Xinhai Sun

Xinhai Sun is a researcher in robotics and machine learning whose work focuses on vision-language-action systems, reinforcement learning, and robot learning. He is affiliated with Synthoid.ai and is best known for contributions including RDGen, a method that leverages trained RL policies to generate high-quality demonstration trajectories for downstream VLA training, as well as work on contrastive meta-adaptation and foveated attention for egocentric representations. His publication record also includes research on depth completion for robotic systems.