Glossary
Plain-language definitions of the terms you keep running into in modern robotics and embodied AI.
- Affordance
- What a robot can actually do in a given situation, as distinct from what it might want to do.
- Behavior cloning
- Train a model to predict the action a human would take, given the same observation.
- Diffusion Policy
- A behavior-cloning policy that predicts short action trajectories via iterative denoising — the dominant imitation-learning baseline in modern robotics.
- Domain randomization
- Randomize simulation parameters during training so the policy works across many possible realities — including the real one.
- Embodied AI
- AI that perceives, reasons about, and acts in the physical world through a body — usually a robot.
- End-effector
- The business end of a robot arm — the gripper, hand, suction cup, or specialized tool that interacts with the world.
- Foundation model
- A single large model trained on a lot of data that's then adapted to many downstream tasks.
- Inverse kinematics
- Given a target pose for the end-effector, find joint angles that produce it.
- Proprioception
- A robot's sense of its own body — joint angles, velocities, motor currents, gripper state.
- ROS 2
- Robot Operating System 2 — the publish/subscribe message bus and tooling that most modern robots use to wire components together.
- Sim-to-real
- The practice of training a robot policy in simulation and deploying it on a real robot.
- SLAM
- Simultaneous Localization and Mapping — building a map of an unknown environment while keeping track of your position in it.
- Teleoperation
- A human controlling a robot in real time, usually via VR headset, joystick, or leader-follower puppetry.
- URDF
- Unified Robot Description Format — the XML-based file that describes a robot's links, joints, kinematics, and visual / collision meshes.
- Vision-language-action (VLA) model
- A robot policy that takes a camera image + natural-language instruction and emits motor actions.