AI is the broader category
Artificial intelligence includes systems for language, vision, prediction, planning and many other tasks. These systems do not necessarily need a physical body to perform their intended function.
What makes intelligence embodied?
Embodied intelligence is grounded in sensing and acting within an environment. The agent's body, sensors and action capabilities influence what it can perceive and how it can achieve a goal.
Why this distinction matters in robotics
Robots must convert abstract goals into physical actions. A useful system therefore needs representations that connect perception, reasoning and control while respecting morphology, dynamics and uncertainty.
The connection to physical AI
Physical AI is a practical expression of this idea: AI systems operating in physical environments. Robot learning, VLA models, world models and cross-embodiment learning all address different parts of the same interface between intelligence and action.
Explore the field
Continue through the research map, embodied AI guide and cross-embodiment learning guide.