Reference
Robotics Glossary
59 terms, jargon, and abbreviations — explained in plain English. New to robotics? Start here.
A
- Actuator
- The component that produces motion — electric motors, hydraulics, or pneumatics that move a robot's joints.
- AGV (Automated Guided Vehicle)
- A mobile robot that follows fixed routes (wires, magnets, or markers) — the predecessor of the free-ranging AMR.
- AMR (Autonomous Mobile Robot)
- A mobile robot that plans its own paths around people and obstacles instead of following fixed routes.
- AprilTag
- A visual fiducial marker (like a simplified QR code) robots use for precise localization and calibration.
C
- Cobot (Collaborative Robot)
- A robot arm designed to work safely alongside people without cages, using force limits and rounded surfaces.
- Computer vision
- Software that extracts meaning from camera images — detection, tracking, segmentation, and pose estimation.
D
- Dead reckoning
- Estimating position from motion history (wheel turns, IMU data) without external references; drifts over time.
- Degrees of freedom (DoF)
- The number of independent ways a robot can move. A typical industrial arm has 6; a humanoid hand alone can have 20+.
- Depth camera
- A camera that measures distance per pixel (stereo, structured light, or time-of-flight), giving robots 3D sight.
- Dexterous manipulation
- Fine, coordinated control of objects — the hard problem behind robot hands doing human tasks.
- Digital twin
- A virtual replica of a robot or facility used to simulate, test, and optimize before touching hardware.
E
- Embodied AI
- AI that learns and acts through a physical body in the real world, rather than purely in software.
- Encoder
- A sensor that measures joint or wheel rotation — the feedback that makes precise motion control possible.
- End effector
- The business end of a robot arm — gripper, suction cup, welder, or any task-specific tool.
- Exoskeleton
- A wearable robot that augments human strength or assists rehabilitation.
F
- Fleet management
- Software that orchestrates many robots at once — assigning tasks, routing traffic, and monitoring health.
- Force-torque sensor
- A wrist-mounted sensor measuring forces and torques, letting a robot feel how hard it is pushing.
- Forward kinematics
- Computing where the end effector is from the robot's joint angles.
- Foundation model (robotics)
- A large model trained on broad robot data intended to generalize across many tasks and embodiments.
G
- Gait
- The pattern of leg movements a walking robot uses — and the control problem of keeping it stable.
- Gripper
- An end effector that grasps objects — two-finger, vacuum, soft, or multi-fingered.
H
- Haptics
- Technology for touch feedback — letting teleoperators feel what the robot feels.
- Harmonic drive
- A compact, zero-backlash gear system used in most robot arm joints.
- Humanoid
- A robot with a human-like body plan — two legs, two arms, a torso — built to work in spaces designed for people.
I
- Imitation learning
- Training robots by demonstration — from teleoperation recordings or human video — instead of hand-coding behavior.
- IMU (Inertial Measurement Unit)
- A chip combining accelerometers and gyroscopes that tells a robot how it is moving and tilting.
- Inverse kinematics
- Computing the joint angles needed to put the end effector at a desired position — the reverse of forward kinematics.
L
- Lidar
- A spinning or solid-state laser sensor that maps surroundings in precise 3D — a staple of mobile robot navigation.
- Localization
- A robot figuring out where it is on its map.
M
- Manipulator
- A robot arm — a chain of joints and links that positions an end effector.
- Mobile base
- The wheeled or tracked platform a mobile robot is built on.
- Motion planning
- Computing a collision-free path from here to there, for arms and mobile robots alike.
O
- Odometry
- Motion estimation from wheel encoders, IMUs, or vision — the input to dead reckoning.
P
- Path planning
- Finding an efficient route through a map, before motion planning smooths and executes it.
- Payload
- The maximum weight a robot can carry or lift at rated performance.
- PID control
- The classic feedback control loop (proportional–integral–derivative) behind most servo and motor control.
- Point cloud
- A set of 3D points from lidar or depth cameras — the raw material of robot perception.
- Pose
- Position plus orientation — the six numbers that fully describe where something is in 3D space.
R
- Reinforcement learning (RL)
- Training by trial and error against a reward signal, often millions of times in simulation.
- Repeatability
- How precisely a robot returns to the same position — often ±0.02 mm for industrial arms.
- ROS (Robot Operating System)
- The open-source middleware most robots are developed on — messaging, drivers, tools, and a huge package ecosystem. ROS 2 is the current generation.
S
- Servo
- A motor with built-in position feedback and control — from hobby servos to industrial servo drives.
- Sim-to-real
- Transferring behavior learned in simulation to a physical robot — and the "reality gap" that makes it hard.
- Singularity (kinematic)
- An arm configuration where the robot loses a degree of freedom and control becomes unstable — planners route around them.
- SLAM
- Simultaneous Localization and Mapping — building a map while tracking your own position in it.
- Soft robotics
- Robots built from compliant materials that bend and squish — safer contact and gentler grasping.
- Speed and separation monitoring
- A cobot safety mode that slows or stops the robot based on how close a person is.
- Stepper motor
- A motor that moves in fixed increments without feedback — cheap, precise, common in 3D printers and small robots.
- Swarm robotics
- Many simple robots coordinating to accomplish tasks no single robot could — inspired by insects.
T
- Tactile sensing
- Touch sensors — pressure, slip, and texture — that let robot hands feel what they hold.
- Teach pendant
- The handheld terminal used to program industrial robots point by point.
- Teleoperation
- A human directly controlling a robot remotely — also how much humanoid training data gets collected.
- Time-of-flight (ToF) sensor
- Measures distance by timing light's round trip — used in depth cameras and rangefinders.
U
- URDF
- Unified Robot Description Format — the XML file that describes a robot's body for ROS and simulators.
V
- VLA (Vision-Language-Action) model
- A model that maps camera images and natural-language instructions directly to robot actions.
W
- Waypoint
- An intermediate target position along a planned path.
- Whole-body control
- Coordinating every joint of a legged robot at once to balance, walk, and manipulate simultaneously.
- Workspace
- The volume of space a robot arm can reach.
Z
- Zero-shot
- Performing a task with no task-specific training — the generalization goal of robot foundation models.
No terms match — try a different word.
Frequently asked questions
What is SLAM in robotics?
SLAM (Simultaneous Localization and Mapping) is how a robot builds a map of an unknown space while tracking its own position within it — the foundation of most mobile robot navigation. See the full term list above.
What is a cobot?
A cobot (collaborative robot) is a robot arm designed to work safely alongside people without safety cages, using force limits, rounded surfaces, and speed monitoring. They're the fastest-growing category of industrial robot.
What is a VLA model?
A vision-language-action model maps camera images and natural-language instructions directly to robot actions — the architecture behind the current wave of general-purpose robot intelligence. Related terms like foundation model, imitation learning, and sim-to-real are defined above.