Wenye Yu

Real-world robot policies can learn autonomously from sparse rewards by treating value predictions as task-progress memory, outperforming offline training by $32.8\%$ success rate.

Robots can now evolve their own code, learning from task execution to autonomously repair and reuse skills, achieving superior performance on complex manipulation benchmarks.

LIFT accelerates learning in vision-language-action policies by injecting reactive force feedback, achieving superior performance in contact-rich tasks.

Robots can now learn complex manipulation tasks directly from human demonstrations using only a pair of smart glasses, achieving zero-shot transfer without specialized hardware.