Songen Gu

Retaining future imagination through compact latent actions allows LAWA to outperform existing models while slashing inference latency by nearly 43%.

Vid2WAM achieves superior task generalization and data efficiency by leveraging video diffusion priors, even with minimal expert demonstrations.

Tactile dynamics are crucial for contact-rich manipulation, and VT-WAM outperforms existing models by 26.67% to 35.84% by effectively integrating visual and tactile cues.

TacForeSight enables robots to anticipate contact changes in real-time, outperforming traditional methods in dynamic manipulation tasks.

$\tau_0$-WM outperforms traditional models by seamlessly integrating action prediction and evaluation, leading to superior performance in complex robotic tasks.

Achieve robust robot manipulation across diverse viewpoints without camera calibration by synthesizing novel views with a geometry-aware video diffusion model.

Pocket-sized VLA models can now achieve state-of-the-art robot manipulation performance by pre-training on a curated multimodal dataset and injecting manipulation-relevant representations into the action space.

Robots can now manipulate objects with greater dexterity and adaptability thanks to a new world model that leverages both vision and high-frequency tactile feedback to predict and react to contact dynamics.