The Ekinocs team at Inrae and the DIVA team at Télécom Paris are recruiting for two fully-funded PhD positions in visualization and HCI as part of a PEPR AgroEcoNum project (TwinFarms) for deploying digital twins at farm scale to promote agro-ecological innovation.
*** Doing a Master’s internship at Ekinocs/DIVA is recommended, though not mandatory for applying. Funding for two M2 internships is also available here ***
Context
The concept of digital twins (DTs), derived from engineering, has enabled improvements in many areas including for physical and chemical processes. In DT, there is a coexistence of a physical object and a digital object. The digital object is presented as a “copy” of the real object, hence the term “twin”.
With the increasing number of sensors in farms, the amount of data collected, combined with already acquired knowledge, could support the agroecological transition in the face of climate change. Although DT’s for agriculture are rare (for a review of digital twins in agriculture, see [6)], biological and social systems are beginning to see the birth of their digital twin.
The goal of the TwinFarm project is to empower farmers to make informed decisions by simulating outcomes in a virtual environment, thereby potentially reducing real-world risks and encouraging more experimental and innovative farming strategies.
The expected contributions of this digital twin are thus two fold: (1) to optimize management decisions of the real twin in real time based on the information collected by the sensors located there; and (2) to carry out digital experiments to test the consequences of modifications before implementing them in a real system.
Scope and objectives
Visualisation plays a central role in digital twins by providing a graphical user interface to represent the real-world system and its digital counterpart. Although there is a lack of a systematic study of visual analytic techniques for digital twins, recent examples demonstrate how visualisation can be used for model output correction, state monitoring and display, equipment simulation and prediction, and more generally for human-machine interaction and collaboration [2]. Despite the promise of DTs, they must overcome two main barriers: (i) techniques need to present the right data in an intuitive and insightful manner for users in contextually relevant ways; (ii) uncertainty is inherent in all data, arising for example from measurement noise and errors, modeling errors and approximations, etc [5]. Both of these barriers are exacerbated by the mediation of the ML models between raw data and final representation.
Following a user-centered design methodology, the recruited students will design, implement and evaluate an interactive user interface for ML results visualisations, to assist operators in quickly and reliably monitoring the achievable goals (given the current process state), providing timely opportunities to change the course of action.
PhD topic 1 aims to develop methods for presenting the digital twin in an intuitive and insightful way, with a focus on interactive exploration [4] and user-customizable systems [1] that enable domain-expert non-programmers to tailor data representations to their specific needs.
PhD topic 2 focuses on uncertainty communication and monitoring user trust [3]. The goal is to propose and validate effective methods to communicate the intrinsic uncertainties that arise throughout the data visualisation pipeline, from sensor to visualisation.
PhD Student profile
Academic Background:
- A Master's degree (Bac +5) or equivalent in Computer Science, Human-Computer Interaction, or a closely related field.
Required skills include:
- Web development;
- Programming (JavaScript/D3.js, Python, other)
- Interest in working with real-world data and domain experts
- Excellent communication skills, both written and oral
Additional skills:
- Knowledge in machine learning is not required but experience is a plus
- Experience with user-centered design & information visualization is a plus
- French language is not required, but a plus. There is a possibility to take french lessons for the recruited student.
Work environment
- Supervisors: Nadia Boukhelifa and James Eagan
- Start date: ideally in January 2026, or soon thereafter
- Location: Palaiseau (AgroParisTech Saclay Campus and Télécom Paris, about a 5-minute walk apart).
The two research labs are located in a newly built campus inaugurated in 2019. The campus is accessible by public transportation (2 regional train stations, several bus lines, and opening of a new metro line in 2026). Adjacent to the Vallée de Chevreuse national park, it is approximately 30 minutes from Paris. Paris is one of the most visited cities in the world and offers a wide variety of museums, theaters, art and design exhibitions as well as lively artistic, athletic, music, scientific, technology, and nightlife scenes.
To Apply
Send CV & cover letter to nadia.boukhelifa@inrae.fr and james.eagan@telecom-paris.fr
References
[1] Badam S.K. et al. (2019). Vistrates: A Component Model for Ubiquitous Analytics. IEEE TVCG, 25(1). [2] Botin-Sanabria D.M. et al. (2022). Digital Twin Technology Challenges and Applications: A Comprehensive Review. Remote Sensing, 14(6), 1335. [3] Boukhelifa, N. et al. (2017). How data workers cope with uncertainty: A task characteristion study. ACM SIGCHI Conference on Human Factors in Computing Systems, 3645–3656. [4] Boukhelifa, N, et al. (2019). An exploratory study on visual exploration of model simulations by multiple types of experts. ACM SIGCHI Conference on Human Factors in Computing Systems, (pp. 1--14). [5] Van der Horn E et al. (2021). Digital Twin: Generalization, characterization and implementation. Decision Support Systems, 145, 113524. [6] Verdouw, C. et al. (2021). Digital twins in smart farming. Agricultural Systems, 189, 103046.