Our research direction — learning world models from interactive experience toward embodied agents — and the work we build on.
AI & ML interests
World models · Embodied agents · Multimodal real-time companions · Learning from human–AI interaction · Latent action representations · Efficient on-device inference
