Like other fields in engineering, robotics seems to have invented a language of its own. But don’t worry—here we explain in plain language what many of these obscure terms mean. Learn some of these and you too will be able to speak like a robotics professor.
A.
ACCELEROMETER: A sensor that measures acceleration forces, typically used to figure out which way your robot is going. Won’t tell your robot which way is correct, though.
ACTUATOR: An electric, hydraulic, or pneumatic device that makes a robot move. Examples include a simple motor that makes a wheel turn; a cylinder with a shaft inside that creates motion in a straight line; and a combination of motors and gears that make a robot leg kick. See also “Compliant” and “Servo.”
ALGORITHM: A step-by-step procedure to calculate something, process some data, or make a decision. Implemented as code running on the robot’s control system.
ANDROID: A robot designed to look and act like a human. Generally, an android has a flesh-like appearance, as opposed to a metal or plastic body of humanoid robots. Think Lieutenant Commander Data versus Honda Asimo. See also “Uncanny Valley.”
ANIMATRONIC: A robotic system designed to mimic something living. Usually refers to something non-anthropomorphic. Usually is a dinosaur.
ANTHROPOMORPHISM: The tendency of people to attribute human characteristics to things that aren’t human. If you look at the stereo cameras on a robot and think that’s where its face is, that’s anthropomorphizing.
ARDUINO: A low-cost microcontroller board, popular with hobbyists. Can serve as the brain of a robot that you build yourself from scratch.
AI: Artificial intelligence. The art and science of making machines smart. The holy grail of AI is achieving a level of reasoning, learning, and knowledge comparable to that of a human. In practice, the most notable AI systems are specialized tools and a computer program that pretends to be a psychotherapist.
AUTOMATON: A machine that operates by itself, generally differentiated from a robot in that an automaton doesn’t necessarily include any form of intelligence, but simply carries out a predefined mechanical action.
AUV: Autonomous underwater vehicle. Typically looks like a torpedo and is used for science and exploration. Designation can also include ROVs, or remotely operated underwater vehicles, which don’t look like torpedoes but rather resemble a car engine with stubby arms.
B.
BACKDRIVABLE: A motor or actuator that can be moved by an external force when it’s operational. If you push on a robot arm that is backdrivable, it will sense the forces and move in the direction you’re pushing. See also “Compliant.”
BACKLASH: Gearing systems are never perfect, and backlash refers to the amount of slop or play that you get due to the gear teeth not intermeshing perfectly. Sometimes backlash is desirable (to allow lubricants to circulate), and sometimes it’s not (if it causes vibrations). Most times it’s inevitable.
BIOROBOTICS: Robots that incorporate features or behaviors found in biological organisms. Also known as bio-inspired or biomimetic robots.
BMI: A brain-machine interface, also called a brain-computer interface, uses either invasive or noninvasive sensors to enable humans to directly control robotic prostheses and other systems using brain activity alone.
C.
CARTESIAN ROBOT: A type of robot that only moves in straight lines at right angles, following coordinates on a—you guessed it—Cartesian plane. This simplicity makes them straightforward to build and program, especially in large sizes. Also known as a linear robot.
CHEMBOTS: A DARPA program seeking to create robots that are soft, flexible, and capable of stuffing themselves through openings much smaller than they are.
COMPLIANT: A robot is said to be compliant when it’s not completely rigid and can sense and control the forces it applies on things. If you put your hand in front of a compliant robot, the resistance from your hand will make it stop. If you put your hand in front of a non-compliant robot, it’s going to hurt. See also “Series Elastic Actuator.”
CONTROL: Automatic control refers to the theoretical and practical tools used to make machines behave like we want them to. Control involves creating abstract models of a system, using mathematical tools and computer simulations to understand its behavior, and devising a controller to make it act in a desired way. See also “Feedback” and “Nyquist Plot.”
CYBERNETIC: Robotic systems that contain active biological components, or a biological entity that makes use of mechanical components. A cyborg is a hybrid of robotics and biology.
D.
DARPA: The U.S. Defense Advanced Research Projects Agency. Funds major programs in robotics to build everything from spy hummingbird drones to Terminator-like humanoids. Unofficially known as the Pentagon’s mad-science wing.
DEGREES OF FREEDOM (DOF): Anything independently movable on a robot counts as a degree of freedom, so one actuator is typically worth one degree of freedom. Adding more degrees of freedom to a robot can make it potentially more useful yet terribly difficult to control.
DELTA ROBOT: See “Parallel Robot.”
DRONE: A general term for an autonomous vehicle, most commonly applied to unmanned aerial vehicles. See “UAV.”
E.
ENCODER: Converts the angular position of the shaft of a motor to a digital code that a computer can understand. Used by robots to tell how much their wheels have turned, helping them track the distance that they’ve traveled.
END-EFFECTOR: The gripper or tool at the end of a robot arm designed to interact with the environment. Examples include a pinch-type gripper, a multi-finger device similar to a human hand, and a welding head.
EMERGENCY STOP: Also known as “e-stop.” Big red button that you push to shut down a robot when something goes wrong. Every robot needs one. You never want to have to push it.
EXOSKELETON: Powered exoskeletons are robotic suits worn by humans that exert direct control over our limbs. They can be used to augment your existing strength, or to restore function to disabled limbs.
F.
FEEDBACK: A method used for improving control of machines. It involves measuring a certain parameter (the speed of a mobile robot, for example), comparing it to a desired value, and based on the discrepancy, making a correction (accelerating or decelerating the robot). This process repeats continually and is known as feedback loop. See also “Control” and “Nyquist Plot.”
FORCE FEEDBACK: See “Haptics.”
G.
GPS: Global positioning system. Uses signals sent from a constellation of satellites to pinpoint the coordinates of a receiver anywhere on Earth. Robots use GPS for absolute positioning, but it’s not generally accurate enough to give a position better than a few meters, and it doesn’t work indoors.
GYRO: Short for gyroscope. Uses a rotating or vibrating mechanism that detects when orientation changes. Mobile and aerial robots use gyros for detecting and maintaining their position in space.
H.
HAPTICS: The study of people’s sense of touch and how to use it to create better control systems. Haptic devices apply forces and vibrations to a user’s body to help convey what the robot is doing. If a remote-controlled manipulator hits an object, for example, the operator will feel resistance on the joystick (what’s known as “force feedback”). Conversely, a device like a haptic glove can measure forces exerted by the operator and translate those measurements to a robot hand.
HARMONIC DRIVE: A special type of gearbox that offers virtually no backlash, a compact and lightweight form factor, good resolution, and high torque. For a price. See also “Backlash.”
HOLONOMIC: Usually used in reference to mobile robots, holonomic means that the robot can move (translate) in any direction without having to turn first.
HUMANOID: A robot with a subset of features designed to be similar to humans. Distinguished from androids in that a humanoid robot is not intended to appear to be human, but simply to mimic the basic structure of a human. Humanoid robots usually include two arms, a head, and a torso. See also “Android.”
HRI: Subfield of robotics that focuses on every aspect of human-robot interaction. See also “Roboethics.”
HYDRAULIC: A hydraulic actuator is powered by pressurized fluid, with the high pressure produced by a pump. Hydraulic actuators are generally strong but slow. And leaky.
I.
ICRA: The IEEE International Conference on Robotics and Automation, one of the two largest robotics conferences the world has ever seen. Organized by the IEEE Robotics & Automation Society and held annually in a different city around the world. Pronounced “e-cra” for no apparent reason. See also “IROS.”
IMU: Inertial measurement unit. A sensor system that typically combines accelerometers, gyroscopes, and electronic compasses. It measures orientation, velocity, and inertial forces. Used as navigation systems in UAVs or as orientation sensors in mobile robots and humanoids.
INVERSE KINEMATICS: A set of equations that describe how the joints of a robot should move to position the end-effector at a desired location in space. See also “Singularity” and “XYZ Coordinates.”
IROS: IEEE International Conference on Intelligent Robots and Systems, the other of the two largest robotics conferences the world has ever seen. Organized by the IEEE Robotics & Automation Society and the Robotics Society of Japan. Held annually, with location alternating between a Japanese and a non-Japanese city. See also “ICRA.”
J.
JAMMING: An actuation technique that involves applying vacuum to a pouch filled with a granular material; when the particles get packed together, they “jam,” or lock into one another. Researchers have used jamming to demonstrate a soft robot gripper that could hold objects by conforming to their shape.
K.
KINECT: A piece of sensor hardware from Microsoft made up of a color camera, an infrared depth sensor, a microphone array, and software that can combine all these data to make 3D maps and track motion and gestures. Originally developed for gaming, its high performance and low cost have led to rapid adoption by roboticists.
KINEMATICS: A series of equations that describe how objects move. Can be applied to just about anything, including robotics. It doesn’t take into account forces causing the objects to move, because then it’s called dynamics.
L.
LAPLACE TRANSFORMATION: A tool that transforms mathematical expressions based in the time domain into expressions based in an abstract, esoteric domain with no direct relation to reality where—believe it or not—designing a control system is easier.
LIDAR: Light detection and ranging. A LIDAR system (or laser range finder) sends out beams of light and measures the time that the beam takes to bounce off of an object and return to the sensor. This gives the distance from the sensor to the object. By sending out lots of beams very fast, some LIDARs can build highly accurate 3D maps of their surroundings. A LIDAR is more expensive than a Kinect sensor, but it’s more accurate and can “see” much farther.
M.
MANIPULATOR: A jointed arm equipped with an end-effector designed to perform tasks. See also “End-effector.”
MATLAB: A mathematical software package used for numerical computing and simulation of dynamic systems. Helps transform ideas into real things; many robots wouldn’t exist if it weren’t for it.
MAV: Micro aerial vehicle. A MAV is a type of UAV that’s smaller than normal, whatever normal happens to be. Generally, it’s probably safe to say that a MAV is a UAV that’s man-portable or smaller.
MCKIBBEN MUSCLE: A pneumatic artificial muscle that contracts when air fills a bladder and extends when air is released. Named after Joseph L. McKibben, a physician who invented the device in the 1950s. In the 1980s, Japanese engineers redesigned the McKibben muscle and came up with a new name for it: the Rubbertuator. It’s still known as McKibben.
MECHATRONICS: A field that combines mechanical engineering, electrical engineering, and computer science to design robots and other intelligent machines.
MOTION PLANNING: A technique that divides a desired movement into smaller, discrete motions. For a mobile robot, a motion planning algorithm determines how to navigate a room from a starting to a finishing point. For a manipulator, a motion planning algorithm determines how to move the joints to position the end-effector at a desired point.
N.
NYQUIST PLOT: Control engineers use the Nyquist plot to determine whether the systems they are designing work (in particular, whether they’re stable and won’t spiral out of control). The plot has an axis with imaginary numbers, but its usefulness is very real indeed. See also “Control.”
O.
OCU: Operator control unit. It can be a highly portable handheld device, a somewhat portable laptop-based controller, or a completely non-portable refrigerator-sized system.
P.
PARALLEL ROBOT: An industrial robot with three to four arms that converge on an end-effector. The robot typically hangs above a working area and positions the end-effector along a plane parallel to the surface. It achieves high speeds and is used for handling and packaging everything from metal widgets to chicken fillet.
PENDANT: Device used to program an industrial robot. You push buttons to move the robot and set desired positions, areas that should be avoided, and actions such as opening or closing the gripper. Pendants may look like game controllers but using them definitely feels more like work than fun.
PLANT: Not the green stuff. In robotic parlance, it refers to a machine or process that you want to control.
PLC: Programmable logic controller. Rugged control systems generally used for industrial automation.
PNEUMATIC: A pneumatic actuator is powered by compressed air or other gases. See also “McKibben Muscle.”
PID: Proportional integral derivative controller. The PID is the workhorse of control systems, used in everything from simple research robots to complex industrial machinery. If you program a robot to drive at 1 m/s, for example, the PID will try to maintain that speed by countering any disturbances with three strategies: The proportional component (P) makes corrections based on the error between the desired speed and the actual speed; the integral component (I) makes corrections based on the accumulation of past errors; and the derivative component (D) makes corrections based on how fast the error is changing. Tweak the P, I, and D parameters and you can almost always get a good controller. See also “Control” and “Feedback.”
Q.
QUADROTOR: Also called “quad” or “quadcopter,” a flying robot that uses four horizontal rotors for lift and control. Simple, cheap, and robust, quadrotors come in a variety of sizes, and more powerful variations (using six, eight, or more rotors) are also used.
R.
RADAR: A sensor that determines the distance of an object by emitting radio waves and measuring the time that the waves take to bounce off of the object and return. Autonomous vehicles and drones use radar to locate objects in their surroundings and avoid crashing on them.
ROBOCUP: An international robotic soccer competition where researchers gather to play with their cool robots and, as a result, advance the state-of-the-art in robotics and artificial intelligence.
ROBOETHICS: Research field that explores the ethical implications of robotics and attempts to prevent the Robopocalypse. Focuses not only on how robots should act but also on how humans who design robots should act. See also “Robopocalypse” and “Three Laws of Robotics.”
ROBOT: A machine that senses its environment, computes a decision based on a stored program, and performs an action. That description, however, can also describe an automatic door, which most certainly is not a robot. A more practical definition is, if you think something is a robot, you can almost always call it a robot.
ROBOPOCALYPSE: The coming robot apocalypse. Results from a robot uprising that ends the world as we know it. Also, a novel about a robot uprising that ends the world as we know it. And a future Spielberg movie about a robot uprising that ends the world as we know it. See also “Three Laws of Robotics” and “Roboethics.”
ROS: Robot Operating System. An open-source software platform used to give robots a variety of capabilities, including vision, navigation, and manipulation. Popular among researchers, ROS provides operating system-like functionality and is flexible and robust. A doctorate degree in robotics is not required to use ROS, but it helps.
S.
SCARA: Selective compliant assembly robot arm. A robot arm with 4 degrees of freedom, typically used in industrial automation.
SEMANTICS: The relationships between things. Robots use what’s called “semantic mapping” to relate objects to each other; for example, the knowledge that cheese is food, and that Twinkies are also food, and that cheese and Twinkies go really well together would be a simple (but rather unpleasant tasting) semantic map.
SENSOR: A device that measures something on a robot’s own body or the world around it. See also “GPS,” “LIDAR,” “Radar”.
SERIAL ROBOT: A robot that consists of a series of links connected to each other with joints that are powered by motors. Popular in industrial automation as pick-and-place robots. See also “Manipulator” and “SCARA.”
SERIES ELASTIC ACTUATOR: An actuator that, in addition to a motor and gearbox, uses a spring to make the device less rigid. By measuring the deflection of the spring, the actuator can determine the forces applied to it. This means that a robot can control its movements by sensing forces rather than just going from one position to another, making it potentially safer for interacting with humans.
SLAM: Simultaneous localization and mapping. SLAM is a technique that robots use to create a map of an unknown environment that they find themselves in. The concept behind SLAM is that a robot needs a map to know where it is, but to make a map, the robots need to know where it is. SLAM attempts to tackle both problems at the same time by combining remote sensing (like LIDAR) and sensors on the robot itself that track motion and orientation (like accelerometers or IMUs).
SWARM ROBOTICS: The study of the collective behavior of dozens or hundreds of robots, drawing parallels with ant colonies and other swarms found in nature.
SERVO: An electric motor that has some mechanism for providing feedback about how much it has moved. Servos are distinct from motors in that a servo is aware of its position as opposed to just on or off, and are used when accurate position control is necessary.
SINGULARITY: A point in space where a robot manipulator may become uncontrollable. Alternately, a point in human history when robots achieve greater-than-human superintelligence and take over the world.
SOCIAL ROBOTS: Robots that are not content in just serving their human masters but also want to befriend them.
SONAR: A type of sensor that uses ultrasound to detect objects and measure the distance to them. Cheaper than radar and LIDAR, but not as good.
STOCHASTIC: A fancy word for random, stochastic refers to a process whose outcome can’t be deterministically predicted and requires statistical tools to studied and accounted for.
SUBSUMPTION ARCHITECTURE: A divide-to-conquer approach to robotics programming popularized in the 1980s. Based on simple modules whose collective action generates more complex behavior.
T.
TELEOPERATION: The act of operating a robot remotely, based on information from cameras and other sensors on the robot. See also “Haptics.”
TELEPRESENCE: Similar to teleoperation, except the robot is intended (on some level) to act as a surrogate for the human operating it. It’ll let you be at the office without actually going there, but it won’t do your work for you.
THREE LAWS OF ROBOTICS: Formulated by science fiction writer Isaac Asimov in 1942. They are: First Law: A robot may not injure a human being or, through inaction, allow a human being to come to harm. Second Law: A robot must obey the orders given to it by human beings, except where such orders would conflict with the First Law. Third Law: A robot must protect its own existence as long as such protection does not conflict with the First or Second Laws.
TORQUE: A measure of the amount of turning force on an object, in newton-meters. In robots, typically used to describe the power of electric motors.
TURING TEST: A test of machine intelligence postulated by Alan Turing. In a generalized interpretation, if a human and a machine are tested in some way such that a second human can’t tell the difference between them, the machine is considered to have passed the Turing Test.
U.
UAV: Unmanned aerial vehicle. The term UAV applies to systems with various degrees of autonomy, from remote controlled aircraft to drones that fly themselves from takeoff to landing.
UCAV: Unmanned combat aerial vehicle. Like a UAV, except armed and dangerous.
UNCANNY VALLEY: A concept in robotics suggesting that if the appearance of a humanoid robot approaches but not attains that of a human, that robot will elicit an uncomfortable reaction from people. In one word: creepy.
V.
VICON: A company that manufactures motion tracking systems. Uses infrared cameras to very precisely track the motion of tagged objects in precalibrated environments; commonly used to control quadrotors indoors.
W.
WORKSPACE: If a robot can reach a point in space, said point is within the robot’s workspace. If a robot can’t reach a point in space, you have to move the robot.
X.
XYZ COORDINATES: The Cartesian coordinate system we know and love. Named after René Descartes, who once wrote: “[A]nd yet what do I see from the window beyond hats and cloaks that might cover artificial machines, whose motions might be determined by springs?” That’s right: Descartes talking about androids indistinguishable from real people. But we digress...
Y.
YAW, PITCH, AND ROLL: Terms to describe motion along an aircraft main axes. A yaw motion moves the nose of the aircraft from side to side. A pitch motion moves the nose of the aircraft up or down. A roll motion keeps the nose straight while moving one wing up and the other down.
Z.
ZMP: Zero moment point is a control method widely used with dynamic robots to help them maintain stability. It positions the robot such that the ground contact point (like the bottom of a humanoid robot’s foot) has zero horizontal motion.
