Looking ahead, I am most interested in foundation models for language and vision, how the two connect, and what it takes to adapt them to robotics, 3D scene understanding and spatial reasoning, with a particular focus on vision-language-action models.