vision and advanced navigation to automate the harvesting process, including row alignment, obstacle avoidance and optimised harvesting routes (Cheng et al., 2023[118]). These robots are primarily wheeled, with some use of caterpillar tracks and drones for specific applications such as aerial spraying. Their development is driven by the need to reduce labour, increase precision and support sustainable digital agriculture.
Fruit and vegetable robots are specialised for high-value horticultural tasks and comprise five main types. Transplanting robots automate the accurate and stable planting of seedlings using advanced manipulators and control systems. Patrolling robots autonomously navigate orchards or greenhouses to monitor crop maturity, environmental parameters and pest presence, often using AI and IoT technologies. Pesticide spraying robots deliver precise and targeted pesticide applications using servo-controlled nozzles, flow control systems and machine vision, reducing chemical use and environmental impact. Gardening robots perform tasks such as pruning, lawn mowing, irrigation and fertilisation, often adapting to dynamic garden environments with advanced navigation and sensor systems. Finally, picking robots use machine vision, DL and sometimes soft robotic grippers to detect, localise and gently harvest fruits and vegetables, aiming to minimise damage and maximise efficiency (Cheng et al., 2023[118]). These robots are designed to address labour shortages, improve consistency and enhance the sustainability of fruit and vegetable production.
Animal husbandry robots automate and optimise key livestock and poultry management tasks. Breeding robots, such as automated disinfection and environmental monitoring systems for poultry and livestock houses, often use IoT and remote control for improved animal health and productivity. Feeding robots deliver precise rations to livestock and poultry, reduce feed waste and may use force feedback, LiDAR and advanced obstacle avoidance for efficient operation. Milking robots allow cows to be milked autonomously at optimal times using vision systems, sensors and data analytics to improve milk yield and quality. Finally, egg-collecting robots autonomously navigate poultry houses to collect eggs, monitor flock health and avoid obstacles. These systems are designed to reduce labour, enhance animal welfare and enable data-driven management in modern animal husbandry operations (Cheng et al., 2023[118]).
UGVs performing agricultural tasks are often equipped with a navigation system that allows for localisation, mapping, motion control and path planning (Chakraborty et al., 2022[140]) . Traditionally, these ground robots perform simple path planning based on their Global Positioning System (GPS) system and maps of the agricultural field. As noted earlier, “FarmDroid” in the Netherlands autonomously handles weeding tasks using GPS and can switch between weeding and seeding operations (Strodt, 2022[141]).
Despite their widespread usage, GPS systems equipped in agricultural robots face difficulty in high-precision navigation in agricultural terrain, which is often complex, unstructured and unpredictable. They also have low satellite signal, particularly in covered areas, greenhouses or mountainous regions (Chakraborty et al., 2022[140]). The application of ML in UGVs addresses these navigation challenges by equipping robots with adaptive decision making to dynamically incorporate real-time information to work around unforeseen obstacles and reduce error (Alrowaily et al., 2024[142]). For instance, the simultaneous localisation and mapping (SLAM) algorithms locate the state of a ground robot using input sensor data, while building a multi-dimensional map of its surrounding environment (Aguiar et al., 2020[143]). By considering data such as historical yield records, soil variability and environmental conditions, ML algorithms allow robots to adapt to particular crop environments, leading to more precise and adaptable execution of agricultural tasks. Combining ML with real-time sensor data, UGVs can autonomously correct faults in path planning and improve performance over time.
UGVs often perform various crop production tasks like seeding, weeding, spraying and harvesting. They have been tested on crops such as grapes, peppers, cucumbers, tomatoes, asparagus, sunflowers, sugar beet and hazelnuts (Moysiadis et al., 2021[137]). AI-powered ground robots improve precision agriculture by maximising crop output. To that end, they optimise planting depths and help farmers fine-tune their planting (Alrowaily et al., 2024[142]). All-terrain vehicles for agriculture, for example, have been equipped with powerful navigation systems and sensors to carry out autonomous tasks like planting, harvesting and monitoring (Padhiary et al., 2024[144]).
Classical image processing and traditional ML techniques, such as SVM, DT, RF, AdaBoost and RANSAC, are commonly used in agricultural UGVs (Agelli et al., 2024[145]). The Visual SLAM (vSLAM) algorithm identifies obstacles through images collected through cameras without having to use multiple sensor data and integrate numerous algorithms (Agelli et al., 2024[145]). These methods, however, face considerable limitations in resource-constrained agricultural environments. They require a controlled environment to ensure accurate and consistent results, along with a meticulous selection of algorithms for separate tasks in feature extraction, object detection and classification.
To overcome these challenges, DL techniques are increasingly used to analyse large visual datasets collected through cameras. In contrast to classic computer vision algorithms, DL models can extract complex features from images of their background environment with less interference. This allows for a more accurate and robust execution of detection and recognition tasks in agriculture. The YOLO (You Only Look Once) model, for example, has been widely adopted for real-time agricultural object detection. It includes tasks like fruit counting, pest identification and weed discrimination (Badgujara, Pouloseb and Gan, 2024[146]).
Moreover, computer vision foundation models that can enable agriculture-specific applications are increasingly being developed. Recently, multimodal approaches such as Apple’s FastVLM extend these capabilities by enabling natural language explanations of visual agricultural scenes in real time. This reduces the need for extensive domain-specific training (Apple Machine Learning Research, 2025[147]).
The main hindrance to adopting the above-mentioned techniques and solutions is the limited availability of relevant high-quality datasets.
Expected benefits from UGVs when they reach their potential are numerous. For example, they will reduce labour, which can also decrease operational costs. In addition, they can offer precise application of fertilisers and pesticides, which might lessen environmental impact and produce better products. Finally, the small size of the UGVs compared to heavy machinery will avoid massive soil compaction and reduce energy consumption. However, the potential for rebound effects, where efficiency gains can trigger greater agri-chemical applications or higher machinery usage, erode some of the environmental gains.
Agricultural robotics use AI to boost their situational awareness when performing agricultural tasks (Rejeb et al., 2022[24]). Their momentum to commercialisation stems from their wide range in utility in spraying, crop monitoring, field surveying and harmful weed detection (Elbasi and Mostafa, 2023[27]). Their ability to inform farmers about weather conditions, crop health diagnostics and resource use is leading them to gain traction.
Insights from interviews
Most commonly adopted use-cases
AI-driven agricultural robotics are increasingly seen as a transformative force in EU agriculture for their potential to address labour shortages and optimise the efficiency and precision of farming operations. Interviewees described adoption of agricultural robotics in the European Union as being in an incipient but accelerating stage. This stage emphasised automation features embedded within machinery rather than fully autonomous vehicles. For instance, one interviewee described the use of AI in vision systems of combine harvesters. These systems use DL to classify grain quality and cleanliness, estimate straw chop quality and enable real-time, automatic adjustment of machine settings, such as maintaining ideal harvesting speed. Another interviewee mentioned that such innovations have shown measurable benefits, leading up to productivity increases of 20%, grain loss reductions of 33% and improvements in throughput of 25%. These features can reduce the need for constant human supervision and mitigate the risk of operator fatigue over long working hours.
Another use-case of AI adoption is guidance automation for machinery such as balers and tillage robots. One interviewee identified balers that have achieved up to 15% higher productivity and 7% lower fuel consumption through their GPS-based autoguidance algorithm. Similarly, vision-guided tractors use AI to automate row-following and allow operators to focus on higher-value tasks. Notably, green-on-brown spraying systems, a computer vision technology that identifies weeds on fields by relying on colour differences (Korolkova, 2024[148]), can reduce herbicide use by up to half by spraying only where crops are present. This use-case is particularly valued in high-value specialty crop sectors such as vineyards and orchards, where precise documentation of sustainable practices can enhance product value and marketability. An interviewee stated that intelligent sprayers also employ computer vision to selectively target weeds, illustrating the trend towards integrating AI into semi-autonomous machinery to reduce manual labour and input costs.
Key barriers and challenges
Market hesitation: despite the advancements of AI-driven systems in agriculture, interviewees noted that most EU systems remain operator-supervised. In Europe, many farmers are highly pragmatic and seek clear evidence of economic payback before investing in advanced systems.
The diversity of farm sizes and types: AI solutions must be adaptable to both large-scale and smallholder contexts. The adoption of AI-enabled robotics is more pronounced among larger farms and those cultivating high-value specialty crops such as vineyards and orchards, where the cost-ratio is more favourable. Smaller farms often remain reliant on traditional practices due to cost and scale considerations.
The small size and heterogenous landscape of European farms and fields: several interviewees noted the European Union is a slow market compared to the United States, Canada and South America, where farmers are willing to test new technology in their large fields. One interviewee stated that traction for ground robots comes from countries like Ukraine, known for its vast agricultural fields, and Germany, with its medium-sized fields and farms. Another interviewee added that it is much easier to use drones and transferable technologies in other countries, such as Ukraine, where certification and documentation requirements are less complex and restrictive than in the European Union.
Regulatory uncertainty: alongside market hesitation, regulatory uncertainty has blocked full autonomy and widespread commercial use of robotics in agriculture. With the EU AI Act entering into force, there are areas where industry would benefit from further guidance and clarification. These include issues such as classification of agricultural robots operating in open fields and autonomous vehicles in urban environments as high-risk AI systems, as well as on interplay between the AI Act with EU Machinery Regulation. Interviewees emphasised the urgent need for clear, streamlined and agriculture-specific regulatory guidance.
The absence of a fast-track homologation pathway for agricultural robotics: without a streamlined mechanism for testing and certifying new technologies, companies face significant delays and costs in bringing new products to market. This is especially challenging for small and medium-sized enterprises (SMEs), which lack the resources of larger original equipment manufacturers. Interviewees expressed the concern that overregulation risks replicating the experience of the biotechnology sector. In that sector, EU policies led to an outflow of talent and investment to other markets, such as the United States and China. They noted that a similarly cautious approach could make the region dependent on foreign technology and digital infrastructure.
Gaps in rural connectivity: spotty connections in rural areas and data interoperability also present significant obstacles. This is especially true for edge-AI applications, such as crop monitoring and precision spraying, which require real-time data processing. Scaling up AI in agriculture requires robust digital infrastructure, high-quality data and affordable access to cloud and graphics processing unit resources. Running multiple AI models and equipping them into robotics systems can require extensive computational resources, and present additional challenges in processing speed in rural areas (Agelli et al., 2024[145]). Interviewees stressed the growing need for scalable, high-performance computing infrastructure to support development and deployment of advanced AI models across diverse farming contexts. They also noted that rural broadband coverage remains uneven across the European Union, limiting the operational effectiveness of AI-driven equipment on farms. While EU policy frameworks such as the Digital Decade targets and the CAP aim to support rural digitalisation, implementation remains slow and fragmented.
Quality and interoperability of agricultural data: factors such as the quality and interoperability of agricultural data are more limiting than the sheer volume of data. Much farm-level data are siloed across different proprietary platforms and incompatible formats, making them difficult to aggregate and analyse effectively for AI applications. This challenge is exacerbated in the EU context by the high degree of heterogeneity in farming systems. Differences in language, farm size, topography, crop types and national regulatory frameworks mean that agri-tech solutions often require customisation and localisation, limiting their scalability. As a result, the European Union does not function as a single, unified market for agri-digital technologies.
Outlook
The European market exhibits unique characteristics that shape the adoption of agricultural robotics. In North America, larger fields and a focus on throughput drive demand for high-capacity machines. Conversely, European farmers, especially those with smaller holdings, tend to prioritise minimising losses and maximising yield from limited acreage. As a result, there is a greater appetite in Europe for machines with onboard technology tailored to each farmer’s needs and crops, particularly among specialty growers. However, smaller farms, which are prevalent in many EU countries, often seek cost-effective solutions. They may be more cautious in adopting high-tech equipment without a clear economic payback.
Interviewees expressed optimism about the potential for wider adoption of AI-driven agricultural robotics in the European Union. However, they emphasised that building trust and demonstrating clear value will be crucial to accelerating uptake among farmers. To overcome trust gaps, policies should be geared towards promoting farmer on-field experimentation with digital tools. Sharing the risk with technology providers could inventivise risk-averse farmers to participate in these field tests (McFadden, Casalini and Antón, 2022[149]).
On-field experiments can be coupled with standardisation initiatives. If standardisation is inappropriate, voluntary codes of conduct or third-party certification can help boost the trust of farmers in new digital equipment. The OECD Tractor Codes, for example, implemented a harmonised standard and certification process to assess tractor performance. They decreased information asymmetry across governments, boosted consumer trust and increased international tractor trade (McFadden, Casalini and Antón, 2022[149]; OECD, 2025[150]).
One interviewee stated that complete autonomy of AI-driven agricultural robotics remains the long-term objective. The interviewee saw the market as increasingly receptive to AI solutions that allow farmers to shift from labour-intensive routine field operation to managerial and decision-making tasks. Ongoing R&D focuses on achieving full autonomy, developing modular platforms and integrating edge-AI to enable real-time analytics and task execution.
Key recommendations
Regulatory clarity and sector-specific guidance for AI-powered agricultural robotic solutions could enhance adoption significantly. Moreover, there is a need for investment in digital infrastructure; support for data sharing and interoperability; and a balanced approach to regulation that avoids stifling innovation or driving it abroad. The following recommendations are suggested to facilitate wider AI adoption in agricultural robots tailored to the European Union:
Promote demonstration projects and peer-to-peer learning to build trust among EU farmers. Share successful cases highlighting the economic and sustainability benefits of agricultural robots in increasing yields. Interviewees noted that robotics adoption requires new skills for both farm operators and the broader workforce. Synergise these efforts by establishing standardisation and third-party certification processes for agricultural robotics.
Provide grants for start-ups and SMEs that are developing robotics solutions tailored to European small farms and specialty crop farms.
Prioritise the development and adoption of standards to facilitate data sharing and prevent monopolisation by large equipment manufacturers.
Predictive analytics
Main use-cases reported in literature
Predictive analytics is a cornerstone of precision agriculture, enabling farmers to optimise agricultural production and resources, mitigate climate risks and meet sustainability targets. Most importantly, it enables them to boost their agricultural production and minimise profit loss. In the EU market, AI is used to perform detailed analytics in a variety of tasks, such as crop yield forecasting, disease risk prediction and breeding optimisation.
AI-based crop yield forecasting
AI-powered crop yield forecasting is emerging as a promising tool for EU farmers, with early applications showing potential to plan, manage and optimise their agricultural production. Modern yield forecasting systems leverage a combination of remote sensing data (e.g. Sentinel-2 satellite data), in-field sensors, and ML and DL techniques to generate accurate, field- and region-specific predictions before harvest. Unlike traditional statistical or purely process-based models, these AI-driven approaches can capture complex, non-linear interactions between soil properties, crop growth stages and climatic variables, leading to more robust and timely yield estimates (Paudel et al., 2022[151]). For example, ML models integrating satellite imagery, meteorological data and soil characteristics outperform conventional linear crop models in predicting yields for five different crops in Germany, France and the Netherlands (Paudel et al., 2021[152]; Paudel et al., 2022[151]). As an example, GeoPard – a Germany-based agri-tech company – applies similar approaches operationally. It combines remote sensing data, soil lab results and machinery inputs to create digital field twins and generate predictive yield and input maps tailored to field-specific variability (Box 2.2).
Numerous studies have demonstrated the effectiveness of ensemble ML learning, a technique that aggregates two or more ML models to improve accuracy (Murel and Kavlakoglu, 2024[153]). Methods such as RF, Gradient Boosting and AB, consistently outperform classic regression and single-algorithm approaches for yield prediction across diverse crops and regions (Asadollah, Jodar-Abellan and Pardo, 2024[154]). RF models provided more accurate yield predictions for wheat, maize and potato at both global and regional scales (Hasan et al., 2023[155]).
Across the European Union, national studies confirm the utility of AI models in diverse agro-climatic settings:
In Spain, Segarra et al. (2022[156]) found RF to be the most accurate in predicting wheat yields among three ML algorithms (RF, SVM, Bayesian Regression).
In Greece, Bebie et al. (2024[157]) used Sentinel-2 data over multiple years to demonstrate the robustness of RF and k-nearest neighbours’ models to estimate wheat yield in the Eastern Mediterranean.
Darra et al. (2023[158]) showed that an ensemble method combining automatic relevance determination regression and support vector regression achieved best accuracy in predicting tomato yield from Sentinel-2 image data. They used an open-source AutoML technique to automate the selection of ensemble ML models, highlighting the value of AutoML (Darra et al., 2023[158]).
In Hungary, Amankulova et al. (2022[159]) applied RF to forecast reliable sunflower yield three to four months before harvest.
In Austria, Pejak et al. (2022[160]) demonstrated that the combination of high-resolution remote sensing data and advanced ML methods can accurately forecast operational, field-scale yields in Austrian soybean production. Their approach found the stochastic gradient descent algorithm to provide the best predictive performance (Pejak et al., 2022[160]).
These examples highlight the potential of advanced ML models, particularly ensemble methods such as RF, combined with Earth observation data (Sentinel-2 satellite imagery), to deliver accurate and timely yield estimates for a range of crops and environments across the European Union. Accurate yield forecasting enables European farmers to make more informed decisions about input use, harvest logistics and market planning. Ultimately, this supports greater resilience and efficiency in agricultural production. Accurate early- and mid-season forecasts can allow for better fertiliser and irrigation scheduling.
At the regional and national levels, ML-driven yield forecasts can help governments identify potential shortfalls or surpluses, improving food security and enabling timely interventions. Countries that face high climate variability or labour shortages can proactively adopt AI to develop regional ML models to provide more granular insights. For example, regional ML models for six major crops (soft wheat, spring barley, sunflower, grain maize, sugar beets and potatoes) in nine EU countries consistently outperformed linear trend models, especially for early-season predictions (Paudel et al., 2022[151]).
The Joint Research Centre of the European Commission is also advancing the use of expert-informed, explainable AI models for yield forecasting. It is integrating decades of agro-climatic knowledge with Big Data to detect climate hazards and forecast their impact on yields across the EU (Joint Research Centre, 2025[161]). These models provide probabilistic outputs and clear explanations of key risk factors, such as droughts or heatwaves during critical crop growth phases. In so doing, they empower farmers and policymakers to make informed decisions about resource allocation, risk mitigation and interventions (Essenfelder, Toreti and Seguini, 2025[162]).
GeoPard Agriculture is a Germany-based, cloud-powered precision agriculture platform designed to help farmers, agronomists and agribusinesses implement data-driven, sustainable farming practices.
Its platform processes a wide range of agricultural data, such as spatial and temporal data, sourced from satellite imagery (such as Sentinel-2 from the European Space Agency), commercial remote sensing providers like Planet, agricultural machinery data (e.g. from harvesters), soil laboratory results and topographic scans. AI and machine learning are used across several core areas, including for crop boundary detection, yield prediction, nitrogen use efficiency-analysis and detection of limiting soil or environmental factors.
The company often works closely with farmers and agribusinesses to develop customised models and validate them with ground-truth data. They have developed partnerships with universities and scientific institutions to enhance their model development. In addition, they rely on internal data augmentation pipelines to train robust AI systems when data are sparse.
The platform is accessible as a browser-accessible interface, which allows users to perform advanced field variability analyses, build digital field twins and generate zone-specific prescription maps for nutrient and input application. These tools enable more precise management of field operations, potentially increasing productivity and resource efficiency. Users can also run custom analytics using pre-set equations or design their own formulas within the platform.
As an independent provider, GeoPard emphasises data privacy and neutrality, allowing users to retain full control over their data.
Source: GeoPard (2025[163]), “Independent precision agriculture platform”, https://geopard.tech/.
Disease risk prediction
AI-enabled disease risk prediction systems can provide farmers with powerful tools to detect and manage crop diseases and pest outbreaks before visible symptoms appear (Gupta et al., 2024[164]). These systems leverage data from IoT sensors, remote sensing and ML to monitor plant health, environmental conditions and pest populations in real time. This enables timely and targeted interventions that reduce both economic losses and environmental impact.
Sensoring and AI Algorithms for Early Crop Disease Detection (SAIA), part of the SmartAgriHubs Flagship Innovation Experiment, is a notable example. Focused on the Iberian Peninsula, the project developed risk maps to facilitate early detection of plant pests in Mediterranean vineyards, cork trees and olive groves (Volta, 2022[165]). It leveraged IoT sensors, AI algorithms and weather forecast models (SmartAgriHubs, 2022[166]) in collaboration with the Digital Innovation Hub in Andalucía, and research teams at the University of Malaga and Córdoba (European Commission, 2025[167]).
Additional European research highlights the practical potential of AI in plant disease management. In Romania, IoT sensor networks combined with ML enabled precise detection and early intervention for grapevine diseases. By collecting environmental and plant-related data and applying ML algorithms, researchers achieved high-precision monitoring of grapevine health, enabling early intervention and effective disease management (Hnatiuc et al., 2023[168]). In Sweden, CNNs have been used to analyse leaf images and detect grapevine diseases such as downy and powdery mildew, achieving validation accuracies of up to 75% for binary classification tasks (Vanglund, 2021[169]). These AI-driven approaches provide decision support to farmers, helping identify infected plants at various disease stages and improving overall vineyard management.
Farmonaut’s AI-powered platform offers another example of real-world adoption. The platform is widely used across Western and Eastern Europe to optimise pesticide use and promote sustainable farming practices (Farmonaut, 2025[55]). It integrates satellite imagery, ML and weather data to detect pest outbreaks and recommend targeted pesticide applications. Farmonaut’s system detects pest issues early and recommends targeted pesticide applications, reducing overall usage by up to 30% without compromising yields (Farmonaut, 2025[55]). Such predictive analytics can enable farmers to achieve reductions in input use and improve resource efficiency.
AI-driven breeding optimisation
Beyond disease prediction, AI is transforming plant breeding by accelerating development of drought- and climate-resilient crop varieties. To that end, it analyses large-scale genomic, phenotypic and environmental datasets to optimise breeding decisions (Petrović et al., 2024[170]). In Central Europe, the Czech University of Life Sciences in Prague collaborated with international partners like the International Crops Research Institute for the Semi-Arid Tropics and Phenospex. Together, they apply advanced AI analytics and high-throughput phenotyping to screen thousands of wheat and maize genotypes for drought adaptation (ICRISAT, 2023[171]). These efforts have enabled rapid identification of “stay-green” traits in sorghum and drought-tolerant chickpea genotypes. Improved varieties boosted yields by up to 25% during end-season droughts and supported food security for farmers facing climate extremes.
AI-driven breeding significantly reduces the time required to bring new, resilient varieties to market, allowing farmers to adapt more quickly to shifting climate patterns and emerging pests. For example, AI models can predict how certain wheat or maize hybrids will perform under simulated drought or heat stress, enabling breeders to select the most promising crosses for further development (Khan et al., 2022[172]; Farooq, Mazhar and Jan, 2025[173]). This not only stabilises yields and reduces the risk of crop failure, but also supports the diversification of crops such as protein-rich legumes or specialty grains. This, in turn, helps farmers access premium markets and increase their resilience to both environmental and economic shocks (WUR, 2023[174]). The integration of AI into breeding is a key enabler for sustainable intensification and adaptation in EU agriculture.
Insights from interviews
Most commonly adopted use-cases
As a key use-case of precision farming, predictive analytics is being used to support decision making in crop management, input optimisation and breeding. Several interviewees mentioned that predictive analytics is used for in-season crop monitoring, disease risk detection and farm work management.
One interviewee illustrated that remote sensing, soil and machinery data can be integrated to create digital field twins and generate field-specific recommendations for input use, crop rotation and harvest logistics. Several interviewees highlighted the use of predictive models for disease and pest risk management. These systems can ultimately help both large agribusinesses and smaller farms to optimise resource allocation and reduce unnecessary field visits for farmers.
One interviewee explained that AI models can be used to link genetic, phenotypic and environmental data, enabling them to predict the yield of new crop varieties and shorten breeding cycles by several years. This predictive approach is crucial for developing climate-resilient crops and supporting sustainable crop rotations.
Key barriers and challenges
Availability and quality: access to high-quality, harmonised and sufficiently granular data remains limited, especially for smaller farms and in regions with fragmented datasets. Interviewees noted that certain public datasets such as Sentinel satellite imagery remain valuable. However, key agronomic data, such as soil properties and field-level management records, are often proprietary, inconsistently formatted or simply unavailable. This limits the robustness and transferability of AI models across geographies and crops.
Economic and technical barriers: several interviewees noted that the front runners integrating AI solutions into their farm systems tend to be big farm owners or specialised farmers with the financial capacity to afford these technologies. While younger farmers are eager to invest in new AI-equipped tools, access to capital and difficulty in securing loans pose huge challenges. Many interviewees stressed that the return on investment for predictive analytics solutions is not always clear to farmers, particularly given the upfront costs of hardware, sensors and digital infrastructure. One interviewee stressed that applications like predictive analytics and disease monitoring and prevention require a collective approach; individual farmers often lack the resources or skills to fully leverage sophisticated analytics tools on their own. Farm advisory services and co‑operatives need to collaborate with farmers to support adoption at local levels. They must work together to build an ecosystem that enables data sharing, data merging and sharing of best practices.
The need for investment in on-farm sensors, edge devices and system integration: without the physical components of on-farm sensors, edge devices and system integration, the benefits of AI analytics cannot be fully realised. However, farmers often find it difficult to assess which systems are compatible with their operations and lack support in integrating new technologies into their workflows.
Scepticism among European farmers: several farmers identified scepticism as an additional hurdle. They noted the extra steps required to help farmers understand the tangible value that digital tools bring in terms of yield and profit. The complexity of integrating new tools into farm workflows, the need for digital skills and the diversity of farm sizes and practices across the European Union further complicate adoption. In addition, uncertainty around data ownership and usage rights affect trust. One interviewee stressed that AI in agriculture must be human-centred, with farmers actively involved in design and deployment. Engaging farmers’ feedback in the process of design, development, deployment and verification of the AI output is critical. This will allow for faster integration into farms through trust and also for development of a more robust and accurate AI system tailored for farm management.
Low levels of public and private funding for innovative agri-tech start-ups: low funding levels for agri‑tech start-ups, especially for those working on hardware and AI integration, further hinder progress. One interviewee noted that investors often favour pure software solutions due to their scalability and faster return potential, leaving hardware-intensive innovations underfunded and unable to scale effectively. Additionally, interviewees noted that EU funding mechanisms are resource intensive, requiring participation in multiple stages that are cumbersome and long, with low success rates. Furthermore, the EU agri-tech sector is fragmented across many countries, each with its own regulations, languages and market dynamics. This fragmentation makes it difficult for start-ups to scale quickly and for investors to see market opportunities.
Regulatory complexity, especially regarding data privacy, data sharing, model validation and interoperability, create additional hesitation. The lack of clear, harmonised interpretation of regulations across Member States, combined with uncertainty regarding the interplay between different regulations, make long-term investments in AI solutions riskier for both farmers and developers.
Outlook
Despite these challenges, interviewees were optimistic about the future of AI-based solutions such as predictive analytics in EU agriculture, anticipating both technological and cultural shifts. There is a clear trend towards integration of edge computing, cloud-based analytics and generative AI into farm management platforms, making predictive tools more accessible and actionable for a wider range of users. Agri-tech companies are leveraging generative AI equipped with general agronomist knowledge to upskill users and internal staff. In fact, several interviewees stated that companies are developing AI-enabled digital agronomists and chatbots that offer personalised, context-aware recommendations. Others are working to embed AI directly into field hardware for real-time decision support. An interviewee predicted that such wider use of LLMs will lower access barriers to agronomy, noting that commercialisation of LLM‑based chatbots may be difficult considering distinct EU regional and local conditions.
Interviewees expect adoption to accelerate as younger, more tech-savvy farmers take over operations. Demonstration projects, peer learning and regulatory clarity will also help build trust in AI solutions. There is also a growing recognition that predictive analytics can support not just productivity, but also resilience against diseases, compliance and risk management.
However, progress will depend on continued investment in digital infrastructure, data standardisation and farmer education, as well as on development of business models that make predictive analytics affordable and relevant for small and medium-sized farms. One interviewee also stressed the need for fair and transparent relationships across the ecosystem. While the Data Act provides a horizontal legal framework for data-access rights, further clarity is needed about certain sector-specific situations. This could include common use of agricultural machinery or farmer advisory services (e.g. irrigation recommendation services, fertiliser prescription, plant health monitoring) (Ryan et al., 2024[175]). To that end, sector-specific guidance may be needed, building on and updating initiatives such as the 2018 voluntary EU Code of Conduct on agricultural data sharing to ensure alignment with the Data Act (Ryan et al., 2024[175]).
Key recommendations
Recommendations are suggested to enable wider AI adoption in predictive analytics systems across EU agriculture, reflecting the key challenges and barriers noted by interviewees.
Invest in rural digital and cloud infrastructure: multiple interviewees emphasised the need for reliable, affordable broadband and cloud computing resources in rural areas. Infrastructure is essential to deploy AI-powered analytics systems and to enable real-time processing of large datasets, especially for smaller and medium-sized farms that lack the appropriate resources and infrastructure.
Accelerate development of interoperable data-sharing frameworks: several interviewees highlighted the importance of accessible, harmonised agronomic data. They recommended that data need to be made more readily accessible across regions and crop types. A wider promotion and increased awareness of the data spaces as part of the CEADS initiative can serve farmers and start-ups with improved access to data, bringing about faster innovation in predictive analytics tools. Incentives to share high-quality, field-level data must also accompany these data-sharing frameworks to foster wider AI adoption and development of models tailored to the European Union.
Provide targeted funding and incentives for start-ups and SMEs developing predictive analytics solutions, and for farmers adopting these tools: dedicated and simplified funding streams are clearly needed to support both innovation and adoption. Public funding should support start-ups and SMEs that are developing AI solutions for agriculture through impact-oriented, flexible funding mechanisms. These should target the development of long-term, sustainable business models. In parallel, and in addition to existing support for capacity building, financial incentives (such as subsidies, tax relief or cost-sharing schemes) could be offered to farmers investing in AI technologies, especially for acquiring sensors, infrastructure and integration support.
Build trust for farmers by focusing on the benefits of AI: to overcome cultural scepticism and improve uptake, efforts must focus on building trust and demonstrating the tangible benefits of AI tools. Policymakers should promote partnerships between technology providers, agronomists, extension services and farmer organisations to co-design, pilot and showcase successful AI applications. Farm advisory systems and co‑operatives should be strengthened to share best practices and support adoption at local levels.
Crop, soil and livestock monitoring
Main use-cases reported in literature
Computer vision and DL algorithms are used to process images of crops taken by cameras, UAVs or robots at different stages to monitor the health and growth of crops, soil and livestock (Tian et al., 2019[37]). Computer vision, and particularly image processing, is valuable in many health monitoring tasks in agriculture. Analysing images captured requires image processing techniques before farmers derive useful information. For this purpose, cameras in visible-spectrum, near-infrared, multispectral, hyper-spectral, thermal, laser scanners or synthetic aperture radar can be used depending on the desired application (Moysiadis et al., 2021[137]).
Crop monitoring
AI-powered sensors, cameras and remote sensing can be used to assess plant health, disease risk and crop monitoring tasks. The EU-funded STELLA project integrates AI models and modern sensing technology to enable early detection of plant diseases (STELLA, 2025[176]). The project developed 19 novel AI models for plant pest monitoring strategies using crowdsourced images across Europe and New Zealand (STELLA, 2025[176]). It is testing for eight different crop diseases across seven pilots covering arable, orchard and vineyard crops across Greece, Italy, Lithuania and France (STELLA, 2025[177]). Additionally, Bayer’s “MagicTrap” can automatically detect, categorise and count pest species using AI image recognition (Crop Science UK, 2024[178]). It has been in commercial use in Germany for three years, collecting over 800 000 images (Crop Science UK, 2024[178]). The Polish start-up Cropler offers a plant-level monitoring approach, using solar-powered cameras and AI to assess crop health and growth dynamics (Box 2.3).
VITO and Brussels-based start-up Superlinear (2025[179]) partnered to develop a crop classification and monitoring system across Europe to support the CAP. The partnership developed a DL-based model that uses multispectral data to predict crop types and create an accurate, up-to-date crop map spanning Europe. This tool reportedly allows farmers to accurately detect and map crops accounting for diverse crop growth patterns across different Member States.
Additionally, German start-up CORAmaps harnesses AI-based interpretation of satellite data to provide insights on crop fields and land conditions (Copernicus, 2024[180]). As part of its services, it assesses agricultural areas, providing information on crop types, quantities and quality. Leveraging radar data from Copernicus Sentinel-1, it enables easier crop monitoring through large-scale crop maps (Copernicus, 2024[180]).
Cropler, a Polish agri-tech start-up, provides AI-driven crop monitoring solutions through a solar‑powered, mobile camera system. Unlike traditional field or satellite monitoring, Cropler’s solution focuses on individual plants instead of fields.
At the heart of Cropler’s technology is its autonomous agri-camera system, which captures high‑resolution images multiple times per day. These solar-powered sensors also collect environmental data such as temperature, humidity and pressure, transmitting this information via GSM networks to a centralised platform.
Using a combination of image classification, growth stage detection and leaf damage recognition, Cropler provides farmers and agronomists with AI-based insights, including personalised recommendations. This set-up enables farmers and agronomists to monitor crop health remotely, track growth stages and detect early signs of stress, disease or pest infestations.
Cropler’s impact includes reducing field visits by up to 50% through remote monitoring; enhancing chemical application efficiency by up to 30% by aligning with precise plant growth stages; enhancing product quality by 15%; and improving yield planning and harvest sequencing.
Source: Cropler (2025[181]), “Let your crops speak: AI-powered digital agronomical assistant, come heat or rain”, https://www.cropler.io/about-us.
Soil monitoring
The EU-funded AI4SoilHealth project leverages AI to advance soil health monitoring for farmers across Europe. In partnership with 28 European institutions, AI4Soil Health tests soil from 13 pilot sites across Europe to develop a free, AI-enabled digital tool to measure soil health (AI4SoilHealth, 2025[182]). NEIKER, a project partner, is piloting the measurement technologies in 11 European regions to improve the design and reliability of the system (NEIKER, 2023[183]). One study used an AI-driven approach to quantify soil salinity in the European Union using datapoint from the LUCAS Soil survey, the largest soil dataset for Europe (Orgiazzi et al., 2017[184]; Shokri, 2024[185]), and the GBoost algorithm predicted soil salinity with the most accuracy (Shokri, 2024[185]).
Livestock health monitoring
AI-driven livestock monitoring is gaining traction, particularly in poultry and dairy operations. DunavNET’s poultryNET uses computer vision and edge computing to estimate chicken weight, detect dead birds and monitor flock behaviour (Box 2.4). Weight measurements achieve accuracy within 2-3% of manual scales, enabling more frequent, accurate monitoring of bird health. Its poultry management system also provides wellness analytics; its ML algorithm analyses chicken vocalisation and behaviour to detect signs of stress and early disease (AgroNET, 2025[186]). Smart digital technologies are being more widely adopted across Europe to enhance animal health, reduce antibiotic use and improve productivity. Farms in several countries have equipped dairy cows with smart ear tags that allow farmers to track and identify each animal each time it visits a smart robotic feeder (Gray, 2020[187]).
Additionally, Smart Pig Health, an EU project under Horizon 2020, leveraged AI and digitised wearable sensors on pigs to monitor behaviour and health symptoms to predict disease, reducing antibiotic use (SmartAgriHubs, 2020[188]). Through data collected from sensors, farmers could track anomalies in water consumption, temperature change and change in CO2 concentration, minimising risk of respiratory infections and boosting animal health (SmartAgriHubs, 2020[188]).
DunavNET, a technology company based in Serbia and Ireland, specialises in the design and implementation of solutions based on IoT and ML/AI technologies for various industrial sectors. Initially focused on IoT applications such as disease and irrigation models in vineyards and orchards, the company has since integrated AI technologies across its offerings, including computer vision, deep learning and generative AI.
Key AI solutions provided include the following:
agroNET: a cloud-based farm management platform that integrates IoT devices and AI analytics (computer vision) to assess crop maturity, size and pest presence. The platform allows optimisation of crop production, irrigation, pest control and supply chain transparency. It supports various agricultural operations, including vineyards, orchards, arable crops and greenhouses.
poultryNET: an AI-driven platform designed to enhance efficiency and control in poultry farming. It collects data within poultry barns and throughout the supply chain and uses edge-based computer vision to estimate chicken weight, detect dead birds and analyse flock movement and behaviour. The platform provides real-time decision support to improve meat quality, reduce environmental impact and ensure animal health.
agroBOT: a decision support agent (“digital agronomist”) in development that will integrate LLMs, farm-specific sensor data and user history to provide personalised advice via WhatsApp or the Internet.
Source: DunavNet (2025[189]), “From farm to fork: Digital farming”, https://dunavnet.eu/solutions/farming/; DunavNet (2025[190]), “AI-driven solutions for a smarter future”, https://wedoai.eu/.
Insights from interviews
Most commonly adopted use-cases
Interviewees reported that European agri-tech firms are adopting computer vision to address several key challenges in monitoring and management. One company has developed a solar-powered, multispectral camera that uses image classification and object detection to recognise plant growth stage across a range of crops, including vegetables, wheat and corn. Its AI-equipped camera can analyse daily and nightly images to minimise noise from shadows and improve accuracy. The system can detect leaf damage and provide actionable insights for optimising chemical application and harvest. The technology is sought out by agronomists, chemical companies and research organisations, highlighting its versatility and the growing demand for plant-level data in the European Union. For soil monitoring, one interviewee explained that ML is applied to satellite imagery, weather data and proprietary soil data to monitor regenerative agriculture practices and quantify carbon sequestration. This supports the participation of farmers in carbon markets and guides their transition to more sustainable practices.
Other interviewees described the integration of computer vision with IoT sensors for disease prediction and irrigation management in orchards and vineyards. Their solutions leverage an object detection CNN to process images from field cameras, estimate crop maturity and detect anomalies such as dead birds in poultry farms or pest infestations in insect traps. These systems enable remote, continuous monitoring, reducing the need for frequent field visits and allowing for more precise, timely interventions.
Key barriers and challenges
Despite the promise of computer vision for crop, soil and livestock monitoring, several barriers to widespread adoption were identified.
Lack of large, high-quality data across crop, soil and livestock: as the most significant challenge, high-quality crop-specific image datasets are needed to train robust AI models. One interviewee noted that collecting such data takes up both time and resources, particularly given the diversity of crops, field conditions and regional practices across the European Union. Another interviewee highlighted the difficulty of obtaining soil data because such information is often proprietary and requires the co‑operation of farmers. More soil samples are needed to measure and predict soil health, as soil data rely mostly on satellite imagery and are impossible to scale due to their scarcity. Moreover, as previously noted, lack of interoperability in public datasets acts as an important hindrance. Hardware costs and the complexity of scaling camera-based systems to thousands of fields also present obstacles, especially for small and medium-sized farms. Additionally, interviewees highlighted the difficulty of integrating computer vision data with other agronomic information (e.g. soil sensors, weather data) in a user-friendly way that supports actionable decision making.
Slow adoption of new technologies: EU farmers tend to be cautious adopters, often requiring extensive demonstration and evidence of return on investment before committing to new technologies. Adoption tends to be slower in crop farms than in livestock farms. One interviewee recounted how the owner of a small vineyard (10 ha) took three years to trust the technology: the first year was spent getting accustomed to the approach; in the second year, farming practices were compared to the system’s recommendations; in the third year, the owner started applying recommendations to the field.
Language diversity, fragmented farm sizes and varying digital literacy levels: the use of multiple languages in the European Union; the variety of farm sizes; and often low levels of formal education and digital literacy further slow adoption.
Regulatory hurdles: strict certification requirements, such as for drones and data security standards, add to the complexity and cost of deploying computer vision solutions at scale. Interviewees noted that investor interest in hardware is weaker than for software-only solutions, making it more difficult for start-ups to raise funds for camera-based products.
Outlook
Advances in edge computing and sensor miniaturisation are making it feasible to process images directly on the device, reducing bandwidth requirements and enabling real-time analytics, even in areas with limited connectivity. Advanced computer vision models are already enabling real-time remote monitoring of plant growth, crop and livestock disease detection. One interviewee noted they are already working to embed AI models directly into their cameras using energy-efficient microchips, a move expected to further differentiate their solution and lower operational costs. Partnerships with major input providers and research institutions will expand the reach and credibility of computer vision platforms.
There is a consensus that as more data are collected and models are refined, the accuracy and utility of computer vision systems will improve, making them indispensable tools for AI-enabled precision agriculture. Interviewees highlighted the potential of AI-powered chatbots and digital agronomists to provide personalised, context-aware recommendations to farmers, integrating local data from weather stations, soil sensors and in-field cameras. Interviewees anticipate that future developments will include integration with LLMs for personalised agronomic advice, broader use of subscription-based analytics platforms and greater interoperability with other digital farm management tools. One interviewee projected that more farmers will use AI-enabled mobile apps to manage farms through audio commands, set up irrigation systems and perform burdensome tasks on livestock and mixed farms. However, achieving these goals will require continued investment in infrastructure, data sharing and regulatory harmonisation, as well as targeted support for demonstration projects and farmer training.
Key recommendations
Support development and deployment of affordable, scalable camera and sensor hardware: increase investment in hardware, boosting computer vision innovations that are crucial to monitor crop and soil health.
Invest in digital infrastructure: expand broadband and edge-computing resources to enable real‑time AI analytics on farms and livestock farms, especially in underserved rural areas.
Support education and trust-building: more investment in farmer education projects, training courses and peer-to-peer workshops is needed. These spaces should showcase new technologies and equipment, allowing farmers to share and compare their experiences. This should lower adoption barriers and build confidence in AI solutions.
Facilitate access to high-quality datasets: support the creation and sharing of open high-quality, big datasets to better detect crop, soil and livestock health.
Address farmers’ concerns regarding data ownership and privacy: promote mechanisms for agricultural data sharing so farmers can retain control and transparency over their data on crops, soil and livestock.
Encourage stakeholder collaboration: foster partnerships among technology providers, farmer organisations and policymakers to co-develop AI solutions tailored to local needs. One interviewee expressed the desire to leverage farmers’ associations for easier access to target farmers.
Key recommendations to enhance AI uptake in agriculture in the European Union
Data availability and access
Invest in open, high-quality datasets: support public collection and dissemination of soil, weather and crop performance data to lower entry barriers and spur innovation, including development of AI models tailored to the European Union’s needs.
Safeguard farmers’ control over agricultural data: provide sectoral-specific guidance on data sharing through advancement of responsible data-sharing frameworks.
Increase awareness of and stakeholder engagement in the Common European Agricultural Data Space (CEADS): promote and disseminate open agricultural data spaces to enhance accessibility of high-quality, field-level data to farmers and start-ups.
Promote standards to reduce fragmentation: encourage use of open data formats, application programming interfaces and protocols across platforms, equipment and systems.
Infrastructure and connectivity
Expand digital infrastructure: improve broadband connectivity, cloud access and affordability, and edge- computing capacity to support real-time AI analytics, particularly in underserved rural areas and by small and medium-sized farms.
Regulatory and policy frameworks
Adopt a comprehensive EU strategy on agricultural digitalisation: integrate funding, regulation, infrastructure and skills development.
Clarify regulatory requirements for the sector: provide specific guidance to facilitate compliance, particularly for start-ups, and small and medium-sized enterprises (SMEs).
Provide guidance on the interplay and application of the EU AI Act and Machinery Regulation: clarify how AI regulations apply to agricultural machinery.
Skills, trust and collaboration
Make AI accessible through user-centred design: develop intuitive interfaces and local language options, especially for older or less tech-savvy farmers.
Share best practices and success stories: leverage multistakeholder platforms and farmers’ associations to strengthen farm advisory services and co‑operatives; promote demonstration projects and peer-to-peer learning to build trust among EU farmers and share information on successful use cases.
Invest in digital skills and training: deliver hands-on capacity-building, including workshops, demonstrations and peer learning among farmers. Support “farmer ambassadors” to lead by example.
Provide grants for start-ups and SMEs: develop robotics solutions tailored to European small farms and specialty crop farms.
Prioritise development and adoption of standards: facilitate data sharing and prevent monopolisation by large equipment manufacturers.
References
[80] 4Growth (2025), “Understanding the market to forecast future growth”, https://4growth-project.eu/ (accessed on 23 April 2025).
[65] Abioye, E. et al. (2022), “Precision irrigation management using machine learning and digital farming solutions”, AgriEngineering, Vol. 4/1, pp. 70-103, https://doi.org/10.3390/agriengineering4010006.
[145] Agelli, M. et al. (2024), “Unmanned ground vehicles for continuous crop monitoring in agriculture: Assessing the readiness of current ICT technology”, Machines, Vol. 12/11, https://doi.org/10.3390/machines12110750.
[53] AgriBIT (2020), “Intelligence system applied to precision farming”, https://h2020-agribit.eu/ (accessed on 12 February 2025).
[128] agROBOfood (2024), “agri-food & robots: Connecting robotic technologies with a dynamic agri-food network of DIHs”, https://agrobofood.eu/ (accessed on 18 June 2025).
[186] AgroNET (2025), “Poultry farms”, https://agronet.solutions/packages/poultry-farms/ (accessed on 12 April 2025).
[69] Agrow (2025), “Irrigation optimization”, https://agrowanalytics.com/en/water-solutions/smart-irrigation-system/ (accessed on 23 February 2025).
[143] Aguiar, A. et al. (2020), “Localization and mapping for robots in agriculture and forestry: A survey”, Robotics, Vol. 9/4, p. 97, https://doi.org/10.3390/robotics9040097.
[182] AI4SoilHealth (2025), “Testing on our pilot sites”, https://ai4soilhealth.eu/pilot-sites/ (accessed on 12 April 2025).
[40] AI4SoilHealth (2025), “Welcome to AI 4 soil health”, https://ai4soilhealth.eu/ (accessed on 23 April 2025).
[142] Alrowaily, M. et al. (2024), “Application of extreme machine learning for smart agricultural robots to reduce manoeuvering adaptability errors”, Alexandria Engineering Journal, Vol. 109, pp. 665-668, https://doi.org/10.1016/j.aej.2024.09.062.
[159] Amankulova, K., N. Farmonov and L. Mucsi (2022), “Time-series analysis of Sentinel-2 satellite images for sunflower yield forecasting”, SSRN Electronic Journal, https://doi.org/10.2139/ssrn.4021889.
[45] Amos, Z. (2024), “How AI is used for crop disease diagnosis”, https://www.agritechtomorrow.com/story/2024/05/how-ai-is-used-for-crop-disease-diagnosis/15554/ (accessed on 14 January 2025).
[147] Apple Machine Learning Research (2025), “FastVLM: Efficient vision encoding for vision language models”, https://machinelearning.apple.com/research/fast-vision-language-models (accessed on 23 April 2025).
[154] Asadollah, S., A. Jodar-Abellan and M. Pardo (2024), “Optimizing machine learning for agricultural productivity: A novel approach with RScv and remote sensing data over Europe”, Agricultural Systems, Vol. 218/103955, https://doi.org/10.1016/j.agsy.2024.103955.
[94] ATLAS (2025), “ATLAS: Agricultural interoperability and analysis system”, https://www.atlas-h2020.eu/ (accessed on 28 July 2025).
[72] AXIOBIT (2025), “InferenceAI: The general-purpose AI agent for enterprise automation and intelligence”, https://axiobit.com/ (accessed on 20 June 2025).
[126] BACCHUS (2022), “Use cases”, https://bacchus-project.eu/cases/ (accessed on 12 June 2025).
[146] Badgujara, C., A. Pouloseb and H. Gan (2024), “Agricultural object detection with You Look Only Once (YOLO) algorithm: A bibliometric and systematic literature review”, arXiv, Vol. 2401.10379, https://arxiv.org/abs/2401.10379.
[157] Bebie, M. and A. Kyparissis (2024), “A comparative analysis of machine learning and vegetation index-based modeling approaches for durum wheat yield assessment using Sentinel-2 imagery”, Proceedings, Remote Sensing for Agriculture, Ecosystems, and Hydrology XXVI, No. 1319107, https://doi.org/10.1117/12.3031645.
[46] Bivocom (2024), “How IoT and AI are revolutionizing fungal crop disease management for farmers”, https://www.bivocom.com/blog/iot-ai-fungal-crop-disease (accessed on 23 February 2025).
[52] Boechel, T. et al. (2022), “Prediction of harvest time of apple trees: An RNN-based approach”, Algorithms, Vol. 15/3, https://doi.org/10.3390/a15030095.
[19] Campi, M. et al. (2024), “The evolving profile of new entrants in agriculture and the role of digital technologies”, OECD Food, Agriculture and Fisheries Papers, No. 209, OECD Publishing, Paris, https://doi.org/10.1787/d15ea067-en.
[110] CEADS (2025), “The common European agricultural data space”, https://ceads.eu/ (accessed on 23 June 2025).
[28] Ceccarelli, T. et al. (2022), “Leveraging automation and digitalization for precision agriculture: Evidence from the case studies”, Background Paper for The State of Food and Agriculture 2022, Food and Agriculture Organization of the United Nations, Rome, https://doi.org/10.4060/cc2912en.
[7] CEMA (2025), “Expert workshop: AI in agriculture: Presentations AI workshop”, European Agricultural Machinery Association, https://digital-strategy.ec.europa.eu/en/library/expert-workshop-ai-agriculture (accessed on 20 June 2025).
[140] Chakraborty, S. et al. (2022), “A comprehensive review of path planning for agricultural ground robots”, Sustainability, Vol. 14/15, https://www.mdpi.com/2071-1050/14/15/9156#sec3-sustainability-14-09156.
[118] Cheng, C. et al. (2023), “Recent advancements in agriculture robots: Benefits and challenges”, Machines, Vol. 11/1, https://doi.org/10.3390/machines11010048.
[35] Chowdhury, S. et al. (2024), “Drivers of soil health across European Union – data from the literature review”, Data in Brief, Vol. 57, https://doi.org/10.1016/j.dib.2024.111064.
[135] CLAAS Harvest Centre (2025), “CLAAS and AMAZONE double down on autonomous technology”, https://claasharvestcentre.com/news-and-event/claas-and-amazone-double-down-on-autonomous-technology/ (accessed on 28 February 2025).
[180] Copernicus (2024), “Observer: Revealing hidden land patterns with AI and Copernicus”, https://www.copernicus.eu/en/news/news/observer-revealing-hidden-land-patterns-ai-and-copernicus (accessed on 5 February 2025).
[121] CORDIS (2025), “Accelerating the achievement of EU Green Deal Goals for pesticide and fertilizer reduction through AI, data and robotic technologies”, https://cordis.europa.eu/project/id/101070496 (accessed on 12 April 2025).
[129] CORDIS (2025), “agROBOfood: Business-oriented support to the European robotics and agri-food sector, towards a network of digital innovation hubs in robotics”, https://cordis.europa.eu/project/id/825395 (accessed on 18 June 2025).
[131] CORDIS (2025), “Cognitive robotic system for digitalized and networked (automated) insect farms”, https://cordis.europa.eu/project/id/101016953 (accessed on 18 June 2025).
[122] CORDIS (2025), “Flexible robots for intelligent automation of precision agriculture operations”, https://cordis.europa.eu/project/id/101017111 (accessed on 12 April 2025).
[133] CORDIS (2025), “Natural intelligence for robotic monitoring of habitats”, https://cordis.europa.eu/project/id/101016970 (accessed on 12 October 2025).
[124] CORDIS (2025), “Robots for protecting crops”, https://cordis.europa.eu/project/id/101016807 (accessed on 12 April 2025).
[132] CORDIS (2025), “Soft Milli-robots”, https://cordis.europa.eu/project/id/101016411 (accessed on 28 June 2025).
[125] CORDIS (2024), “MoBile Robotic PlAtforms for ACtive InspeCtion and Harvesting in AgricUltural AreaS”, https://cordis.europa.eu/project/id/871704 (accessed on 21 June 2025).
[127] CORDIS (2024), “RObotics for MIcrofarms”, https://cordis.europa.eu/project/id/773875 (accessed on 23 April 2025).
[60] CORDIS (2022), “VINEyardROBOT”, https://cordis.europa.eu/project/id/737669/fr (accessed on 12 April 2025).
[130] CoRoSect (2021), “CoRoSect pilots”, https://corosect.eu/pilots/ (accessed on 16 March 2025).
[57] Critchley, L. (2025), “Agricultural robots and the current state of smart farming”, https://www.mouser.com/blog/agricultural-robots-current-state-smart-farming (accessed on 30 March 2025).
[178] Crop Science UK (2024), “Cereals launch for ’magic’ pest monitoring tool”, https://cropscience.bayer.co.uk/insights/seed-establishment/cereals-launch-for-magic-pest-monitoring-tool (accessed on 23 March 2025).
[181] Cropler (2025), “Let your crops speak: AI-powered digital agronomical assistant, come heat or rain”, https://www.cropler.io/about-us (accessed on 25 April 2025).
[158] Darra, N. et al. (2023), “Can satellites predict yield? Ensemble machine learning and statistical analysis of Sentinel-2 imagery for processing tomato yield prediction”, Sensors, Vol. 23/5, https://doi.org/10.3390/s23052586.
[95] DEMETER (2025), “The DEMETER approach towards an agricultural interoperability space”, https://h2020-demeter.eu/demeter-approach-towards-an-agricultural-interoperability-space/ (accessed on 12 March 2025).
[71] Digital Green (2023), “Farmer.CHAT by Digital Green – making vetted farmer knowledge accessible”, https://digitalgreen.org/farmerchat/ (accessed on 12 February 2025).
[115] Directorate-General for Agriculture and Rural Development (2025), “The next chapter for the CAP”, https://agriculture.ec.europa.eu/media/news/next-chapter-cap-2025-07-17_en (accessed on 25 June 2025).
[189] DunavNet (2025), “From farm to fork: Digital farming”, https://dunavnet.eu/solutions/farming/ (accessed on 24 April 2025).
[190] DunavNet.AI (2025), “AI-driven solutions for a smarter future”, https://wedoai.eu/ (accessed on 18 March 2025).
[136] EASA (2025), “Drone regulatory system”, European Aviation Safety Agency, https://www.easa.europa.eu/en/domains/drones-air-mobility/drones-air-mobility-landscape/Understanding-European-Drone-Regulations-and-the-Aviation-Regulatory-System (accessed on 12 March 2025).
[139] EASA (2023), “ED Decision 2023/012/R”, European Aviation Safety Agency, https://www.easa.europa.eu/en/document-library/agency-decisions/ed-decision-2023012r (accessed on 23 April 2025).
[92] EIP-AGRI (2025), “European CAP network”, https://ec.europa.eu/eip/agriculture/en/european-cap-network.html (accessed on 12 April 2025).
[194] EIP-AGRI (2018), “Precision farming”, https://ec.europa.eu/eip/agriculture/en/digitising-agriculture/developing-digital-technologies/precision-farming-0.html (accessed on 4 March 2025).
[27] Elbasi, E. and N. Mostafa (2023), “Artificial intelligence technology in the agricultural sector: A systematic review”, IEEE Access, Vol. 11/1, pp. 171-202, https://doi.org/10.1109/access.2022.3232485.
[191] ENRD (2022), “New knowhow for the farm sustainability tool for nutrients”, European Network for Rural Development, https://ec.europa.eu/enrd/news-events/news/new-knowhow-farm-sustainability-tool-nutrients_en.html#:~:text=This%20involves%20generating%20new%20knowhow,of%20Member%20States'%20agricultural%20sector.&text=Source:,ENRD%20Contact%20Point.
[29] EPRS (2023), “Artificial intelligence in the agri-food sector: Applications, risks and impacts”, European Parliamentary Research Service, https://www.europarl.europa.eu/RegData/etudes/STUD/2023/734711/EPRS_STU(2023)734711_EN.pdf.
[68] ESA (2025), “Satellite irrigation platform”, European Space Agency, https://incubed.esa.int/portfolio/irreo-platform/ (accessed on 25 April 2025).
[50] ESA (2019), “YIPEEO: Yield prediction and estimation from Earth observation”, European Space Agency, https://eo4society.esa.int/projects/yipeeo/#:~:text=novel%20machine%2Dlearning%20techniques (accessed on 12 February 2025).
[162] Essenfelder, A., A. Toreti and L. Seguini (2025), “Expert-driven explainable artificial intelligence models can detect multiple climate hazards relevant for agriculture”, Communications Earth & Environment, Vol. 6/207, https://doi.org/10.1038/s43247-024-01987-3.
[93] EU CAP Network (2025), “Innovation, knowledge exchange and EIP-AGRI”, https://eu-cap-network.ec.europa.eu/support/innovation-knowledge-exchange-eip-agri_en (accessed on 12 June 2025).
[3] European Commission (2025), “A vision for agriculture and good shaping together an attractive farming and agri-food sector for future generations”, COM(2025) 75 final, European Commission, Brussels, https://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:52025DC0075.
[167] European Commission (2025), “Andalucía agrotech digital innovation hub”, https://european-digital-innovation-hubs.ec.europa.eu/edih-catalogue/andalucia-agrotech-digital-innovation-hub (accessed on 18 June 2025).
[99] European Commission (2025), “Annex to the Commission implementing decision on the financing of the Digital Europe programme and the adoption of the multiannual work programme for 2025-2027”, https://ec.europa.eu/newsroom/dae/redirection/document/114219 (accessed on 19 July 2025).
[88] European Commission (2025), “Commission presents its roadmap for a thriving EU farming and agri-food sector”, https://ec.europa.eu/commission/presscorner/detail/en/ip_25_530 (accessed on 23 July 2025).
[90] European Commission (2025), “Common agricultural policy funds”, https://agriculture.ec.europa.eu/common-agricultural-policy/financing-cap/cap-funds_en (accessed on 23 April 2025).
[62] European Commission (2025), “Demonstration of an agronomy-integrated see & spray technology for a more sustainable use of pesticide in agriculture”, https://webgate.ec.europa.eu/life/publicWebsite/project/LIFE20-ENV-DE-000650/demonstration-of-an-agronomy-integrated-see-spray-technology-for-a-more-sustainable-use-of-pesticide-in-agriculture (accessed on 12 April 2025).
[102] European Commission (2025), “DIGITAL success stories – AI testing and experimentation facilities: AgrifoodTEF”, https://digital-strategy.ec.europa.eu/en/factpages/digital-success-stories-ai-testing-and-experimentation-facilities-agrifoodtef (accessed on 23 March 2025).
[91] European Commission (2025), “Digitalisation of agriculture and rural areas in the EU”, https://agriculture.ec.europa.eu/overview-vision-agriculture-food/digitalisation_en (accessed on 23 July 2025).
[5] European Commission (2025), “EU agri-food exports reach record levels of €235.4 billion in 2024”, https://agriculture.ec.europa.eu/media/news/eu-agri-food-exports-reach-record-levels-eu2354-billion-2024-2025-04-08_en (accessed on 23 July 2025).
[6] European Commission (2025), Monitoring EU Agri-food Trade: Developments in 2024, European Commission, Brussels, https://agriculture.ec.europa.eu/document/download/8e6be7cf-e1d2-48fd-9114-90c448e09643_en?filename=monitoring-agri-food-trade_dec2024_en.pdf.
[54] European Commission (2024), “Artificial intelliGence applied to pRecision farmIng By the use of GNSS and integrated technologies”, https://cordis.europa.eu/article/id/454267-improving-europe-s-precision-agriculture-with-ai (accessed on 18 March 2025).
[104] European Commission (2024), “Blueprint proposal for the Common European agricultural data space”, https://digital-strategy.ec.europa.eu/en/library/blueprint-proposal-common-european-agricultural-data-space (accessed on 21 July 2025).
[105] European Commission (2024), “Data Act”, https://digital-strategy.ec.europa.eu/en/policies/data-act (accessed on 12 May 2025).
[98] European Commission (2024), “Digital agriculture and forestry: Understanding the market to forecast and support future growth”, https://cordis.europa.eu/project/id/101134855 (accessed on 24 February 2025).
[18] European Commission (2024), EU Agricultural Outlook 2023-35, European Commission, Brussels, https://agriculture.ec.europa.eu/system/files/2024-01/agricultural-outlook-2023-report_en_0.pdf.
[106] European Commission (2024), “European Data Governance Act”, https://digital-strategy.ec.europa.eu/en/policies/data-governance-act (accessed on 23 February 2025).
[67] European Commission (2024), Green dEal cOmpliant iRriGation Increasing Europe’s Agriculture resilience to drought, European Commission, Brussels, https://doi.org/10.3030/101181572.
[103] European Commission (2024), Policy brief – Rolling out the Common European Agricultural Data Space, https://digital-strategy.ec.europa.eu/en/news/policy-brief-rolling-out-common-european-agricultural-data-space.
[87] European Commission (2024), “The digitalisation of the European agricultural sector”, https://digital-strategy.ec.europa.eu/en/policies/digitalisation-agriculture (accessed on 12 April 2025).
[96] European Commission (2023), “Digitalisation of the European agricultural sector: Activities in Horizon 2020”, https://digital-strategy.ec.europa.eu/en/policies/digitalisation-agriculture-horizon-2020 (accessed on 12 April 2025).
[100] European Commission (2023), “eDelivery funding opportunities under CEF and DEP”, https://ec.europa.eu/digital-building-blocks/sites/display/DIGITAL/2023/01/12/eDelivery+funding+opportunities+under+CEF+and+DEP (accessed on 23 May 2025).
[89] European Commission (2023), “The common agricultural policy at a glance”, https://agriculture.ec.europa.eu/common-agricultural-policy/cap-overview/cap-glance_en (accessed on 5 February 2025).
[41] European Commission (2022), “AI4SoilHealth: Accelerating collection and use of soil health information using AI technology to support the soil deal for Europe and EU Soil Observatory”, https://cordis.europa.eu/project/id/101086179 (accessed on 18 April 2025).
[42] European Commission (2018), “Intelligent ‘health-check’ for farms goes head to head with crop diseases”, https://cordis.europa.eu/article/id/229064-intelligent-healthcheck-for-farms-goes-head-to-head-with-crop-diseases (accessed on 21 April 2025).
[20] European Commission (2017), EU Agricultural Outlook: For the Agricultural Markets and Income, 2017-2030, European Commission, Brussels, https://agriculture.ec.europa.eu/system/files/2018-07/agricultural-outlook-2017-30_en_0.pdf.
[108] European Union (2024), “Rural pact community platform”, https://ruralpact.rural-vision.europa.eu/news/advancing-digitalisation-rural-areas-new-position-paper_en (accessed on 23 March 2025).
[138] European Union (2019), “Directive 2009/128/EC of the European Parliament and of the Council of 21 October 2009 establishing a framework for Community action to achieve the sustainable use of pesticides (Text with EEA relevance)”, https://eur-lex.europa.eu/eli/dir/2009/128/oj/eng (accessed on 12 February 2025).
[2] Eurostat (2025), “Economic accounts for agriculture – values at current prices”, https://ec.europa.eu/eurostat/databrowser/view/aact_eaa01/default/bar?lang=en (accessed on 12 October 2025).
[192] Eurostat (2025), “Glossary: Farm typology”, https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Glossary:Farm_typology (accessed on 23 April 2025).
[8] Eurostat (2025), “National accounts and GDP”, https://ec.europa.eu/eurostat/statistics-explained/index.php?title=National_accounts_and_GDP#Developments_for_GDP_in_the_EU:_the_rebound_observed_in_2021_continued_in_2022_and_2023.2C_but_was_progressively_more_subdued (accessed on 23 September 2025).
[1] Eurostat (2025), “Statistics explained: Performance of the agricultural sector”, https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Performance_of_the_agricultural_sector#Resource_performance_of_the_agricultural_sector. (accessed on 12 October 2025).
[11] Eurostat (2023), “EU farms: 5.3 million fewer in 2020 than in 2005”, eurostat, https://ec.europa.eu/eurostat/fr/web/products-eurostat-news/-/ddn-20230403-2 (accessed on 23 March 2025).
[13] Eurostat (2023), “Key figures on the European food chain 2023 edition”, https://ec.europa.eu/eurostat/documents/15216629/18054337/KS-FK-23-001-EN-N.pdf/048e130f-79fa-e870-6c46-d80c9408620b?version=7.0&t=1707290893751 (accessed on 12 Februay 2025).
[17] Eurostat (2022), “Statistics explained: Farmers and the agricultural labour force – statistics”, https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Farmers_and_the_agricultural_labour_force_-_statistics (accessed on 12 April 2025).
[12] Eurostat (2022), “Statistics explained: Farms and farmland in the European Union”, https://ec.europa.eu/eurostat/statistics-explained/index.php?title=Farms_and_farmland_in_the_European_Union_-_statistics&oldid=656883 (accessed on 13 April 2025).
[9] FAOSTAT (2023), Commodities by Regions, (dataset), https://www.fao.org/faostat/en/#rankings/commodities_by_regions_exports (accessed on 23 April 2025).
[10] FAOSTAT (2023), Value of Agricultural Production, (dataset), http://www.fao.org/faostat/en/#data/QV (accessed on 12 April 2025).
[84] Farm Connexion (2023), “CNH Industrial acquiert la start-up Augmenta”, [CNH Industrial acquires the startup Augmenta], https://www.farm-connexion.com/2023/03/13/cnh-industrial-acquiert-la-start-up-augmenta/ (accessed on 20 January 2025).
[85] FarmDroid (2025), “Automate your agriculture”, https://farmdroid.com/ (accessed on 23 April 2025).
[55] Farmonaut (2025), “Revolutionizing European agriculture: How AI-powered solutions are optimizing pesticide use and boosting sustainable farming”, https://farmonaut.com/europe/ai-powered-solutions-boost-sustainable-farming-in-europe (accessed on 23 March 2025).
[113] Farmtopia (2023), “AI in farming: Smarter decisions for EU growers”, https://farmtopia.eu/ai-in-farming-smarter-decisions-for-eu-growers/ (accessed on 24 June 2025).
[116] Farmtopia (2023), “Farmtopia: European project paving the way for digital farming accessible to all”, https://farmtopia.eu/farmtopia-european-project-paving-the-way-for-digital-farming-accessible-to-all/ (accessed on 23 April 2025).
[173] Farooq, H., R. Mazhar and S. Jan (2025), “Mitigating climate change impacts on agriculture through AI-driven crop improvement”, MOJ Biology and Medicine, Vol. 10/1, https://doi.org/10.15406/mojbm.2025.10.00237.
[123] Flexigrobots (2025), “Flexigrobots”, https://flexigrobots-h2020.eu/ (accessed on 20 May 2025).
[36] Forrester, N. (2025), “Healthy soil is the hidden ingredient – here’s how I’m fighting to save it”, Nature Spotlight: Spain, https://doi.org/10.1038/d41586-025-01026-x.
[109] Gao, Y. et al. (2024), “Security threats to agricultural artificial intelligence: Position and perspective”, Computers and Electronics in Agriculture, Vol. 227/1, https://doi.org/10.1016/j.compag.2024.109557.
[75] Garske, B., A. Bau and F. Ekardt (2021), “Digitalization and AI in European sgriculture: A strategy or achieving climate and biodiversity targets?”, Sustainability, Vol. 13/9, p. 4652, https://doi.org/10.3390/su.13094652.
[117] Gebresenbet, G. et al. (2023), “A concept for application of integrated digital technologies to enhance future smart agricultural systems”, Smart Agricultural Technology, Vol. 5, p. 100255, https://doi.org/10.1016/j.atech.2023.100255.
[163] GeoPard (2025), “Independent precision agriculture platform”, https://geopard.tech/ (accessed on 24 April 2025).
[38] Gill, Y. (2025), “Using soil mapping to enhance sustainable agriculture and boost productivity”, AgriTech Tomorrow, https://www.agritechtomorrow.com/story/2025/02/using-soil-mapping-to-enhance-sustainable-agriculture-and-boost-productivity-/16365/ (accessed on 18 June 2025).
[49] Global Change Research Institute (2021), “YIPEEO”, https://yipeeo.czechglobe.cz/#activities (accessed on 12 April 2025).
[114] Goller, M., C. Caruso and C. Harteis (2021), “Digitalisation in agriculture: Knowledge and learning requirements of German dairy farmers”, International Journal for Research in Vocational Education and Training, Vol. 8/2, pp. 208-223, https://doi.org/10.13152/ijrvet.8.2.4.
[187] Gray, R. (2020), “How crop and animal sensors are making farming smarter”, Horizon, https://projects.research-and-innovation.ec.europa.eu/en/horizon-magazine/how-crop-and-animal-sensors-are-making-farming-smarter (accessed on 25 January 2025).
[164] Gupta, D. et al. (2024), “AI-powered revolution in plant sciences: Advancements, applications, and challenges for sustainable agriculture and food security”, Exploration of Foods and Foodomics, Vol. 2, pp. 443-459, https://doi.org/10.37349/eff.2024.00045.
[101] HaDEA (2025), “5G and edge cloud for smart communities”, European Health and Digital Executive Agency, https://hadea.ec.europa.eu/programmes/connecting-europe-facility/about/cef-digital-5g-and-edge-cloud-smart-communities_en (accessed on 30 Septeber 2025).
[155] Hasan, M. et al. (2023), “Ensemble machine learning-based recommendation system for effective prediction of suitable agricultural crop cultivation”, Frontiers in Plant Science, Vol. 14, https://doi.org/10.3389/fpls.2023.1234555.
[168] Hnatiuc, M. et al. (2023), “Intelligent grapevine disease detection using IoT sensor network”, Bioengineering, Vol. 10/9, https://doi.org/10.3390/bioengineering10091021.
[44] Hossain, M. et al. (2024), “Plant disease dynamics in a changing climate: Impacts, molecular mechanisms, and climate-informed strategies for sustainable management”, Discover Agriculture, Vol. 2/132, https://doi.org/10.1007/s44279-024-00144-w.
[171] ICRISAT (2023), “Unleashing the power of innovation at ICRISAT: Advancing crop breeding with breakthrough technologies”, International Crops Research Institute for the Semi-Arid Tropics, https://pressroom.icrisat.org/unleashing-the-power-of-innovation-at-icrisat-advancing-crop-breeding-with-breakthrough-technologies (accessed on 23 April 2025).
[193] International Society of Precision Agriculture (2024), “Precision Agriculture Definition”, https://www.ispag.org/resources/definition (accessed on 12 March 2025).
[161] Joint Research Centre (2025), “AI tool to help detect growing climate hazards for EU agriculture”, https://joint-research-centre.ec.europa.eu/jrc-news-and-updates/ai-tool-help-detect-growing-climate-hazards-eu-agriculture-2025-04-07_en (accessed on 23 July 2025).
[34] Joint Research Centre (2023), “A new tool maps the state of soil health across Europe”, https://joint-research-centre.ec.europa.eu/jrc-news-and-updates/new-tool-maps-state-soil-health-across-europe-2023-03-13_en (accessed on 13 March 2025).
[172] Khan, M. et al. (2022), “Applications of artificial intelligence in climate-resilient smart-crop breeding”, International Journal of Molecular Sciences, Vol. 23/19, p. 11156, https://doi.org/10.3390/ijms231911156.
[48] Klompenburg, T., A. Kassahun and C. Catal (2020), “Crop yield prediction using machine learning: A systematic literature review”, Computers and Electronics in Agriculture, Vol. 177/10, https://doi.org/10.1016/j.compag.2020.105709.
[77] Kondratieva, N. (2021), “EU agricultural digitalization decalogue”, Herald of the Russian Academy of Sciences, Vol. 91/6, pp. 736-742, https://doi.org/10.1134/S1019331621060150.
[148] Korolkova, K. (2024), “How AI Is taking green-on-green spray technology to new levels of productivity and sustainability”, https://www.globalagtechinitiative.com/in-field-technologies/how-ai-is-taking-green-on-green-spray-technology-to-new-levels-of-productivity-and-sustainability/ (accessed on 20 January 2025).
[56] Kugler, L. (2022), “Artificial intelligence, machine learning, and the fight against world hunger”, Communications of the ACM, Vol. 65/2, pp. 17-19, https://doi.org/10.1145/3503779.
[26] Lin, Y. et al. (2019), “SensorTalk: An IoT device failure detection and calibration mechanism for smart farming”, Sensors, Vol. 19/21, p. 4788, https://doi.org/10.3390/s19214788.
[32] McBride, M. (2022), “Estimating soil chemical properties by diffuse reflectance spectroscopy: Promise versus reality”, European Journal of Soil Science, Vol. 73/1, https://doi.org/10.1111/ejss.13192.
[149] McFadden, J., F. Casalini and J. Antón (2022), “Policies to bolster trust in agricultural digitalisation: Issues note”, OECD Food, Agriculture and Fisheries Papers, No. 175, OECD Publishing, Paris, https://doi.org/10.1787/5a89a749-en.
[74] McFadden, J. et al. (2022), “The digitalisation of agriculture: A literature review and emerging policy issues”, OECD Food, Agriculture and Fisheries Papers, No. 176, OECD Publishing, Paris, https://doi.org/10.1787/285cc27d-en.
[79] McFadden, J. et al. (2024), “Global adoption of precision agriculture: An update on trends and emerging technologies”, Proceedings of the 16th International Conference on Precision Agriculture, https://mssoy.org/sites/default/files/2025-01/McFADDEN%20%20PA%20PROC%202024.pdf.
[70] McKinsey and Company (2024), “From bytes to bushels: How gen AI can shape the future of agriculture”, https://www.mckinsey.com/industries/agriculture/our-insights/from-bytes-to-bushels-how-gen-ai-can-shape-the-future-of-agriculture (accessed on 12 March 2025).
[134] Monarch (2024), “Monarch announces establishment of Monarch Tractor Europe N.V.”, https://www.monarchtractor.com/news/monarch-announces-establishment-of-monarch-tractor-europe (accessed on 21 June 2025).
[119] Mordor Intelligence (2024), “Europe agricultural robots and mechatronics market size & share analysis – growth trends & forecasts (2025-2030)”, https://www.mordorintelligence.com/industry-reports/europe-agricultural-robots-and-mechatronics-market (accessed on 5 April 2025).
[120] Morrison, O. (2025), “’The goal is not to remove farmers from the farming process’: CNH talks autonomous vehicle innovation”, AgTech Navigator, https://www.agtechnavigator.com/Article/2025/03/24/cnh-industrial-discusses-autonomous-vehicle-advancements/ (accessed on 20 June 2025).
[137] Moysiadis, V. et al. (2021), “Smart farming in Europe”, Computer Science Review, Vol. 39/100345, https://doi.org/10.1016/j.cosrev.2020.100345.
[76] Munniunker, S., L. Nel and D. Diederichs (2022), “The impact of artificial intelligence on agricultural labour n Europe”, Journal of Agricultural Informatics, Vol. 13/1, https://doi.org/10.17700/jai.2022.13.1.638.
[153] Murel, J. and E. Kavlakoglu (2024), “What is ensemble learning?”, https://www.ibm.com/think/topics/ensemble-learning (accessed on 15 January 2025).
[83] Naïo Technologies (2025), “Delivering agricultural robotics solutions – Pioneer in agricultural robotics solutions, Naïo Technologies provides robots for easier farming”, https://www.naio-technologies.com/en/home/ (accessed on 12 March 2025).
[97] NCP Portal (2025), “Agriculture of data partnership”, https://horizoneuropencpportal.eu/ncp-networks/cluster-6/agriculture-data-partnership (accessed on 23 April 2025).
[183] NEIKER (2023), “Artificial intelligence to know the health of Europe’s soils”, https://neiker.eus/en/news/artificial-intelligence-to-know-the-health-of-europes-soils/ (accessed on 25 April 2025).
[63] Ngo, H. et al. (2019), “Weeds detection in agricultural fields using convolutional neural network”, International Journal of Innovative Technology and Exploring Engineering, Vol. 8/11, https://doi.org/10.35940/ijitee.K1327.0981119.
[150] OECD (2025), “Tractors”, https://www.oecd.org/en/topics/tractors.html (accessed on 12 June 2025).
[107] OECD (2023), Policies for the Future of Farming and Food in Spain, OECD Agriculture and Food Policy Reviews, OECD Publishing, Paris, https://doi.org/10.1787/a93d26be-en.
[21] OECD (2023), Policies for the Future of Farming and Food in the European Union, OECD Agriculture and Food Policy Reviews, OECD Publishing, Paris, https://doi.org/10.1787/32810cf6-en.
[22] OECD (2023), Policies for the Future of Farming and Food in the European Union, OECD Agriculture and Food Policy Reviews, OECD Publishing, Paris, https://doi.org/10.1787/32810cf6-en.
[23] OECD (2023), Policies for the Future of Farming and Food in the European Union, OECD Agriculture and Food Policy Reviews, OECD Publishing, Paris, https://doi.org/10.1787/32810cf6-en.
[73] OECD (2022), “OECD Framework for the Classification of AI systems”, OECD Digital Economy Papers, No. 323, OECD Publishing, Paris, https://doi.org/10.1787/cb6d9eca-en.
[111] OECD (2021), “Mapping data portability initiatives, opportunities and challenges”, OECD Digital Economy Papers, No. 321, OECD Publishing, Paris, https://doi.org/10.1787/a6edfab2-en.
[86] OECD.AI (2025), Total VC Investments in AI by Country and Industry, data from Prequin, (dataset), https://oecd.ai/en/data?selectedArea=investments-in-ai-and-data&selectedVisualization=total-vc-investments-in-ai-by-country-and-industry (accessed on 16 October 2025).
[4] OECD/FAO (2024), OECD-FAO Agricultural Outlook 2024-2033, OECD Publishing, Paris/Food and Agriculture Organization of the United Nations, Rome, https://doi.org/10.1787/4c5d2cfb-en.
[184] Orgiazzi, A. et al. (2017), “LUCAS Soil, the largest expandable soil dataset for Europe: A review”, European Journal of Soil Science, Vol. 69/1, pp. 140-153, https://doi.org/10.1111/ejss.12499.
[82] Overstory (2025), “AI-powered vegetation intelligence for a more resilient grid”, https://www.overstory.com/ (accessed on 18 April 2025).
[144] Padhiary, M. et al. (2024), “Enhancing precision agriculture: A comprehensive review of machine learning and AI vision applications in all-terrain vehicle for farm automation”, Smart Agricultural Technology, Vol. 8/100483, https://doi.org/10.1016/j.atech.2024.100483.
[58] Pathan, M. et al. (2020), “Artificial cognition of applications in smart agriculture: A comprehensive review”, Artificial Intelligence in Agriculture, Vol. 4, pp. 81-95, https://doi.org/10.1016/j.aiia.2020.06.001.
[25] Patricio, I. and R. Rieder (2018), “Computer vision and artificial intelligence in precision agriculture for grain crops: A systematic review”, Computers and Electronics in Agriculture, Vol. 153, pp. 69-81, https://doi.org/10.1016/j.compag.2018.08.001.
[152] Paudel, D. et al. (2021), “Machine learning for large-scale crop yield forecasting”, Agricultural Systems, Vol. 187, p. 103016, https://doi.org/10.1016/j.agsy.2020.103016.
[151] Paudel, D. et al. (2022), “Machine learning for regional crop yield forecasting in Europe”, Field Crops Research, Vol. 276, p. 108377, https://doi.org/10.1016/j.fcr.2021.108377.
[160] Pejak, B. et al. (2022), “Soya yield prediction on a within-field scale using machine learning models trained on Sentinel-2 and soil data”, Remote Sensing, Vol. 14/9, p. 2256, https://doi.org/10.3390/rs14092256.
[170] Petrović, B. et al. (2024), “Application of precision agriculture technologies in Central Europe-review”, Journal of Agriculture and Food Research, Vol. 15, https://doi.org/10.1016/j.jafr.2024.101048.
[51] Ramos, M. et al. (2025), “Improving early prediction of crop yield in Spanish olive groves using satellite imagery and machine learning”, PLOS One, Vol. 20/1, https://doi.org/10.1371/journal.pone.0311530.
[112] Reimsbach-Kounatze, C. and A. Molnar (2024), “The impact of data portability on user empowerment, innovation, and competition”, OECD Going Digital Toolkit Notes, No. 25, OECD Publishing, Paris, https://doi.org/10.1787/319f420f-en.
[24] Rejeb, A. et al. (2022), “Examining the interplay between artificial intelligence and the agri-food industry”, Artificial Intelligence in Agriculture, Vol. 6/21, https://doi.org/10.1016/j.aiia.2022.08.002.
[43] Ristaino, J. et al. (2021), “The persistent threat of emerging plant disease pandemics to global food security”, PNAS, Vol. 118/23, p. e2022239118, https://doi.org/10.1073/pnas.2022239118.
[61] Robs4Crops (n.d.), Robs4Crops: Cultivating Tomorrow’s Agriculture With Robotics, https://www.robs4crops.eu/robs4crops-cultivating-tomorrows-agriculture-with-robotics/#:~:text=The%20Robs4Crops%20project%20leveraged%20cutting,sugar%20beet%20and%20pumpkin%20cultivation. (accessed on 10 January 2025).
[16] Román, A. (2024), “Ukrainian agriculture: From Russian invasion to EU integration”, Briefing, European Parliamentary Research Service, https://www.europarl.europa.eu/RegData/etudes/BRIE/2024/760432/EPRS_BRI(2024)760432_EN.pdf.
[175] Ryan, M. et al. (2024), “The future of agricultural data-sharing policy in Europe: Stakeholder insights on the EU Code of Conduct”, Humanities and Social Sciences Communications, Vol. 11, https://doi.org/10.1057/s41599-024-03710-1.
[156] Segarra, J., J. Araus and S. Kefauver (2022), “Farming and Earth observation: Sentinel-2 data to estimate within-field wheat grain yield”, International Journal of Applied Earth Observation and Geoinformation, Vol. 107, https://doi.org/10.1016/j.jag.2022.102697.
[185] Shokri, N. (2024), “Integrating LUCAS data with AI-driven models for predicting soil salinization across the EU”, https://events.interpore.org/event/46/contributions/6492/ (accessed on 15 January 2025).
[166] SmartAgriHubs (2022), “Sensoring and AI algorithms for early crop disease detection – SAIA”, https://www.smartagrihubs.eu/flagship-innovation-experiment/21-fie-sensoring-and-ai-algorithms-for-early-detection-of-crop-diseases (accessed on 16 January 2025).
[188] SmartAgriHubs (2020), “Pig health assessment based on applied sensors – smart pig health”, https://www.smartagrihubs.eu/flagship-innovation-experiment/11-FIE-pig-health-assessment-based-on-applied-sensors (accessed on 12 February 2025).
[81] Source.ag (2025), “AI solutions for smarter greenhouse cultivation”, https://www.source.ag/ (accessed on 26 June 2025).
[78] Sparrow, R., M. Howard and C. Degeling (2021), “Managing the risks of artificial intelligence in agriculture”, NJAS: Impact in Agricultural and Life Sciences, Vol. 93/1, pp. 172-196, https://doi.org/10.1080/27685241.2021.2008777.
[15] State Statistics Service of Ukraine (2023), 2022 Agriculture of Ukraine, State Services of Ukraine, https://ukrstat.gov.ua/druk/publicat/kat_u/2023/zb/09/S_gos_22.pdf.
[47] Stelar (2024), “Key challenges in accurate crop yield prediction”, https://stelar-project.eu/key-challenges-in-accurate-crop-yield-prediction/ (accessed on 23 March 2025).
[176] STELLA (2025), “Our vision”, https://stella-pss.eu/stella-project/our-vision/ (accessed on 12 April 2025).
[177] STELLA (2025), “Use case pilots”, https://stella-pss.eu/use-case-pilots/ (accessed on 12 May 2025).
[141] Strodt, S. (2022), “Futuristic fields: Europe’s farm industry on cusp of robot revolution”, https://projects.research-and-innovation.ec.europa.eu/en/horizon-magazine/futuristic-fields-europes-farm-industry-cusp-robot-revolution (accessed on 18 January 2025).
[179] Superlinear (2025), “VITO partners with Superlinear to transform agricultural monitoring with AI”, https://superlinear.eu/impact/vito-partners-with-superlinear-to-transform-agricultural-monitoring-with-ai (accessed on 12 March 2025).
[66] Synelixis (2025), “GEORGIA (Green dEal cOmpliant iRriGation Increasing Europe’s Agriculture resilience to drought)”, https://synelixis.com/portfolios/georgia/ (accessed on 12 May 2025).
[64] Talaviya, T. et al. (2020), “Implementation of artificial intelligence in agriculture for optimisation of irrigation and application of pesticides and herbicides”, Artificial Intelligence in Agriculture, Vol. 4, pp. 58-73, https://doi.org/10.1016/j.aiia.2020.04.002.
[37] Tian, H. et al. (2019), “Computer vision technology in agricultural automation – a review”, Information Processing in Agriculture, Vol. 7/1, https://doi.org/10.1016/j.inpa.2019.09.006.
[169] Vanglund, I. (2021), “Using machine learning to detect grapevine disease in wine production”, Master’s thesis in Mathematical Sciences, Lund University, https://lup.lub.lu.se/luur/download?func=downloadFile&recordOId=9046108&fileOId=9046117.
[59] Vine Robots (2025), “Learn to make your own!”, https://www.vinerobots.org/ (accessed on 5 April 2025).
[165] Volta (2022), “Sensoring and AI algorithms for early crop disease detection – SAIA”, https://voltatv.com/smart-farming/SmartAgriHubs-FIEs/Sensoring-and-AI-Algorithms-for-Early-Crop-Disease-Detection-SAIA (accessed on 23 May 2025).
[39] Wadoux, A. (2025), “Artificial intelligence in soil science”, European Journal of Soil Science, Vol. 76/2, p. e70080, https://doi.org/10.1111/ejss.70080.
[31] Wadoux, A. et al. (2020), “A note on knowledge discovery and machine learning in digital soil mapping”, European Journal of Soil Science, Vol. 71/2, pp. 133-136, https://doi.org/10.1111/ejss.12909.
[33] Wakchaure, M., B. Patle and B. Mahindrakar (2023), “Application of AI techniques and robotics in agriculture: A review”, Artificial Intelligence in the Life Sciences, Vol. 3, https://doi.org/10.1016/j.ailsci.2023.100057.
[14] World Bank (2021), Agricultural Land, (dataset), https://data.worldbank.org/indicator/AG.LND.AGRI.K2?most_recent_value_desc=true (accessed on 12 February 2025).
[174] WUR (2023), “New institute to develop ’smart breeding’ method for more resilient and climate-adaptive agriculture”, https://www.wur.nl/en/research-results/research-institutes/plant-research/show-wpr/new-institute-to-develop-smart-breeding-method-for-more-resilient-and-climate-adaptive-agriculture.htm (accessed on 12 February 2025).
[30] Zhang, Q. (ed.) (2023), Artificial Intelligence in Agriculture, Springer, Cham, https://doi.org/10.1007/978-3-030-89123-7_183-2.
← 1. A farm in the European Union is classified into one of three types: a crop-specialist holding where crop production is the dominant activity; a livestock-specialist holding where livestock production dominates; or a mixed-farming holding where neither livestock nor crop production dominates (Eurostat, 2025[192]).
← 2. Farmers’ training level is categorised into three levels: practical agricultural experience (where the farm manager’s experience was acquired through practical work on an agricultural holding); basic agricultural training (if the manager took any training at a general agricultural college specialising in certain subjects); or full agricultural training (if the manager took any training course for at least two years full time after the end of compulsory education and graduated from an agricultural institution).
← 3. Precision farming focuses on real-time “observation, measurement and responses to the variability in crops, field and animals” to enhance farming practices and productivity and mitigate environmental impacts (EIP-AGRI, 2018[194]).
← 4. In 2019, the International Society of Precision Agriculture (2024[193]) adopted a definition of precision agriculture, which was then updated in January 2024: “Precision Agriculture is a management strategy that gathers, processes and analyzes temporal, spatial and individual plant and animal data and combines it with other information to support management decisions according to estimated variability for improved resource use efficiency, productivity, quality, profitability and sustainability of agricultural production”.
← 5. Open-source electronic prototyping platform enabling users to create interactive electronic objects.
← 6. Single-board computer with wireless LAN and Bluetooth connectivity.
← 7. FaST is a digital agriculture tool platform that provides recommendations on crop fertilisation through a nutrient management plan, analysing manually inputted data from farmers (ENRD, 2022[191]).