Romain Vuillemot

Papers

Total Citations

H-Index

About

Romain Vuillemot is a leading researcher in visual analytics and human-computer interaction, with a focus on making complex AI and robotics systems interpretable. His work centers on the critical challenge of understanding and debugging the "black box" of machine learning models, particularly in deep reinforcement learning and robotics. His most notable contribution is **DRLViz** (2019, 37 citations), a visual analytics interface that reveals the internal memory and decision-making processes of agents trained via deep reinforcement learning. By visualizing the large, temporal vectors that govern a robot's actions, Vuillemot provides researchers with a powerful tool to demystify why an agent makes certain choices. He further extends this work to address the **Sim2Real gap** with **SIM2REALVIZ** (2021), a system that visualizes discrepancies between simulated training environments and real-world deployment, a critical problem for reliable robotics. Through these contributions, Vuillemot is pioneering the field of explainable AI for embodied agents, enabling more transparent, trustworthy, and robust autonomous systems.