Marc Duclusaud

Institut Polytechnique de Bordeaux, Laboratoire Bordelais de Recherche en Informatique

Papers

Total Citations

H-Index

About

Marc Duclusaud is a leading researcher in humanoid robotics, with a primary focus on locomotion control, fall recovery, and high-fidelity physics simulation. His most impactful work includes the development of **FRASA**, an end-to-end reinforcement learning agent that enables humanoid robots to autonomously recover from falls and stand up—a critical challenge in dynamic environments where traditional Model Predictive Control and keyframe-based methods fall short. This work has already garnered 4 citations and represents a significant step toward more resilient, real-world capable robots. Duclusaud also contributed to **extended friction models for servo actuators**, improving the accuracy of physics simulations used in reinforcement learning pipelines, a foundational tool for modern robot control. As a core member of the **Rhoban Football Club**, he helped lead the team to victory in the RoboCup Humanoid Kid-Size League in 2023, a prestigious international competition that benchmarks the state of the art in autonomous humanoid soccer. With a growing citation record and a focus on bridging simulation and reality, Duclusaud is shaping the next generation of agile, self-recovering humanoid robots.