Valentin Duruisseaux
Papers
1
Total Citations
3
H-Index
1
About
Valentin Duruisseaux is a rising researcher at the intersection of machine learning, physics-informed deep learning, and robotic control. His work focuses on embedding fundamental physical laws and geometric structures—particularly Lie group symmetries—directly into neural network architectures to create more efficient, generalizable models for dynamical systems. His most cited paper, "Lie Group Forced Variational Integrator Networks for Learning and Control of Robot Systems" (2022), introduces a novel framework that combines variational integrators with Lie group theory to learn accurate robot dynamics while preserving conservation laws and geometric properties. This approach significantly improves sample efficiency and long-term prediction accuracy compared to black-box models, with applications in model-based control and simulation. Though early in his career, his work has already garnered attention for bridging the gap between classical mechanics and modern deep learning. Duruisseaux’s contributions are particularly valuable for robotics and autonomous systems, where data-efficient, physically consistent models are critical for safe and reliable control.
Research Focus
Key Achievements
Top Papers
- 1