Black Scarab Reference Guide
Term in focus
Consistency, accuracy, and reading a robot specification.
Read the full explanationSoftware and Connectivity
5G is the fifth generation of cellular network technology, designed to support faster data transmission, greater network capacity, and lower communication delays than previous generations.
Sensors and Perception
An absolute encoder is a sensor that measures the position of a rotating shaft or mechanical component and assigns a distinct value to that position.
Sensors and Perception
An accelerometer is a sensor that measures acceleration, helping a machine detect changes in movement, vibration, impacts, and orientation relative to gravity.
Industry and Safety
Acceptance testing is the process of verifying that a machine, robot, or automation system meets the requirements agreed upon between a supplier and its customer.
Robotics and Actuation
Robot accuracy measures how closely a robot reaches its intended or commanded position.
Sensors and Perception
Acoustic sensing is the use of sound waves to detect, measure, or understand events in the surrounding environment.
Models and Learning
An action token is a unit of information used by certain artificial intelligence models to represent part of a physical action.
Models and Learning
Active learning is a machine learning approach in which a model helps select which examples should be labeled and added to its training data.
Models and Learning
Actor-critic is a family of reinforcement learning methods that uses two complementary functions to improve an AI system's decision-making.
Robotics and Actuation
An actuator is a device that converts energy into physical movement or force.
Foundations
Adaptation is the ability of a system to adjust its behavior or operation when conditions change.
Navigation and Control
Admittance control is a robotics control method that adjusts a robot's movement in response to an external force.
Robotics and Actuation
An automated guided vehicle, or AGV, is a mobile machine designed to transport materials along predefined routes with limited direct human control.
Compute and Memory
An AI accelerator is a processor or specialized hardware component designed to perform artificial intelligence calculations efficiently.
Foundations
An AI agent is a software system that uses artificial intelligence to pursue goals, make decisions, and perform actions with some degree of independence.
Foundations
Artificial intelligence, or AI, is the field of developing computer systems capable of performing tasks associated with human intelligence, such as recognizing patterns, understanding language, making predictions, and choosing actions.
Compute and Memory
An arithmetic logic unit, or ALU, is a component inside a processor that performs basic mathematical and logical operations.
Robotics and Actuation
An autonomous mobile robot, or AMR, is a robot that uses sensors, software, and navigation technology to move through an environment with limited direct human control.
Software and Connectivity
An application programming interface, or API, is a defined way for software systems to exchange information or request functions from one another.
Compute and Memory
An application-specific integrated circuit, or ASIC, is a chip designed to perform a particular task or group of related tasks efficiently.
Industry and Safety
Assembly automation is the use of machines, robots, and control systems to join components or perform tasks involved in building a product.
Foundations
Automation is the use of machines, software, or control systems to perform tasks with reduced direct human intervention.
Foundations
An autonomous system is a machine or software-based system capable of carrying out defined tasks and making certain decisions without continuous direct human control.
Foundations
Autonomy is the ability of a machine or system to make and execute decisions within a defined task or environment without requiring continuous human direction.
Robotics and Actuation
Backdrivability describes how easily an external force can move a motor-driven mechanism or robot joint in the reverse direction through its transmission.
Robotics and Actuation
Backlash is the amount of free movement that can occur between mechanical components when the direction of motion or applied force changes.
Models and Learning
A batch is a group of data examples processed together by a machine learning model during training or inference.
Models and Learning
Behavior cloning is a machine learning method that teaches a system to perform tasks by learning from examples of the desired behavior.
Robotics and Actuation
A bearing is a mechanical component that supports movement between parts while reducing friction and controlling how loads are transferred.
Robotics and Actuation
A brushless direct current motor, or BLDC motor, is an electric motor that uses electronic control instead of mechanical brushes to manage the electrical switching needed for rotation.
Software and Connectivity
Bluetooth Low Energy, or BLE, is a wireless communication technology designed to exchange information while consuming relatively little power.
Industry and Safety
Brownfield deployment is the installation of new technology, equipment, or automation systems within an existing operational facility.
Compute and Memory
Cache is a small, fast area of memory that stores information a processor is likely to need again soon.
Sensors and Perception
A camera is a device that captures light and converts it into images or video that a computer can process.
Software and Connectivity
Controller Area Network, commonly called CAN or CAN bus, is a communication system that allows electronic controllers and devices to exchange messages over a shared network.
Robotics and Actuation
A Cartesian robot is a robotic system that moves along straight-line axes, typically arranged according to the X, Y, and Z directions.
Industry and Safety
Changeover is the process of preparing production equipment to switch from making one product or performing one operation to another.
Compute and Memory
A chiplet is a relatively small semiconductor die designed to operate as part of a larger chip package containing multiple dies.
Navigation and Control
Closed-loop control is a method of controlling a system by continuously measuring its actual behavior and adjusting commands to reduce the difference from a desired result.
Software and Connectivity
Cloud computing is the delivery of computing resources, such as processing power, storage, and software, through network-accessible infrastructure that can be provisioned when needed.
Industry and Safety
A collaborative robot, commonly called a cobot, is a robot designed with features that can support applications involving interaction or shared workspaces between humans and robots.
Foundations
Cognition refers to the processes involved in acquiring information, understanding situations, reasoning, remembering, and making decisions.
Simulation and Data
Collision detection is the process of determining whether objects are touching, intersecting, or are likely to collide based on their positions and shapes.
Industry and Safety
Commissioning is the process of verifying, configuring, testing, and preparing installed equipment or systems for their intended operation.
Compute and Memory
A compiler is a software tool that translates computer code or another structured program representation into a form that a target system can execute or process.
Robotics and Actuation
Compliance is the ability of a mechanical system or controlled robot to yield or move in response to an applied force.
Industry and Safety
Condition monitoring is the process of collecting and analyzing information about equipment to assess its operating condition and identify signs of deterioration.
Simulation and Data
A contact model is a mathematical representation of what happens when objects touch or interact physically.
Software and Connectivity
Containerization is a software technology that packages an application with the files, libraries, and dependencies it needs to run in a consistent computing environment.
Models and Learning
A context window is the amount of information an artificial intelligence model can process as part of a single interaction or computation, commonly measured in tokens.
Models and Learning
Continual learning is a machine learning approach that allows an AI model to acquire new knowledge or capabilities over time while attempting to preserve what it has previously learned.
Robotics and Actuation
A continuum robot is a robot with a flexible structure that bends continuously along part of its body instead of moving exclusively through a small number of rigid joints.
Foundations
Control is the process of directing a machine's behavior so that its actual operation follows a desired objective.
Navigation and Control
Control gain is a numerical parameter that determines how strongly a controller responds to an input, measurement, or difference between desired and actual behavior.
Navigation and Control
Control loop frequency is the number of times a controller updates its calculations and commands each second.
Navigation and Control
A coordinate frame is a reference system used to describe the position and orientation of objects in space.
Navigation and Control
A costmap is a representation of an environment in which different locations are assigned numerical costs that help a robot determine where it should travel.
Foundations
A cyber-physical system, or CPS, is a system in which computing, communication, and physical processes are closely connected.
Compute and Memory
A central processing unit, or CPU, is the general-purpose processor responsible for executing program instructions and coordinating many of a computer's operations.
Compute and Memory
A compute unit is a group of processing resources within a computer chip that executes computational work.
Compute and Memory
CUDA, short for Compute Unified Device Architecture, is NVIDIA's parallel computing platform and programming model for running general-purpose calculations on compatible NVIDIA GPUs.
Sensors and Perception
A current sensor is a device that measures the flow of electrical current through a conductor or circuit.
Models and Learning
Curriculum learning is a machine learning approach that organizes training examples or tasks into a purposeful sequence, often progressing from easier problems to more difficult ones.
Sensors and Perception
Computer vision is a field of artificial intelligence and computing focused on extracting useful information from images and video.
Compute and Memory
Compute Express Link, or CXL, is a high-speed interconnect standard designed to support communication and memory sharing between processors, accelerators, and memory devices.
Industry and Safety
Cybersecurity is the practice of protecting digital systems, networks, software, and information from unauthorized access, disruption, manipulation, and other threats.
Industry and Safety
Cycle time is the time required to complete a production operation, make one unit, or finish a repeated process.
Robotics and Actuation
A cycloidal drive is a mechanical speed-reduction system that uses the movement of specially shaped discs and pins to convert relatively fast input rotation into slower, higher-torque output rotation.
Navigation and Control
Damping is the process by which energy is dissipated or otherwise removed from motion, reducing oscillation and vibration over time.
Simulation and Data
Data annotation is the process of adding labels, descriptions, or other structured information to raw data so it can be used more effectively for machine learning.
Simulation and Data
Data augmentation is a machine learning technique that creates additional training examples by modifying existing data.
Simulation and Data
Data generation is the process of producing or collecting information for use in software development, machine learning, simulation, testing, or analysis.
Simulation and Data
Data governance is the set of policies, responsibilities, and processes that determine how an organization collects, manages, protects, and uses data.
Simulation and Data
A data lake is a storage system designed to hold large amounts of structured, semi-structured, and unstructured information, often in its original or lightly processed form.
Simulation and Data
A data pipeline is a system that moves and processes information between different locations, applications, or stages of a workflow.
Simulation and Data
Data provenance is information describing where data originated, how it was collected, and what happened to it before it reached its current form.
Simulation and Data
Data quality describes how accurately, completely, consistently, and reliably information represents what it is intended to describe.
Simulation and Data
A data split is the division of a dataset into separate groups used for training, validating, and testing a machine learning model.
Simulation and Data
A data warehouse is a system that brings together information from multiple sources and organizes it for querying, reporting, and analysis.
Simulation and Data
A dataset is an organized collection of information used for analysis, machine learning, testing, or research.
Software and Connectivity
Data Distribution Service, or DDS, is a communication standard designed to exchange information between software applications and devices using a publish-subscribe model.
Navigation and Control
Dead reckoning is a navigation method that estimates a machine's current position by starting from a known location and calculating how far and in what direction it has moved.
Models and Learning
A decoder is a component of an artificial intelligence model that converts an internal representation or input sequence into an output.
Robotics and Actuation
A delta robot is a parallel robot that uses multiple lightweight arms connected to a common moving platform.
Industry and Safety
Depalletizing is the process of removing products, boxes, containers, or other items from a pallet.
Sensors and Perception
A depth camera is a sensor that measures or estimates the distance between the camera and points in the surrounding scene.
Sensors and Perception
Depth estimation is the process of determining how far objects or surfaces are from a camera or another reference point.
Navigation and Control
Deterministic computing refers to computing behavior that follows defined, predictable rules. In real-time industrial systems, the term commonly emphasizes the ability to complete important operations within predictable timing limits.
Software and Connectivity
Device management is the process of monitoring, configuring, updating, securing, and maintaining connected electronic equipment.
Industry and Safety
Diagnostics are the methods and information used to identify problems, determine their likely causes, and evaluate the operating condition of equipment or software.
Robotics and Actuation
Differential drive is a method of moving a wheeled robot by controlling the speeds of wheels on opposite sides independently.
Models and Learning
A diffusion model is a type of generative artificial intelligence model that learns to create data by progressively removing noise from an initially noisy representation.
Simulation and Data
A digital thread is a connected flow of information that links the design, manufacturing, operation, maintenance, and other stages of a product's lifecycle.
Simulation and Data
A digital twin is a digital representation of a physical asset, process, or system that is connected to relevant information about its real-world counterpart.
Robotics and Actuation
Direct drive is a mechanical arrangement in which a motor drives its load without an intervening speed-reduction gearbox.
Foundations
Deep learning is a form of machine learning that uses neural networks with multiple layers to recognize patterns, make predictions, and learn representations from data.
Compute and Memory
Direct memory access, or DMA, is a computing mechanism that allows compatible hardware devices to transfer data to or from memory without requiring the CPU to copy every piece of information itself.
Robotics and Actuation
Degrees of freedom, or DOF, describe the number of independent ways a mechanical system can move.
Models and Learning
Domain adaptation is a machine learning approach that helps a model perform effectively when the data it encounters differs from the data used during its original training.
Simulation and Data
Domain randomization is a machine learning technique that deliberately changes elements of a training environment so a model learns to handle a range of operating conditions.
Industry and Safety
Downtime is the period during which equipment or a production process is unavailable for its intended operation.
Compute and Memory
A data processing unit, or DPU, is a specialized processor designed to handle infrastructure tasks such as networking, data movement, storage operations, and security.
Compute and Memory
Dynamic random access memory, or DRAM, is a common type of computer memory used to temporarily store information while a system is operating.
Compute and Memory
A digital signal processor, or DSP, is a processor designed to perform mathematical operations on digital signals efficiently.
Sensors and Perception
Dynamic range describes the span between the smallest and largest signal levels a sensor can measure usefully.
Robotics and Actuation
Dynamics is the study of how forces and torques affect the motion of physical objects.
Industry and Safety
An emergency stop, commonly called an E-stop, is a manually initiated safety function intended to stop a hazardous machine operation in an emergency.
Software and Connectivity
Edge AI is the execution of artificial intelligence models or algorithms on computing devices located close to where information is generated.
Software and Connectivity
Edge computing is an approach to computing that places processing and storage resources close to the devices or locations where information is generated.
Simulation and Data
Edge data is information generated, collected, or stored close to machines, sensors, and other devices operating in the physical world.
Software and Connectivity
An edge gateway is a device or software system that connects local equipment, sensors, and machines to other networks or computing services.
Compute and Memory
An edge processor is a computing chip or processing system used to execute tasks close to where information is generated, rather than relying entirely on remote computers.
Navigation and Control
An extended Kalman filter, or EKF, is a mathematical method used to estimate the state of a system when its motion or measurements involve nonlinear relationships.
Robotics and Actuation
An electric actuator is a device that uses electrical energy to produce mechanical movement.
Models and Learning
An embedding is a numerical representation of information that captures useful characteristics or relationships in a form an AI model can process.
Foundations
Embodied AI is artificial intelligence developed to perceive, learn, reason, or act through interaction with a body and its environment.
Foundations
Emergent behavior describes a pattern or capability that arises from interactions within a system rather than from a single explicit instruction defining that behavior.
Sensors and Perception
An encoder is a sensor that measures the position or movement of a rotating or linear mechanical component.
Models and Learning
An encoder is a component of an AI model that transforms input information into an internal representation the model can use for further processing.
Robotics and Actuation
End-of-arm tooling, or EOAT, refers to the tools and devices attached to the working end of a robotic arm.
Models and Learning
An epoch is one complete pass through a training dataset during machine learning.
Navigation and Control
An error signal is the difference between a desired value and the value measured or estimated by a control system.
Software and Connectivity
EtherCAT, short for Ethernet for Control Automation Technology, is an industrial communication technology designed for fast and precisely coordinated data exchange between automation devices.
Software and Connectivity
EtherNet/IP is an industrial networking protocol that uses standard Ethernet and Internet Protocol technologies together with the Common Industrial Protocol, or CIP, to exchange information between automation devices.
Sensors and Perception
An event camera is a type of image sensor that reports changes in brightness at individual pixels rather than capturing complete images at regular intervals.
Sensors and Perception
Exteroception is the process of sensing information about the external environment.
Sensors and Perception
Extrinsic parameters describe the position and orientation of a camera or sensor relative to another coordinate system.
Industry and Safety
A fail-safe system is designed so that specified failures cause it to enter or maintain a condition that reduces the risk of harm.
Industry and Safety
A Factory Acceptance Test, or FAT, is a structured evaluation of equipment performed at the supplier's facility before it is shipped or installed at the customer's site.
Industry and Safety
Fault detection is the process of identifying when equipment, software, or a system is behaving abnormally or no longer satisfying defined operating requirements.
Industry and Safety
Fault tolerance is the ability of a system to continue performing a defined function even when one or more of its components fail.
Models and Learning
Federated learning is a machine learning approach that allows multiple devices or organizations to participate in training a shared model while keeping raw training data at its original location.
Foundations
A feedback loop is a process in which information about a system's actual behavior is used to influence what the system does next.
Navigation and Control
Feedforward control is a control method that calculates an action using a model, known inputs, or anticipated disturbances without waiting for an error to appear.
Models and Learning
Fine-tuning is the process of continuing the training of an existing machine learning model using additional data to improve its performance on a particular task or domain.
Software and Connectivity
Firmware is software stored on an electronic device that controls or supports its underlying hardware functions.
Sensors and Perception
A fisheye camera is a camera equipped with an ultra-wide-angle lens designed to capture a large portion of the surrounding scene.
Robotics and Actuation
Forward kinematics is the mathematical process of calculating the position and orientation of a robot's tool or another component from known joint positions.
Compute and Memory
Flash memory is a type of nonvolatile electronic memory that retains information even when electrical power is removed.
Compute and Memory
FLOPS stands for floating-point operations per second, a measure of how many floating-point calculations a computing system can perform in one second.
Software and Connectivity
A fleet management system, or FMS, is software used to coordinate, monitor, and manage a group of robots or vehicles.
Navigation and Control
Force control is a robotics control method that regulates the force a robot applies to an object or surface.
Models and Learning
A foundation model is an artificial intelligence model trained on a broad collection of data so that it can be used or adapted for many different tasks.
Sensors and Perception
Field of view, or FOV, describes the extent of the surrounding environment that a camera or sensor can observe.
Compute and Memory
FP16 is a 16-bit floating-point numerical format used to represent numbers in computing systems.
Compute and Memory
FP8 is a family of 8-bit floating-point numerical formats designed to represent numbers using less memory than formats such as FP16 or FP32.
Compute and Memory
A field-programmable gate array, or FPGA, is a semiconductor device whose internal digital logic can be configured after manufacturing.
Sensors and Perception
Frame rate is the number of individual images a camera captures or a video system processes each second.
Sensors and Perception
A force-torque sensor is a device that measures forces and twisting moments applied to a mechanical component.
Industry and Safety
Functional safety is the part of a system's overall safety that depends on safety-related equipment and control functions operating correctly when required.
Robotics and Actuation
A gantry robot is a robotic system that moves a tool or payload along a framework of rails, often positioned above or around a working area.
Robotics and Actuation
Gear ratio describes the relationship between the rotational speeds of connected components in a gear transmission.
Robotics and Actuation
A gearbox is a mechanical transmission that uses gears to change rotational speed, torque, or direction between a motor and a driven component.
Foundations
Generalization is the ability of an AI model to perform well on examples or situations it did not encounter exactly during training.
Industry and Safety
Geofencing is the use of a virtual geographic or spatial boundary to detect, restrict, or respond to the movement of a device or machine.
Navigation and Control
Geolocation is the process of determining the geographic or spatial location of a device, person, object, or observation.
Navigation and Control
A global planner is a component of a robot navigation system that calculates an overall route from the robot's current position to a destination.
Sensors and Perception
A global shutter is a camera exposure method in which all pixels in an image sensor capture light during the same time interval.
Navigation and Control
GNSS stands for Global Navigation Satellite System, the general term for satellite-based systems that provide positioning, navigation, and timing information.
Compute and Memory
A graphics processing unit, or GPU, is a processor designed to perform large numbers of calculations in parallel.
Industry and Safety
Greenfield deployment is the introduction of technology, equipment, or infrastructure into a new facility or system that is being developed from the ground up.
Robotics and Actuation
A gripper is a robotic end effector designed to hold, move, or manipulate objects.
Simulation and Data
Ground truth is the best available information about the actual or correct answer to a measurement, observation, or prediction task.
Sensors and Perception
A gyroscope is a sensor that measures angular velocity, or how quickly an object rotates around an axis.
Sensors and Perception
A Hall effect sensor is an electronic device that detects magnetic fields using a physical phenomenon called the Hall effect.
Robotics and Actuation
A Harmonic Drive is a type of precision speed-reduction mechanism based on strain-wave gearing, which uses the controlled deformation of a flexible gear element.
Industry and Safety
A hazard is a source, situation, or condition with the potential to cause harm.
Industry and Safety
Hazard analysis is the systematic process of identifying potential sources of harm and examining the situations in which they could cause injury or damage.
Compute and Memory
High-bandwidth memory, or HBM, is a type of computer memory designed to transfer large amounts of data between memory and processing hardware at very high rates.
Compute and Memory
A heat sink is a component designed to absorb and transfer heat away from electronic or mechanical equipment.
Compute and Memory
Heterogeneous computing is an approach to computing that combines different types of processors or computing resources within one system.
Simulation and Data
Hardware-in-the-loop, or HIL, is a testing method that connects real electronic control hardware to a simulated machine or physical process.
Foundations
Human-in-the-loop, or HITL, describes a system or workflow in which people actively contribute to decisions, training, or operations performed by automated systems.
Industry and Safety
A human-machine interface, or HMI, is the system through which a person monitors, communicates with, or controls a machine.
Foundations
Human-on-the-loop, or HOTL, describes an oversight approach in which an automated system performs defined tasks without requiring a person to approve every action, while humans retain a supervisory role and the ability to intervene.
Foundations
Human-robot interaction, or HRI, is the study and design of how people and robots communicate, cooperate, share environments, and respond to one another.
Robotics and Actuation
A humanoid robot is a robot designed with a body structure or appearance resembling that of a human.
Robotics and Actuation
A hydraulic actuator is a device that converts energy from pressurized liquid into mechanical movement or force.
Robotics and Actuation
Inverse kinematics, or IK, is the mathematical process of determining which joint positions will place a robot's tool at a desired position and orientation.
Sensors and Perception
Image classification is a computer vision task that assigns one or more categories to an image or selected image region.
Sensors and Perception
Image resolution describes the number of pixels used to represent a digital image, typically expressed as its width multiplied by its height.
Models and Learning
Imitation learning is a machine learning approach in which a system learns behavior from examples of how a task should be performed.
Navigation and Control
Impedance control is a robotics control method that defines how a robot should respond to differences in movement and applied forces, often by making it behave like a virtual spring, damper, and mass.
Sensors and Perception
An inertial measurement unit, or IMU, is a sensor module that measures information about a device's motion and rotation.
Sensors and Perception
An incremental encoder is a sensor that produces electrical signals representing changes in position as a shaft rotates or a component moves.
Foundations
Industrial automation is the use of control systems, machines, and software to perform manufacturing or industrial processes with reduced direct human intervention.
Models and Learning
Inference is the process of using a trained artificial intelligence model to generate an output from new input information.
Compute and Memory
An inference engine is software that executes a trained artificial intelligence model to produce outputs from input data.
Compute and Memory
InfiniBand is a high-performance networking technology designed to connect computing systems with high bandwidth and low communication latency.
Compute and Memory
INT8 is an eight-bit integer numerical format that represents whole-number values using eight bits of information.
Foundations
An intelligent machine is a machine that uses computing, sensing, or artificial intelligence to interpret information and make decisions about its behavior.
Compute and Memory
An interconnect is a communication pathway that transfers data between components of a computing system.
Software and Connectivity
Interoperability is the ability of different systems, devices, or applications to exchange information and work together effectively.
Sensors and Perception
Intrinsic parameters are numerical values describing how a camera projects three-dimensional points onto its image sensor.
Compute and Memory
Jitter is variation in the timing of repeated events, communications, or computing operations.
Navigation and Control
A Kalman filter is a mathematical algorithm that estimates the changing state of a system by combining predictions with imperfect measurements.
Robotics and Actuation
Kinematics is the study of motion using position, orientation, velocity, and acceleration without directly analyzing the forces that produce that motion.
Models and Learning
Knowledge distillation is a machine learning technique in which a model is trained using information produced by another model, often with the goal of transferring capabilities from a larger model to a smaller one.
Compute and Memory
Latency is the delay between an input, request, or event and the corresponding response from a system.
Models and Learning
Latent space is a numerical representation in which an AI model organizes underlying patterns or characteristics learned from data.
Models and Learning
Learning rate is a setting in machine learning that controls the size of the adjustments made to a model's parameters during training.
Robotics and Actuation
A legged robot is a robot that moves using two or more articulated legs rather than relying exclusively on wheels or tracks.
Foundations
Level of autonomy describes how much of a defined task a machine can perform without direct human intervention.
Sensors and Perception
LiDAR, short for Light Detection and Ranging, is a sensing technology that uses light to measure distances to objects and surfaces.
Industry and Safety
A safety light curtain is a protective device that uses a field of light beams to detect when a person or object enters a monitored area.
Robotics and Actuation
A linear actuator is a device that produces movement along a straight line.
Robotics and Actuation
A linear motor is an electric motor designed to produce movement directly along a straight path instead of rotating a shaft.
Models and Learning
A large language model, or LLM, is an artificial intelligence model trained on large amounts of language-related data to process and generate text or other token sequences.
Navigation and Control
A local planner is a component of a robot navigation system that determines short-term movement using the robot's current state, nearby obstacles, and navigation objectives.
Navigation and Control
Localization is the process of estimating a robot's position and, often, orientation relative to a map or reference coordinate system.
Navigation and Control
Loop closure is the process by which a robot's mapping system recognizes that it has returned to a previously visited location.
Models and Learning
A loss function is a mathematical calculation that measures how well an AI model's output matches a desired result or training objective.
Compute and Memory
A multiply-accumulate operation, commonly abbreviated MAC, is a mathematical operation that multiplies two numbers and adds the result to an accumulated value.
Industry and Safety
Machine tending is the process of loading, unloading, or otherwise supporting production equipment during its operation.
Sensors and Perception
Machine vision is the use of imaging hardware and software to inspect, measure, identify, or guide processes performed by machines.
Sensors and Perception
A magnetometer is a sensor that measures magnetic fields.
Robotics and Actuation
A robot manipulator is a mechanical system designed to position, orient, or move a tool or object within a workspace.
Navigation and Control
Map matching is the process of comparing estimated positions or observed environmental features with a known map to determine the most likely corresponding location.
Navigation and Control
Mapping is the process of creating a representation of an environment using observations from sensors or other information sources.
Compute and Memory
Matrix multiplication is a mathematical operation that combines two matrices, or rectangular arrangements of numbers, according to defined multiplication and addition rules.
Compute and Memory
A microcontroller unit, or MCU, is a compact computer integrated into a single chip, typically containing a processor, memory, and interfaces for communicating with electronic devices.
Compute and Memory
Memory bandwidth measures how much information can be transferred between memory and a processor during a given period.
Compute and Memory
Memory capacity is the amount of information a memory system can store at one time.
Compute and Memory
A memory controller is a component that manages communication between a processor or computing system and its memory.
Compute and Memory
Memory hierarchy is the arrangement of computing memory and storage into different levels according to characteristics such as speed, capacity, cost, and proximity to the processor.
Industry and Safety
A manufacturing execution system, or MES, is software used to monitor, coordinate, and document manufacturing operations on the factory floor.
Software and Connectivity
A message broker is software that receives messages from applications or devices and routes them to the appropriate recipients.
Sensors and Perception
A microphone is a sensor that converts sound waves into electrical signals that electronic systems can process.
Sensors and Perception
A microphone array is a group of microphones arranged so their recordings can be processed together.
Software and Connectivity
Middleware is software that helps different applications, devices, or computing components communicate and work together.
Compute and Memory
Mixed precision is a computing technique that uses different numerical formats within the same workload to balance performance, memory consumption, and numerical accuracy.
Foundations
Machine learning, or ML, is a branch of artificial intelligence that develops models capable of learning patterns from data and using those patterns to make predictions or decisions.
Robotics and Actuation
A mobile manipulator is a robot that combines a mobile base with one or more robotic arms or manipulation mechanisms.
Robotics and Actuation
A mobile robot is a robot capable of moving through its environment rather than remaining fixed in one location.
Software and Connectivity
Modbus is an industrial communication protocol used to exchange information between electronic devices, controllers, sensors, and software systems.
Models and Learning
Model-based reinforcement learning is a machine learning approach in which an agent uses a model of its environment to help learn or select actions.
Models and Learning
A model checkpoint is a saved record of a machine learning model's state at a particular stage of training.
Models and Learning
Model compression is the process of reducing the memory, storage, or computational resources required by a machine learning model.
Models and Learning
Model-free reinforcement learning is a machine learning approach in which an agent learns useful behavior from experience without explicitly using a model that predicts how the environment changes after each action.
Models and Learning
A model parameter is a numerical value within a machine learning model that determines part of how the model processes information.
Models and Learning
Model weights are numerical values within a machine learning model that influence how it processes information and generates predictions.
Sensors and Perception
A monocular camera is a camera that observes a scene from a single viewpoint.
Navigation and Control
Motion planning is the process of determining how a robot can move from one configuration or location to another while satisfying relevant constraints.
Robotics and Actuation
A motor controller is an electronic or software-based system that regulates the operation of an electric motor.
Robotics and Actuation
A motor driver is an electronic circuit or device that delivers controlled electrical power to a motor.
Navigation and Control
Model predictive control, or MPC, is a control method that uses a mathematical model to predict how a system may behave over a future time period and calculate suitable control actions.
Software and Connectivity
MQTT is a lightweight messaging protocol used to exchange information between devices and applications through a publish-subscribe communication model.
Models and Learning
Multimodal AI refers to artificial intelligence systems capable of processing or generating more than one type of information, such as text, images, audio, video, or sensor measurements.
Navigation and Control
A navigation stack is a collection of software components that work together to help a mobile robot determine its location, plan routes, avoid obstacles, and execute movement commands.
Software and Connectivity
Network bandwidth is the amount of data a communication connection or network can carry per unit of time, commonly expressed in bits per second.
Software and Connectivity
Network topology describes how devices and communication connections are arranged within a computer network.
Models and Learning
A neural network is a machine learning model made up of interconnected mathematical units that process information and learn patterns from data.
Compute and Memory
A neural processing unit, or NPU, is a processor designed to accelerate calculations commonly used by artificial neural networks.
Compute and Memory
NVLink is NVIDIA's high-speed interconnect technology designed to exchange information between compatible computing processors, particularly GPUs.
Sensors and Perception
Object detection is a computer vision task that identifies objects in an image and estimates where they are located.
Navigation and Control
Obstacle avoidance is the process of detecting obstacles and selecting movement that reduces the likelihood of unwanted contact.
Navigation and Control
An occupancy grid is a map that divides an environment into small cells and represents whether each cell is occupied, free, or unknown.
Industry and Safety
An operational design domain, or ODD, defines the specific conditions under which an automated system is designed to operate.
Industry and Safety
Overall Equipment Effectiveness, or OEE, is a manufacturing performance metric that measures how effectively equipment produces acceptable output during planned production time.
Robotics and Actuation
Omnidirectional drive is a mobile robot drive arrangement that allows movement in multiple planar directions, including sideways movement without first turning the robot.
Software and Connectivity
On-device AI is the execution of artificial intelligence models directly on the device using them, rather than depending on a remote server for every prediction or decision.
Models and Learning
Online learning is a machine learning approach in which a model updates its parameters incrementally as new training examples become available.
Software and Connectivity
OPC UA, short for Open Platform Communications Unified Architecture, is an industrial communication standard designed to allow machines, controllers, and software systems to exchange structured information.
Navigation and Control
Open-loop control is a control method in which a system issues commands without using measurements of the resulting output to correct those commands.
Compute and Memory
OpenCL, short for Open Computing Language, is an open standard for programming parallel computing workloads across different types of processors and accelerators.
Industry and Safety
An over-the-air update, or OTA update, is the delivery of software, firmware, or other digital changes to a device through a communication network.
Navigation and Control
Overshoot occurs when a system temporarily moves beyond its intended target or final value before settling.
Software and Connectivity
Packet loss occurs when information divided into network packets fails to reach its intended destination.
Industry and Safety
Palletizing is the process of arranging and stacking products onto pallets for storage, handling, or transportation.
Compute and Memory
Parallel computing is an approach to computing in which multiple calculations or tasks are performed at the same time using several processing resources.
Navigation and Control
A particle filter is a mathematical algorithm that estimates the state of a system by maintaining many possible explanations of where or how the system is operating.
Navigation and Control
Path planning is the process of finding a route from a starting location to a destination while accounting for obstacles and other constraints.
Industry and Safety
Payback period is the amount of time required for an investment's accumulated net cash benefits to recover its initial cost.
Robotics and Actuation
Robot payload is the load a robot is designed or rated to carry under specified operating conditions.
Compute and Memory
Peripheral Component Interconnect Express, commonly called PCIe, is a high-speed communication standard used to connect components inside computing systems.
Foundations
Perception is the process of interpreting information from sensors to estimate what is happening in an environment or within a machine.
Industry and Safety
Power and force limiting, or PFL, is an approach to collaborative robot operation that limits the mechanical effects of contact between a robot application and a person.
Sensors and Perception
A photodiode is a semiconductor device that converts incoming light into an electrical signal, usually measured as current.
Simulation and Data
Photorealistic rendering is the process of generating computer-created images intended to resemble photographs of real scenes.
Foundations
Physical AI refers to artificial intelligence systems that perceive, reason about, or act within the physical world through machines, sensors, and other connected technologies.
Foundations
Physical intelligence is the ability of a system to understand and act within the constraints of the physical world.
Simulation and Data
A physics engine is software that calculates how virtual objects move and interact according to mathematical models of physical behavior.
Industry and Safety
Pick and place is an automation task in which a robot collects an object from one location and moves it to another.
Navigation and Control
A proportional-integral-derivative controller, or PID controller, is a feedback-control method that adjusts a system using its current error, accumulated past error, and rate of error change.
Industry and Safety
Performance Level, or PL, is a classification used by ISO 13849 to describe the capability of safety-related parts of a machine control system to perform a defined safety function under specified conditions.
Foundations
Planning is the process of selecting a sequence of actions intended to move a system from its current situation toward a desired goal.
Industry and Safety
A programmable logic controller, or PLC, is an industrial computer designed to monitor inputs and control machinery or industrial processes.
Robotics and Actuation
A pneumatic actuator is a device that uses compressed air or another pressurized gas to produce mechanical movement.
Sensors and Perception
A point cloud is a collection of points representing locations in three-dimensional space.
Models and Learning
A policy is a rule or computational model that determines which action an agent should take based on its current state or observations.
Navigation and Control
Pose describes the position and orientation of an object or robot relative to a reference coordinate system.
Navigation and Control
Position control is a control method used to move a mechanism toward a desired location or angle and regulate its position.
Models and Learning
Post-training is the collection of methods used to adapt, improve, or refine an AI model after its initial pretraining phase.
Compute and Memory
Power efficiency describes how effectively a machine or computing system converts consumed electrical power into useful output.
Industry and Safety
Predictive maintenance is a maintenance approach that uses information about equipment condition and performance to estimate when problems may develop or maintenance may be required.
Sensors and Perception
A pressure sensor is a device that measures the pressure exerted by a gas or liquid.
Models and Learning
Pretraining is the initial training phase in which an AI model learns useful patterns and representations from a broad or foundational dataset before being adapted to particular tasks.
Software and Connectivity
Private 5G is a cellular network deployed for the dedicated use of an organization, facility, or defined group of users.
Software and Connectivity
PROFINET is an industrial Ethernet communication standard used to exchange information between automation controllers, sensors, drives, and other equipment.
Sensors and Perception
Proprioception is the ability of a system to sense information about its own physical state, including the positions and movements of its components.
Software and Connectivity
A protocol is a set of rules that defines how devices or software systems exchange information.
Sensors and Perception
A proximity sensor is a device that detects the presence or approach of an object without requiring direct physical contact.
Models and Learning
Pruning is a model-optimization technique that removes selected weights, connections, or structures from a machine learning model.
Software and Connectivity
Publish-subscribe, often abbreviated pub-sub, is a communication pattern in which software components publish messages to named topics while other components subscribe to receive relevant information.
Models and Learning
Q-learning is a reinforcement learning algorithm that learns how valuable different actions are in particular situations.
Robotics and Actuation
Quasi-direct drive, or QDD, is an actuator design that combines a motor capable of producing substantial torque with a relatively low-ratio mechanical transmission.
Software and Connectivity
Quality of Service, or QoS, refers to mechanisms and settings used to manage how network or software communications handle requirements such as reliability, delay, priority, and message delivery.
Robotics and Actuation
A quadruped robot is a legged robot that moves using four articulated legs.
Industry and Safety
Quality inspection is the process of examining, measuring, or testing products and processes to determine whether they meet specified requirements.
Models and Learning
Quantization is the process of representing numerical values using a reduced set of possible values or lower-precision numerical formats.
Sensors and Perception
Radar, short for Radio Detection and Ranging, is a sensing technology that uses radio waves to detect objects and measure their distance, direction, or movement.
Simulation and Data
Ray tracing is a rendering technique that calculates how rays of light interact with objects and surfaces in a virtual environment.
Robotics and Actuation
Robot reach is a specification describing how far a robotic mechanism can extend its working point from a defined reference location.
Navigation and Control
Real-time computing is an approach to computing in which correctly completing an operation includes meeting a defined timing requirement.
Foundations
Reasoning is the process of using information, relationships, and rules to draw conclusions, evaluate possibilities, or select actions.
Industry and Safety
Redundancy is the use of additional components, systems, or methods to support a function when one element fails or becomes unreliable.
Robotics and Actuation
Robot repeatability measures how consistently a robot returns to the same commanded position or pose when performing repeated movements under specified conditions.
Models and Learning
A reward model is a machine learning model that estimates how desirable an output, action, or result is according to a learned evaluation objective.
Sensors and Perception
An RGB camera is a digital camera that captures color information using red, green, and blue channels.
Sensors and Perception
An RGB-D camera is a sensing device that provides both color images and depth information about the surrounding environment.
Simulation and Data
Rigid body simulation is a computational method that models the movement and physical interaction of objects assumed not to change shape.
Industry and Safety
Risk assessment is the systematic process of identifying hazards, evaluating associated risks, and determining appropriate measures to reduce those risks.
Models and Learning
Reinforcement learning, or RL, is a machine learning approach in which an agent learns to make decisions through interaction with an environment and feedback about the results of its actions.
Models and Learning
Reinforcement learning from human feedback, or RLHF, is a model-training approach that uses human evaluations or preferences to help guide the improvement of an AI system.
Foundations
A robot is a programmed mechanical system capable of performing movement, positioning, or manipulation with some degree of autonomy.
Robotics and Actuation
A robot arm is a mechanical system made of connected links and joints that positions, moves, or orients a tool or object.
Simulation and Data
A robot demonstration is a recorded example of how a task is performed, often including observations, movements, and other information relevant to the task.
Robotics and Actuation
A robot joint is a mechanical connection between parts of a robot that allows or constrains relative movement.
Foundations
Robotics is the field of science and engineering concerned with designing, building, programming, controlling, and operating robots.
Simulation and Data
A robotics simulator is software that creates a virtual environment for developing, testing, and evaluating robots without requiring every experiment to use physical hardware.
Compute and Memory
ROCm, short for Radeon Open Compute, is AMD's open software platform for GPU-accelerated computing.
Industry and Safety
Return on investment, or ROI, is a financial measure that compares the net benefit of an investment with its cost.
Sensors and Perception
A rolling shutter is a camera exposure method in which different rows of an image sensor capture light at slightly different times.
Software and Connectivity
ROS 2, short for Robot Operating System 2, is an open-source software framework and ecosystem used to develop robotic applications.
Software and Connectivity
A ROS action is a communication interface used to manage tasks that take time to complete and may need progress feedback or cancellation.
Software and Connectivity
A ROS bag is a recorded collection of messages from a Robot Operating System application.
Software and Connectivity
A ROS node is a software component within the Robot Operating System that performs one or more defined functions.
Software and Connectivity
A ROS service is a communication interface that allows one software component to send a request to another component and receive a response.
Software and Connectivity
A ROS topic is a named communication channel used by software components to exchange messages through publish-subscribe communication.
Software and Connectivity
ROS, short for Robot Operating System, is an open-source software ecosystem that provides tools, libraries, and communication mechanisms for building robots.
Robotics and Actuation
A rotary actuator is a device that converts energy into rotational movement or torque.
Navigation and Control
Route planning is the process of selecting a route between locations using information such as distance, available paths, restrictions, and operating objectives.
Navigation and Control
Real-Time Kinematic positioning, or RTK, is a satellite positioning technique that uses correction information and precise satellite-signal measurements to improve location accuracy.
Navigation and Control
A real-time operating system, or RTOS, is an operating system designed to support tasks with predictable timing requirements.
Compute and Memory
A runtime is the software environment or supporting system used to execute a program or computational workload.
Industry and Safety
A safety laser scanner is a protective sensing device that monitors a defined area using laser-based distance measurements and initiates a safety-related response when specified conditions are detected.
Industry and Safety
A safety programmable logic controller, or safety PLC, is an industrial controller designed for defined safety-related control functions.
Industry and Safety
A site acceptance test, or SAT, is a structured test performed after equipment has been installed at its intended operating location.
Industry and Safety
SCADA stands for Supervisory Control and Data Acquisition, a category of software and supporting systems used to monitor and supervise industrial processes.
Robotics and Actuation
SCARA stands for Selective Compliance Assembly Robot Arm, a type of industrial robot particularly suited to fast, repetitive operations involving horizontal movement and vertical placement.
Simulation and Data
A scenario is a defined situation used to develop, train, or test a system under particular conditions.
Simulation and Data
Scenario generation is the process of creating operating situations for training, testing, or evaluating automated systems.
Software and Connectivity
A software development kit, or SDK, is a collection of tools, libraries, documentation, and other resources that help developers build software for a particular platform, device, or service.
Robotics and Actuation
A series elastic actuator, or SEA, is an actuator designed with an elastic element placed in series between the motor or transmission and the load.
Models and Learning
Self-supervised learning is a machine learning approach in which training objectives are created from the structure or content of the data itself.
Sensors and Perception
Semantic segmentation is a computer vision task that assigns a category to each pixel in an image.
Compute and Memory
A semiconductor process node is a designation for a generation of chip manufacturing technology.
Sensors and Perception
A sensor is a device that detects or measures a physical, chemical, or environmental property and produces a signal representing that information.
Sensors and Perception
Sensor calibration is the process of establishing how a sensor's measurements relate to known reference values or conditions.
Sensors and Perception
Sensor fusion is the process of combining information from multiple sensors to produce a more useful estimate of a system's state or environment.
Sensors and Perception
Sensor noise is unwanted variation or interference in a sensor's measurements that does not directly represent the physical quantity being measured.
Navigation and Control
Sensor registration is the process of aligning measurements from different sensors or observations into a common spatial reference.
Robotics and Actuation
A servo motor is a motor used within a controlled motion system that uses feedback to regulate position, speed, or torque.
Navigation and Control
A setpoint is the desired value that a control system is intended to achieve or maintain.
Navigation and Control
Settling time is the amount of time a system takes to reach and remain within a specified tolerance around its final target or steady-state value after a change.
Industry and Safety
Safety Integrity Level, or SIL, is a classification used in functional safety to describe the integrity required of a safety-related function.
Simulation and Data
Software-in-the-loop, or SIL, is a testing method in which actual or representative application software is executed against simulated components of a system.
Compute and Memory
A silicon interposer is a thin piece of silicon used in advanced semiconductor packaging to create dense electrical connections between multiple chips.
Models and Learning
Sim-to-real transfer is the process of taking a robot model, control strategy, or learned behavior developed in simulation and making it work on physical equipment.
Simulation and Data
Simulation is the use of mathematical or computational models to reproduce selected characteristics of a real system, process, or environment.
Compute and Memory
System-in-Package, or SiP, is a semiconductor packaging approach that combines multiple electronic components into one compact package or module.
Foundations
Situated intelligence is the idea that intelligent behavior depends on the situation in which an agent operates, including its environment, available information, physical capabilities, and immediate objectives.
Navigation and Control
SLAM stands for Simultaneous Localization and Mapping, a robotics technique that allows a machine to construct or update a map while estimating its own position within that map.
Compute and Memory
A System-on-Chip, or SoC, is an integrated circuit that combines several computing and electronic functions onto one semiconductor chip.
Simulation and Data
Soft body simulation is a computational method used to represent objects that deform when forces act upon them.
Robotics and Actuation
Soft robotics is a field of robotics that develops machines and mechanisms using flexible, compliant, or deformable materials.
Robotics and Actuation
A solenoid is an electromagnetic device that uses an electrical coil to generate a magnetic field.
Models and Learning
Sparsity describes a representation in which many values are zero or absent.
Foundations
Spatial intelligence is the ability to interpret and reason about the positions, shapes, orientations, distances, and relationships of objects in space.
Compute and Memory
SRAM, short for Static Random Access Memory, is a type of semiconductor memory that stores information while electrical power remains available without requiring the periodic refresh operations used by DRAM.
Industry and Safety
A safety-rated monitored stop, or SRMS, is a safety function historically described in industrial robot standards in which a robot stops hazardous movement and its stopped condition is monitored by a safety-related system.
Industry and Safety
Speed and separation monitoring, or SSM, is a collaborative robotics safety approach that maintains an appropriate protective distance between people and hazardous robot movement.
Navigation and Control
Stability describes how a system behaves when its state changes or it experiences a disturbance.
Navigation and Control
State estimation is the process of using measurements and mathematical models to estimate important conditions of a system that may not be directly or perfectly observable.
Robotics and Actuation
A stepper motor is an electric motor designed to rotate through discrete angular increments, commonly called steps.
Sensors and Perception
Stereo vision is a computer vision method that estimates depth by comparing images captured from two different viewpoints.
Robotics and Actuation
Stiffness describes how strongly a mechanical structure or system resists deformation when force or torque is applied.
Industry and Safety
Safe Torque Off, or STO, is a safety-related function that prevents a compatible motor drive from generating torque through its normal motor-power control.
Sensors and Perception
A strain gauge is a sensor that measures deformation in a material by detecting changes in an electrical property, commonly resistance.
Sensors and Perception
Structured light is a three-dimensional sensing technique that projects a known light pattern onto an object or scene and analyzes how the pattern appears from a different viewpoint.
Models and Learning
Supervised learning is a machine learning approach in which an AI model learns from examples containing input information and corresponding target outputs.
Simulation and Data
Synthetic data is information generated artificially rather than recorded directly from the real-world events or objects it represents.
Simulation and Data
A synthetic environment is a digitally constructed representation of a setting used for simulation, training, testing, or visualization.
Simulation and Data
System identification is the process of developing a mathematical model of a system using measured information about its inputs and outputs.
Industry and Safety
A systems integrator is a company or specialist that combines different technologies, equipment, and software into a complete functioning system.
Sensors and Perception
A tactile sensor is a device that measures aspects of physical contact, such as pressure, deformation, vibration, or contact location.
Industry and Safety
Takt time is the rate at which a manufacturing process needs to produce finished units to meet customer demand within the available production time.
Industry and Safety
Total cost of ownership, or TCO, is an estimate of the complete cost of acquiring, operating, maintaining, and eventually replacing or disposing of an asset over a defined period.
Compute and Memory
Thermal Design Power, or TDP, is a processor specification used to guide the design of an appropriate cooling system.
Industry and Safety
Telemetry is the collection and transmission of measurements or operating information from a device to another system for monitoring or analysis.
Simulation and Data
A telemetry log is a recorded history of measurements, operating conditions, and events collected from a machine or system.
Simulation and Data
Teleoperation is the control of a machine or robot by a person located away from the machine itself.
Sensors and Perception
A temperature sensor is a device that measures temperature and produces information that an electronic system can interpret.
Compute and Memory
A Tensor Core is a specialized computing unit found in certain NVIDIA GPUs that accelerates matrix operations commonly used in artificial intelligence.
Simulation and Data
A test set is a collection of examples reserved for evaluating how a trained machine learning model performs on data that was not used to train or select it.
Sensors and Perception
A thermal camera is an imaging device that detects infrared radiation associated with the thermal state of objects and surfaces.
Compute and Memory
Thermal throttling is the automatic reduction of a processor's operating performance to help keep its temperature within specified limits.
Compute and Memory
Throughput is the amount of work a system completes during a specified period.
Simulation and Data
Time series data is a sequence of observations recorded in time order.
Sensors and Perception
A time-of-flight sensor, or ToF sensor, estimates distance by measuring the travel time of an emitted signal between the sensor and a target.
Models and Learning
A token is a unit of information that an artificial intelligence model processes.
Compute and Memory
TOPS stands for trillions of operations per second, a measure of the rate at which computing hardware can perform specified operations.
Robotics and Actuation
Torque is the turning effect of a force applied around an axis.
Navigation and Control
Torque control is a control method that regulates the torque produced by a motor or robotic joint.
Compute and Memory
A Tensor Processing Unit, or TPU, is a specialized processor developed by Google to accelerate machine learning workloads.
Models and Learning
Training is the process through which a machine learning model acquires or improves capabilities using data, examples, or experience.
Simulation and Data
A training set is the portion of a dataset used to adjust a machine learning model's parameters during training.
Navigation and Control
Trajectory planning is the process of creating a time-dependent description of how a robot should move.
Models and Learning
Transfer learning is a machine learning approach that uses knowledge acquired from one task or dataset to help improve performance on another.
Models and Learning
A transformer is a neural network architecture that uses attention mechanisms to process relationships between different parts of its input.
Software and Connectivity
Time-Sensitive Networking, or TSN, is a family of IEEE networking standards designed to make communication over Ethernet-based networks more predictable.
Robotics and Actuation
An uncrewed aerial vehicle, or UAV, is an aircraft operated without a human pilot physically onboard.
Robotics and Actuation
An uncrewed ground vehicle, or UGV, is a vehicle that operates on the ground without a human physically onboard.
Sensors and Perception
An ultrasonic sensor uses high-frequency sound waves to detect objects or estimate their distance.
Compute and Memory
Unified memory is a computing approach that allows different processors, such as CPUs and GPUs, to access information through a shared memory programming model or memory arrangement.
Models and Learning
Unsupervised learning is a machine learning approach that identifies patterns or structure in data without relying on a correct target label for every example.
Industry and Safety
Validation is the process of determining whether a system meets the needs and intended use of its users or application.
Simulation and Data
A validation set is a collection of examples used during machine learning development to evaluate model performance and guide decisions about training or model selection.
Robotics and Actuation
A valve is a mechanical or electromechanical device that regulates the flow, direction, or pressure of a fluid.
Navigation and Control
Velocity control is a control method that regulates how quickly a motor, joint, wheel, or other moving component travels.
Simulation and Data
Virtual commissioning is the process of testing and evaluating industrial automation systems using digital models before or alongside installation of the physical equipment.
Models and Learning
A Vision-Language-Action model, or VLA, is an artificial intelligence model that connects visual observations and language instructions with representations of physical actions.
Models and Learning
A Vision-Language Model, or VLM, is an artificial intelligence model that connects visual information with language.
Navigation and Control
Visual odometry, or VO, is the process of estimating a camera's movement by analyzing changes across a sequence of images.
Compute and Memory
A Vision Processing Unit, or VPU, is a specialized processor designed to accelerate computer vision and related image-processing workloads.
Compute and Memory
VRAM, short for Video Random Access Memory, is the memory associated with a graphics processor for storing information needed during its operation.
Models and Learning
A World Action Model, or WAM, is an emerging type of robotics AI model that connects predictions about how the world changes with the actions a robot can take.
Navigation and Control
A waypoint is a defined location that a robot, drone, or vehicle is instructed to reach during navigation.
Models and Learning
A World Foundation Model, or WFM, is an AI model pretrained on broad information about physical scenes, events, and interactions so its capabilities can be adapted for multiple world-modeling tasks.
Sensors and Perception
Wheel odometry is a method of estimating a mobile robot's movement using measurements of how its wheels rotate.
Robotics and Actuation
A wheeled robot is a mobile robot that uses wheels as its primary means of movement.
Software and Connectivity
Wi-Fi is a family of wireless networking technologies based primarily on IEEE 802.11 standards.
Robotics and Actuation
A robot's workspace is the collection of positions or poses that its mechanism can reach within its mechanical constraints.
Industry and Safety
Zero trust is a cybersecurity approach based on the principle that users, devices, and applications should not receive implicit access simply because they are connected to an organization's internal network.
