'/%3E%3C/svg%3E){kind=small}
Hucebot seminar: Sébastien Kleff (Inria)
26 novembre 2026 @ 15:00 pm - 16:30 pm
On November 26th au 3pm in room C005, the HUCEBOT team will have the pleasure of welcoming Sebastien Kleff, postdoctoral fellow at the Inria Center of Bordeaux, for a seminar on human-robot interaction.
Title: Toward Structured Intelligence for Physical Interaction in Robotics
Abstract: Despite impressive advances in robotics, reliable physical interaction beyond controlled laboratory settings remains a challenge. Whether manipulating unfamiliar objects or collaborating physically with people, robots must adapt their behavior to uncertain physical interactions while maintaining task performance and safety. Yet contact dynamics are difficult to model accurately, while interaction data remain scarce and expensive to collect.
In this talk, I explore how model-based control can address these challenges by exploiting physical and mathematical structure. I first show how efficient numerical optimization enables real-time Nonlinear Model Predictive Control (NMPC) on robotic hardware. Turning to physical interaction, I demonstrate how incorporating force feedback into NMPC enables accurate force regulation alongside predictive planning and explicit physical constraints. Beyond force sensing, I show how physically grounded contact representations enable robust control on unknown surfaces. Finally, I discuss uncertainty-aware human-robot collaboration and share my broader vision of structured intelligence for reliable physical interaction.
In this talk, I explore how model-based control can address these challenges by exploiting physical and mathematical structure. I first show how efficient numerical optimization enables real-time Nonlinear Model Predictive Control (NMPC) on robotic hardware. Turning to physical interaction, I demonstrate how incorporating force feedback into NMPC enables accurate force regulation alongside predictive planning and explicit physical constraints. Beyond force sensing, I show how physically grounded contact representations enable robust control on unknown surfaces. Finally, I discuss uncertainty-aware human-robot collaboration and share my broader vision of structured intelligence for reliable physical interaction.
Bio: Sébastien Kleff received the Diplôme d’Ingénieur (M.Sc.) from École Centrale de Nantes and an M.Sc. in Electrical Engineering from Shanghai Jiao Tong University in 2019. He earned his Ph.D. from New York University in 2024. Between 2020 and 2022, he conducted part of his doctoral research at LAAS-CNRS. He is currently a postdoctoral researcher at the Inria Centre in Bordeaux. His research focuses on model predictive control, force and tactile feedback control, and human-robot collaboration.
