Models & learning
Embodied AI
Embodied AI is artificial intelligence that perceives and acts through a body in an environment, so its actions change the observations available to it next. The body may be a physical robot or, in common research usage, an agent situated in a persistent simulated world. Embodied AI emphasises the coupled loop between morphology, perception, action, learning, and the environment.
Also known as: embodied artificial intelligence, EAI
Updated
Intelligence is coupled to a body and environment
Embodied AI treats the body as part of the intelligent system rather than as a replaceable output device. The body's shape, materials, sensors, actuators and possible movements affect which information is available and which actions are feasible. Pfeifer and Bongard argue that morphology both constrains and enables intelligent behaviour.
The defining mechanism is a sensorimotor loop. An agent observes, acts, changes its relationship with the environment and receives new observations caused partly by that action. The ACM survey of embodied intelligence describes the field through the connections among morphology, action, perception and learning rather than any component in isolation.
Embodied does not always mean physically deployed
Robotics and computer-vision research often calls an agent embodied when it acts in an interactive simulated environment with persistent state. RoboTHOR, for example, was designed to develop embodied agents in simulation and evaluate transfer to corresponding physical environments.
On Humanoids Data, physical AI is the narrower label for systems acting through real hardware. This distinction keeps simulation-based embodied research in scope without treating simulated friction, sensing or safety as equivalent to the physical world. Our longer guide to embodied AI examines that boundary in more detail.
What the term does not guarantee
Running an AI model on a robot does not by itself demonstrate meaningful embodiment. A system may still replay an open-loop sequence or ignore the consequences of its actions. Evidence should show that observations, decisions and control update as the interaction unfolds.
The label also says nothing about generality or autonomy. An embodied policy may be limited to one body, task and environment, with conventional controllers and human supervision handling the rest.
Sources
Related terms
Models & learning
Physical AI
Physical AI is artificial intelligence used in machines that perceive and act in the physical world, including robots, autonomous vehicles, and other autonomous systems. It closes a loop between sensing, decisions, action, and physical consequences. The label is emerging rather than standardised and often overlaps with embodied AI.
Hardware & control
Robot embodiment
A robot embodiment is the particular body and sensorimotor interface through which a robot perceives and acts. It includes morphology and kinematics, actuators, end effectors, sensors, physical limits, and the observation and action conventions exposed to a controller or learned policy. Two robots can perform the same task while having different embodiments.
Models & learning
Robot learning
Robot learning is the use of data and machine-learning methods to acquire or improve robot perception, prediction, planning or control. A robot may learn from demonstrations, rewards, direct interaction, simulation or previously collected datasets. The field includes reinforcement learning and imitation learning, but neither term is a synonym for robot learning as a whole.
Models & learning
Policy
A policy is the decision rule that maps a robot’s current observations or estimated state, and sometimes a task instruction, to an action or probability distribution over actions. It can be hand-designed or learned from demonstrations, rewards or both. In humanoid robotics, its outputs may be joint targets, torques, end-effector changes or higher-level skills.
Models & learning
World model
A world model is a learned predictive model of how an environment changes, often conditioned on a robot’s current state and candidate actions. It may forecast pixels, latent states, rewards or other task-relevant quantities. A policy or planner can use those predictions to compare possible futures before commanding the physical robot.