Autonomy-stack software for humanoid robots that handles command-to-control workflows, manipulation, and locomotion across diverse hardware and tasks. Uses simulation and reinforcement learning to scale policies to the real world with minimal human intervention.