Jian Tang

HAF outperforms conventional VLA models in humanoid loco-manipulation by effectively managing complex motion coordination without the need for extensive computational resources.

GAINS reveals that effectively modeling human imperfection can boost task success rates by over 20% in robot manipulation tasks.

Unified image restoration methods can now be rigorously compared across real-world conditions, with 20 teams showcasing innovative solutions that push the boundaries of restoration accuracy.

Achieving high perceptual quality in video compression at bitrates below 0.005 bpp could redefine the limits of efficient video transmission.

Attention-level generalization in Pelican-VLA 0.5 allows it to focus on relevant objects without any task-specific training, outperforming traditional models in unseen scenarios.

Humanoid robots can complete laboratory tasks but often fail to meet the precision required for scientific validity, exposing a critical gap in current automation efforts.

IOI achieves state-of-the-art simulation performance by decoupling deterministic motion from stochastic physical interactions, enabling robust zero-shot generalization to unseen tasks.

Achieve human-like dexterity in humanoid robots by unifying visual-language cues with learned whole-body proprioceptive dynamics, outperforming prior methods in complex manipulation tasks.