Yunheng Wang

Treating action conditioning as a structured process rather than a global compression could redefine how we model high-dimensional dexterous actions in AI.

Hierarchical control in humanoid robots can solve complex leg-critical tasks, but performance is fragile and heavily reliant on the choice of motion tracker.

Forget finetuning: this training-free method achieves state-of-the-art zero-shot 3D visual grounding, even in messy, real-world environments.