Dexterous hands and humanoid robots are typically developed as distinct embodiments: the former enable contact-rich manipulation at the object scale, whereas the latter provide mobility and whole-body interaction in human-centered environments. We introduce \textbf{Handroid}, a d…
Compliance is essential for dexterous manipulation, yet existing solutions often rely on external tactile or force sensors that are costly, fragile, and difficult to deploy on low-cost robot hands. We propose a proprioception-driven framework that learns contact-aware compliance…