Abstract: In robotic visual servoing, when prior measurement or imaging is impractical, control must rely on a single image of the initial view. Diffusion models can generate 3D shapes from a single ...