Concept
Unified Dynamics Transformer
A unified dynamics Transformer architecture that jointly evolves cross-modal dynamic representations of appearance, geometry, and action within a latent space through shared parameters, with each modality equipped with dedicated encoders and decoders.
Timeline (last 90 days)
Oct 14
Thinker-WM uses a shared Unified Dynamics Transformer to unify RGB, metric depth, and robot actions into the same dynamic representation, while equipping each modality with dedicated encoders and decoders to preserve the output structure of each modality.
Unverified50%