Robotics is the interdisciplinary field concerned with the design, construction, programming, control, and evaluation of robots: physical machines that perform tasks through controlled movement and interaction with their surroundings. It combines mechanical engineering, electrical engineering, computer science, and control theory, with contributions from mathematics and artificial intelligence. Its central problems include representing the environment, choosing actions, and executing them despite physical constraints and uncertainty. Robotics encompasses both autonomous systems and machines operating under human supervision or remote control. (roboticsbook.org)
Scope and historical development
Robotics overlaps with automation, but the two are not identical. Automation includes processes that need no robotic mechanism, while robotics addresses machines whose motion and physical interaction are essential to their function. Robots need not resemble humans, and their capabilities may range from repeating programmed movements to interpreting observations and planning actions. Industrial definitions emphasize reprogrammable, multipurpose manipulators; broader service-robot classifications also encompass systems with limited autonomy or full teleoperation. (ifr.org)
A milestone in research was Shakey, developed at SRI between 1966 and 1972. This mobile platform integrated sensing, reasoning, route-finding, and manipulation of simple objects. Rather than requiring instructions for every movement, it could plan sequences of actions toward specified goals. Shakey illustrated how perception, automated planning, and physical execution could be combined within one robotic system. (sri.com)
Physical structure and actuation
A robot system typically includes a mechanical structure, actuators, sensors, controllers, and a power supply. Manipulators consist of linked segments connected by rotating or sliding joints. Mobile platforms may use wheels or other locomotion mechanisms. An actuator converts supplied energy into movement; common implementations include electric motors and hydraulic or pneumatic mechanisms. An end effector is the device attached to a manipulator’s working end, such as a gripper, welding tool, or suction device. (osha.gov)
Mechanical design determines which movements are possible and what loads the system can carry. Reach, joint limits, payload, and operating speed constrain task execution. The robot’s useful capabilities also depend on surrounding equipment: fixtures, conveyors, tools, and sensors can be integral parts of an industrial robot installation. Consequently, evaluating an isolated robot arm does not establish the performance or safety of the complete application. (osha.gov)
Motion, planning, and control
Robot kinematics describes motion without considering the forces producing it. Forward kinematics calculates the position and orientation of a tool from joint coordinates; inverse kinematics determines joint configurations that produce a desired tool pose. Multiple solutions, unreachable poses, and singular configurations complicate the inverse problem. Robot dynamics adds masses, inertia, forces, and torques, drawing on classical mechanics to predict motion or calculate required actuator inputs. (hades.mech.northwestern.edu)
Motion planning searches for movements that satisfy task requirements while respecting obstacles and mechanical constraints. A geometric path specifies where a robot moves; a trajectory also specifies how that movement unfolds over time. Planning can operate in configuration space, where each point represents a complete set of robot coordinates. Mathematical optimization provides methods for selecting motions according to objectives such as travel time or effort. (hades.mech.northwestern.edu)
Control connects planned motion to actual behavior. Through feedback, measurements are compared with desired states and actuator commands are adjusted. Position, velocity, and force control serve different purposes. Manipulation additionally requires reasoning about contact: grasping depends on object geometry, contact forces, and friction, rather than simply placing a tool at a target location. (northwestern.edu)
Perception and state estimation
Robotic perception turns sensor measurements into information useful for action. Cameras, range sensors, proximity detectors, and inertial instruments provide different observations of the robot and environment. Computer vision supports interpreting images, while sensor fusion combines observations to estimate quantities that no single sensor measures reliably. Sensor noise and imperfect motion models make state estimation a central problem. (roboticsbook.org)
Localization estimates a robot’s position and orientation relative to a reference frame or map. Simultaneous localization and mapping, commonly abbreviated SLAM, estimates location while constructing a map. Methods based on probability and Bayesian inference represent uncertainty rather than treating every observation as exact. These estimates support navigation and planning, but remain subject to measurement errors and modeling assumptions. (roboticsbook.org)
Learning and applications
Machine learning enables robots to derive models or behaviors from data. Deep learning is used in perception, while reinforcement learning and learning from demonstrations address action selection and manipulation skills. Learned components can complement explicit geometric and physical models rather than replace the entire control system. Learning robotic manipulation remains challenging because actions affect the environment and involve contact, uncertainty, and varied objects. (roboticsbook.org)
Industrial robots perform material handling, assembly, welding, painting, and machine loading. Service robots perform useful tasks outside industrial automation, including consumer and professional applications. Space robotics includes planetary rovers and systems assisting astronauts. Such applications combine remote commands with onboard autonomy, allowing robots to execute selected actions without continuous human intervention. (osha.gov)
Safety and evaluation
Robotic safety concerns the complete system, including tools, workpieces, neighboring machinery, and people. Hazards can arise from unexpected motion, crushing, impact, or stored energy. Programming, maintenance, testing, and adjustment are particularly important because people may enter the robot’s operating space during these activities. Protective measures and risk assessment therefore address operating conditions as well as mechanical design. (osha.gov)
Human–robot interaction examines how people communicate and work with robots. Performance evaluation includes task success, accuracy, repeatability, and behavior during collaboration or changing conditions. Simulation supports controlled experiments, but simulated models require validation against physical systems. NIST robotics research uses testbeds and standardized measurements to investigate safety, dexterous manipulation, collaboration, and the ability to retask robotic systems. (nist.gov)