Search papers, labs, and topics across Lattice.
This paper introduces GaussianDream++, an efficient 3D Gaussian world modeling approach that enhances robotic manipulation through the integration of World State Tokens and World Prediction Tokens into Vision-Language-Action (VLA) policies. By leveraging a training-only World Representation Head, the method effectively decouples current world states from future predictions, enabling robust performance in dynamic environments with minimal computational overhead. The results demonstrate a significant improvement in task success rates, achieving 98.6% on LIBERO and 87.8% on LIBERO-Plus, while also enhancing real-robot performance from 29.2% to 52.5%.
Achieving 98.6% accuracy in 3D Gaussian world modeling could redefine efficiency benchmarks for robotic manipulation tasks.
Vision-Language-Action (VLA) policies have advanced language-conditioned robotic manipulation, yet action-imitation objectives provide only weak supervision for metric 3D structure and short-horizon physical evolution. Geometry-enhanced policies mainly improve current-scene grounding, whereas predictive policies often model future dynamics in RGB or latent spaces and may incur substantial deployment cost. GaussianDream demonstrates that training-time current Gaussian reconstruction and future Gaussian prediction provide effective 3D supervision, but its dense VGGT/TGE-based prefix jointly carries state, dynamics, and action-conditioning information. We present \textbf{\methodname}, a compact, policy-native extension that inserts \textbf{World State Tokens} and \textbf{World Prediction Tokens} directly into the VLA backbone. A training-only \textbf{World Representation Head} decodes these tokens into a Current World and coupled Future Prediction over shared Gaussian primitives, while static--dynamic factorization preserves persistent structure and focuses residual motion on interaction-relevant regions. At inference, the head, renderer, auxiliary objectives, and VGGT/TGE pathway are removed, leaving only 20 world tokens without online Gaussian decoding or rollout. \method achieves \textbf{98.6\%} on LIBERO and \textbf{87.8\%} on LIBERO-Plus, with clear gains under Camera and Layout shifts. Real-robot experiments further improve average success from 29.2\% to 52.5\% over reproduced $\pi_{0.5}$ while maintaining efficient closed-loop control.
