Automation is the use of machines, control systems, and software to perform tasks or operate processes with reduced direct human intervention. It encompasses physical operations, such as manufacturing and temperature regulation, and information-processing activities, such as routing documents or executing transactions. Automation may handle a single operation or coordinate an entire production or service workflow. It does not necessarily eliminate human participation: people commonly define objectives, configure systems, maintain equipment, and manage exceptions. The International Society of Automation emphasizes technology that monitors and controls the production and delivery of products and services. (isa.org)
Historical development
Important precursors emerged during the Industrial Revolution. A 1788 steam engine built by James Watt incorporated a centrifugal governor, illustrating automatic regulation through a mechanical mechanism. In 1801, Joseph Marie Jacquard demonstrated a loom whose weaving patterns were controlled by punched cards. The loom separated instructions from the machinery executing them: changing the cards changed the pattern without rebuilding the mechanism. It remained human-powered, demonstrating that automatic control and mechanical power are distinct features. (collection.sciencemuseumgroup.org.uk)
Later systems extended automatic control through electronic computation and programmable equipment. Modern industrial installations use digital controllers to execute sequences, regulate processes, and communicate with supervisory systems. In offices and services, computers enable automation of information handling, while software workflows coordinate operations across applications. These developments broadened automation beyond the movement of physical machinery. (nvlpubs.nist.gov)
Operating principles
A central foundation is control theory, which studies how systems can be directed toward desired behavior. In a typical control loop, a sensor measures the process, a controller computes an action, and an actuator changes the process. Feedback returns information about the resulting behavior to the controller. Measurements may concern temperature, speed, position, or other operating variables. The controller uses an algorithm to determine an appropriate response. (cds.caltech.edu)
An open-loop system acts without using feedback from the controlled output to correct its operation. A closed-loop system measures that output and adjusts its action in response. For example, thermostatic heating uses measured temperature to regulate heating around a desired value. Feedback can improve resistance to disturbances and uncertainty, but poorly designed feedback can also cause instability or introduce measurement noise into system behavior. (cds.caltech.edu)
Automation also includes discrete sequencing rather than continuous regulation. A programmable logic controller (PLC) can execute logic, timing, counting, and input–output instructions for machinery. In larger installations, local controllers operate within distributed control systems or communicate with supervisory control and data acquisition systems. Local control and supervisory monitoring serve different, complementary functions. (nvlpubs.nist.gov)
Industrial and software applications
Industrial automation combines measurement, control, communications, and machinery in manufacturing, utilities, transportation, and buildings. Robotics is one component, particularly where equipment must manipulate objects or perform physical operations. However, many automated installations are not robots: process controllers and building-management systems can operate automatically without articulated mechanical arms. Applications include chemical production, wastewater treatment, environmental control, and energy management. (isa.org)
Software automation executes operations on digital information. Business process management coordinates workflows involving people and applications, while robotic process automation uses software bots to perform repetitive tasks such as entering data, moving files, or interacting with application interfaces. Despite its name, robotic process automation does not require a physical robot. Bots may run unattended or assist a person during an activity; their scope is usually defined by executable instructions and available application access. (ibm.com)
Automation does not inherently require artificial intelligence. Rule-based workflows and conventional controllers can operate without learning from data. AI-supported automation adds capabilities such as prediction, classification, and document interpretation. Machine learning, computer vision, and natural language processing can help systems process information that is less readily handled by fixed rules. These capabilities are often combined with conventional workflow and execution mechanisms rather than replacing them. (ibm.com)
Economic and organizational effects
Automation can increase productivity by reducing the time or resources required for particular tasks and by improving consistency. Its effects on the labor market depend on how work is organized. Replacing a task can reduce demand for the workers performing it, while lower production costs can expand demand for output and for complementary work. Installation, monitoring, maintenance, and process improvement can also create or reshape tasks. Consequently, technical capability alone does not determine the net employment effect. (oecd.org)
An occupation normally comprises multiple tasks, not all equally suitable for automation. OECD research therefore distinguishes the automatability of particular skills and activities from the disappearance of an entire job. Work may be reorganized even when an occupation remains, changing the balance between routine execution, judgment, interpersonal activity, and technical supervision. Estimates of automation exposure describe potential technological effects, not certain forecasts of unemployment. (oecd.org)
Reliability, safety, and security
Automation that interacts with the physical environment is commonly classified as operational technology. Such systems have performance, reliability, and safety requirements that can differ from those of ordinary office computing. A failure may affect equipment, physical processes, or service availability rather than only digital information. System design therefore encompasses control behavior, operating constraints, maintenance, and recovery as well as task execution. (nist.gov)
Connectivity introduces cybersecurity concerns because unauthorized access or altered instructions can interfere with monitoring and control. Security measures must account for operational requirements: controls suitable for conventional information systems can disrupt industrial performance if applied without adaptation. Human responsibilities consequently remain significant, including system configuration, oversight, investigation of abnormal behavior, and management of changes to control logic. (nccoe.nist.gov)