A safety standard for humanoids and other actively balanced robots should fix the goal, not the method: how to meet it is up to the designer. Humanoids are intrinsically unpredictable, so zero uncertainty is impossible;
The Physical AI Voices by Herbert Scale Experts.