Research axises

Research axises

Digital Agriculture requires interdisciplinarity. Indeed, designing new tools and services for digital agriculture with the ambition of disseminating them to users requires not only knowledge of agricultural, technological and digital sciences, but also of social, management and economic sciences. The latter help to improve the success of innovation and understanding of its societal impacts. The scientific communities are organized around 6 research axes:

Axis 1: Impact of information and communication technologies on the rural world

Understanding the impact of ICT on rural societies

Understand the contribution of ICT to improving land and farm management.
Understand how new ICT-related services are changing the role of agricultural players, particularly in terms of farm advisory services.
ICT is profoundly changing our society, and is also expected to revolutionize the agricultural sector. Digital Agriculture can open up new ways of improving the management of farms, territories and agricultural sectors. How these ICTs will contribute to these dynamics of change at different scales is still under discussion. In particular, new services will be developed, and it is vital to analyze how innovations can contribute to organizational, economic and institutional change. The economic impact of such tools needs to be assessed in terms of productivity, competitiveness, territorial governance processes and the effectiveness of public policies.
Farms aiming for productivity gains can adopt precision farming technologies (sensors, automated data processing, decision support systems, etc.), while those practicing non-conventional agriculture can employ other digital technologies to communicate, access information and exchange. #DigitAg focuses on ICT adoption issues in agriculture.

ICTs enhance the creation of horizontal collaborations and the development of new networks. New services are being created and distributed by organizations of all shapes and sizes, including those outside agriculture. It is therefore essential to analyze the extent to which these innovations will contribute to organizational, economic and institutional change.

New relationships between farmers and cooperatives are emerging through the use of ICTs, as are new organizations and regulations at territorial level and in agricultural supply chains. Finally, the effects of such tools need to be assessed in terms of productivity, competitiveness and sustainability. Changes in governance processes in rural areas, and in the effectiveness of public policies linked to support, transparency, control and mobilization of the agricultural sector, also need to be studied.

What are the impacts of ICT-related innovations on practices, natural resource management, farm performance, sustainable agriculture, resilience, marginalization and employment? Do these new services foster the emergence of new collaborative patterns? Are digital technologies changing the relationship between farmers and farm advisory structures? Is there a danger of farmers losing autonomy in their decision-making? Are ICTs changing public-private relations?

Managers: Leïla Temri (Montpellier SupAgro), Jean-Marc Touzard (INRAE)
Scientic fileds:  Mainly economics, management and social sciences; secondarily agronomy (prospective knowledge of production systems), digital technologies

Axis 2:  Innovation in digital agriculture

ICT innovation development: technological, social and legal challenges

Objectives:  Understand how to successfully implement technical and organizational innovations in digital agriculture.
Address the legal and ethical issues surrounding the intellectual property of data and knowledge, and study the consequences for value sharing.

This line of research is dedicated to the social, legal and management issues associated with innovation in digital agriculture. A first scientific approach involves understanding how digital tools and services need to be designed to be adopted by farmers. This raises questions about design science, collective learning processes, innovation diffusion, etc. In addition, this line of research addresses ethical and legal issues concerning data and knowledge ownership. The #DigitAg project will contribute to furthering this research.
The development of innovations in the field of management is of great importance in digital agriculture, and the use, construction and dissemination of these innovations need to be studied.
The challenges of innovation construction arise with the involvement of users in the early stages of the design process. The co-design of decision-support tools with operators and regional players, as well as participatory workshops, are a valuable aid to the adaptation and dissemination of products and services that meet real user needs.Promoting participatory design and a variety of design modes requires the study of different technologies and the diversity of their uses.
In a context where innovation is increasingly open, new forms of innovation support organization need to be explored (innovation ecosystems, living laboratories, etc.).The development of these new services raises questions about operating methods, governance and financing (business models).

Next comes an analysis of the forms and determinants of innovation dissemination among farmers, companies, cooperatives and local players.

The acquisition, analysis and dissemination of data ultimately raises ethical and legal issues concerning intellectual property. Who has access to the data? Who controls and owns the data (public, private), and for what purposes? If intellectual property laws apply to databases, the question then becomes:Are databases free to use, and can the data be disseminated widely?

Managers: Sophie Mignon (Université de Montpellier), Danielle Galliano (INRAE)
Scientific fields:  Mainly management, law and social sciences; secondarily agronomy and digital technologies

Axis 3 : Sensors and data acquisition and management

Promoting the development of sensors and data acquisition systems, including crowdsourcing


Study and design sensors to meet detection challenges (field phenotyping, disease, plant/animal condition).
Develop "frugal" data acquisition using smartphones and satellite images.

Despite 30 years of research into sensors, agriculture still suffers from a lack of satisfactory measurement tools for certain parameters (study of diseases, plants, soil composition, etc.) at the right temporal and spatial resolutions. We need to improve access to high-resolution data and information on agricultural processes, by making better use of satellite images. We also need to promote low-cost innovation and the development of robust, easy-to-use sensors, including connected objects and devices linked to smartphones. Optimal implementation of these new technologies in real-life conditions (sensor networks, integrated and portable sensors with a focus on autonomy and cost) is the key to successful innovation, and requires innovation experts. With #DigitAg, all research on sensors and data acquisition will be supported by innovation experts. Data processing and related issues (uncertainties, validation of results, etc.) will also be studied.

Agriculture is characterized by strict constraints when it comes to data acquisition and sensor development: small markets, low margins, large areas to cover, high variability, complex environment, and a low to medium level of understanding of ICT by users. Sensors and data acquisition systems therefore need to be robust, low-maintenance, easy to use and offer satisfactory metrological performance. Sensors can be installed in a variety of configurations: portable, in the field (on animals, plants, in the soil, etc.), on-board agricultural equipment or aerial devices (drones, aircraft), or even on satellites (space observation). It must also be possible for farmers to analyze their own data, using smartphones or tablets. Today, sensor, data acquisition and processing technologies do not yet meet the needs of agriculture.

#DigitAg aims to fill this gap, in particular with regard to :

Assessing the condition of animals, plants and soil (physiology, composition, health);
Sensor networks (e.g. meteorological sensors for temperature, humidity, rain, hail, etc.);
On-board sensors on agricultural equipment, drones, etc.
Sensors linked to smartphones
Satellite image processing at different resolutions
The scientific questions linked to these shortcomings concern..:

A better understanding of light/matter interactions in optical sensors and radars
Applying Internet of Things concepts to agriculture
Designing low-cost, low-energy (including energy capture) microsystems for measuring and transferring data on physical parameters (temperature, hygrometry, etc.), for sustainable agriculture.
Ergonomics of data acquisition systems on smartphones, to encourage their adoption by farmers.

Managers: Ryad Bendoula (INRAE), Philippe Combette (Université de Montpellier)
Scientific fields:  Mainly physics, engineering sciences, optics, electronics, digital sciences; secondarily agronomy (physiology, modeling, indicators), social and management sciences (user-based design).

Axis 4: Information systems, data storage and transfer

Advancing the design of agricultural information systems

The main challenges facing data management are :

- Scalability (Big data, extended applications),
- Complexity (relevance, uncertainties, validity, multi-scale, etc.),
- Heterogeneity of data sources and semantics,
- Confidentiality and ethics (sensitive data, funding, surveys, etc.),
- Data flows and reproducibility.
- Given the need for interdisciplinarity and the contextual approach to data management, agricultural data must be informed (common vocabulary, ontologies, rules).

Managers: Maguelonne Teisseire (INRAE), Clément Jonquet (Université de Montpellier)
Scientific fields:  Mainly computer science; secondarily agronomy, management sciences

Axis 5: Data mining, data analysis, knowledge extraction


Design new data mining methods for agricultural Big Data and confidentiality requirements.
Develop visual and interactive methods for data analysis, designed for non-specialists.

The real potential residing in the growing masses of agricultural data (axes 3 and 4) is that they can be converted into usable knowledge and provide high value-added decision-support services for agriculture. At present, no single application effectively exploits the wealth of data with specific spatial and temporal characteristics. There is a need for an integrated approach that combines data collected by: on-farm monitoring systems (plants, animals, buildings), farmers, scientists and technicians on technical itineraries and crowd-testers.

At the same time, agricultural data are increasingly numerous, multi-scale (spatial, temporal), random, dynamic and heterogeneous. They can also be highly sensitive, requiring confidentiality protection. Unfortunately, no data mining technique can take all these aspects into account. The working hypothesis here is the construction of a methodology based on all the expertise, skills and knowledge in statistical data analysis, data mining and integration of imperfect knowledge developed by the #DigitAg teams for other fields (medicine). The proposed methods will be co-designed by players in the field (advisors, operators, analysts) and interactions will take place at various levels, whether to provide expertise, specify problems, or present results.

A major challenge will be to design interactive methods for non-specialists (visuals for analysis, explanation and justification). These new methods will lead to major advances in the practical exploitation of masses of agricultural data. A precise framework for assessing results will be put in place, based on statistical validation of models, agronomic expertise (INRA, ACTA) and collaboration with other players. Such a framework is highly innovative in terms of data mining and statistical analysis. Validating current results on an agricultural timescale (seasons, years) will enable us to continuously improve the proposed methods.

Managers: Alexandre Termier (INRIA, à droite sur l'image), Pascal Poncelet (Université de Montpellier)
Scientific fields:  Mainly data science, computer science; secondarily agronomy (knowledge of data types and quality, knowledge extraction), social sciences (user-centered design and visualization methods).

Axis 6: Modeling and simulation

Exploring new methods for integrating and qualifying models

Advance genotype-to-phenotype modeling through new methods of data integration and knowledge input.
Develop methods for integrating different types of information and knowledge (from data, experts, models).
Quantify model uncertainties in agriculture more precisely.

Over the years, dozens of agronomic models have been developed, some of which are already used by Decision Support Systems. However, new tools such as high-resolution detection (axis 3) and data-based knowledge (axis 5) are opening up new opportunities for major developments in model-building methods. The scientific objectives of this axis are to improve the applicability of dynamic models by coupling models of different natures and scales, using real-time data assimilation. The originality here lies in the variety of scales studied (from the plant organ to a restricted region) and in the interdisciplinary approach. The latter is central to :

-Coupling agronomy - ICT - innovation sciences to develop expert models (based on participatory science).

Models are an essential component of Digital Agriculture, as they can be used to transform data into information, diagnostics and agricultural advice. They form the basis of precision agriculture and breeding, as well as decision support systems. An overhaul of our production and impact assessment models is urgently needed: models need to better represent and take into account plant interactions with the environment, the effects of stress and the impact of extreme climatic events, but also the impact of alternative farm management strategies... Today, the main shortcoming of modeling and simulation is the lack of applicability of dynamic models. Our aim is to improve this by coupling different types of model, using real-time data assimilation at different time and space scales (from the field to the agricultural region).

First and foremost, we need to consider the advantages of genetics: genotype-to-phenotype models (the link between the plant's genes and its evolution in a real environment), which will be informed by the results of high-throughput phenotyping methods. To this end, modular modeling will be favored, so that models with explicit genetic information can be added and shared more easily.

Secondly, the use of new-generation sensors in agriculture (IoT, drones) creates the need for new methodologies (spatio-temporal data analysis including advanced statistical inference) for integrating data into models (including seasonal climate prediction) to produce more accurate, real-time predictive information. There is also a need to improve the development of decision-making and management models based on this type of data. Production models need to be coupled with economic models for strategic decision-making, and methods such as fuzzy logic, multi-objective optimization and argumentation will be required for multi-criteria evaluation.

In addition, for production system evaluation purposes, models need to be able to predict a wider range of responses (services) and trade-offs than is currently the case.

Finally, modeling research must also address cross-cutting issues such as quantifying model uncertainty and integrating models. Quantifying uncertainty is in fact in demand by decision-makers and public policy-makers.

Finally, models that integrate different sources of information (quantitative and qualitative) and knowledge (expert or data-driven) are required for decision-making.

Managers: Patrick Taillandier (INRAE), Pierre Martre (INRAE)
Scientific fields:  Strong interdisciplinarity between agricultural sciences, mathematics, artificial intelligence and computer science

Modification date : 08 February 2024 | Publication date : 08 August 2022 | Redactor : GL