About myself

I am Félix Doublet, a French first-year PhD student at Science Tokyo, where I am part of Shinoda Lab. My current research focuses on robotic manipulation using Vision-Language-Action (VLA) models.

Before starting my PhD, I completed Master’s research at the SMILAB lab at Keio University, advised by Prof. Komei Sugiura. My research has covered visual explanation methods for ResNet models and learning-based approaches for robotic decision making.

Background

I have a generalist engineering background from Ecole Centrale de Nantes and French classes préparatoires. I then moved to Japan for a double-degree program combining my interest in artificial intelligence with my long-standing interest in Japanese culture.

Research interests

I am especially interested in building foundation models for embodied agents, with a current focus on VLA-based robotic manipulation. My earlier work focused on explainable AI, including visual explanation generation based on Layer-Wise Relevance Propagation for ResNet. I have also worked on reinforcement learning, world models, and recovery-oriented robotic policies.

My passions

I am passionate about Japanese culture and continue learning Japanese. I am also a big fan of sports, especially football, tennis, and Formula One.