AnchorDream: Repurposing Video Diffusion for Embodiment-Aware Robot Data SynthesisShare on Twitter Facebook LinkedIn Previous Next