Built the Humanoid Olympics manipulation benchmark; scores new systems on a fixed task set by success and time rather than by demo.
The Physical AI Voices by Herbert Scale Experts.