The vision of our project is to propose new modeling and control strategies for intelligent agents, explicitly considering users in the design and evaluation process, to facilitate the shift toward seamless and adaptive human-agent interaction.

• Devise and understand low-complexity models and efficient data-riven modeling strategies

Large-scale models learned from data attain state-of-the-art performance in many applications, but often at a prohibitive cost. In the same way, some AI algorithms have to be embedded inside the system itself to reduce the inference time which could reduce its accuracy. It is therefore crucial to develop and understand strategies for mitigating these drawbacks with little or no performance loss.  Conversely, when a large amount of data is available or the hardware resources are limited, it is mandatory to extract only the information that is crucial for deriving a reliable model of reduced complexity for analysis or control purposes.

• Develop safe reinforcement learning control strategies 

The features of reinforcement learning make it a relevant candidate for controlling dynamical systems due to its ability to adapt control actions to system variations. However, endowing systems controlled by reinforcement learning with stability and robustness guarantees remains a largely open question.

• Define procedures for the development and evaluation of interactive decision support systems, taking particular needs into account.

• Devise mechanisms for trustable decision support, requiring fault-tolerant data collection (among e.g., heterogeneous entities in dynamic environments) and reliable data processing.

• Develop estimation and control approaches

Exploring estimation techniques including unknown input observers, derive accurate real-time physical and data-driven based dynamic models, and implement robust and resilient control strategies. Strategies that can meet individual-specific requirements for assistive systems and that include artificial intelligence methods are still an open research field.

• 🔗 Adenilso Simão, Professor at ICMC USP

• 🔗 André P.L.F. Carvalho, Professor at ICMC USP

• 🔗 Antoine Gallais, Professor at the Université Polytechnique Hauts-de-France

• 🔗 Bruna Carolina Rodrigues da Cunha, Professor at ICMC USP

• 🔗 Cássio Guimarães Lopes, Professor at at Poli USP

• 🔗 Catherine Dezan, Professor at the Université de Bretagne Occidentale

• 🔗 David Espes, Professor at the Université de Bretagne Occidentale

• 🔗 Hamza Ouarnoughi, Professor at the Université Polytechnique Hauts-de-France

• 🔗 Jamal Daafouz, Professor at the Université de Lorraine

• 🔗 João Cavalcanti Santos, Professor at the Université de Montpellier

• 🔗 José Henrique de M. Goulart, Professor at the Toulouse INP

• 🔗 Kalinka R.L. J. C. Branco, Professor at ICMC USP 

• 🔗 Kamila Rios Rodrigues, Professor at ICMC USP

• 🔗 Márcia da Costa Peixoto, Tenure track position at the Université Polytechnique Hauts-de-France

• 🔗 Marie Chabert, Professor at the Toulouse INP

• 🔗 Oswaldo L. V. Costa, Professor at Poli USP

• 🔗 Phillip M. S. Burt, Professor at Poli USP

• 🔗 Rafik Belloum, Professor at the Université Polytechnique Hauts-de-France

• 🔗 Romain Postoyan, Directeur de Recherche at CRAN

• 🔗 Thierry-Marie Guerra, Professor at the Université Polytechnique Hauts-de-France

• 🔗 Youcef Imine, Professor at the Université Polytechnique Hauts-de-France

• 🔗Luciano Frezzato, Professor at Poli USP