What it is
PhysDB uses Physical AI as the top-level map for models, robot bodies, data, simulation, perception, action, safety, and deployment constraints.
Why it matters
The term is useful only when it stays tied to physical constraints: latency, embodiment, contact, energy, uncertainty, and validation.
How not to overread it
Do not collapse it into one vendor model, one humanoid demo, or one benchmark score.
Related edges
World foundation models
Physical AI model taxonomy
One model family does not define the field.
Vision-language-action models
Physical AI model taxonomy
VLA models are components, not whole robots.
Simulation
Robot training and validation
Simulation is an approximation.
Safety evaluation
Physical deployment
No page should imply safety certification.
Onboard compute
Robot deployment
Physical AI claims need power, latency, and thermal context.
Humanoid embodiment
Embodiment classes
Humanoids are not the whole field.