Robotics, autonomy and off-Earth maintenance

Interactive beginner glossary

Use the interactive vocabulary to follow information from sensor to decision to physical action and then into maintenance evidence. Each definition should explain where uncertainty or failure can enter that chain.

• autonomy — Autonomy is a robot's ability to perceive, decide and act locally without continuous human commands, while remaining within validated rules, authority and safety constraints.
• perception — Perception converts raw sensor data into useful estimates of the environment or robot state, such as terrain, obstacles, pose, hazards or equipment condition.
• sensor — A sensor converts a physical quantity into data. Its output always has limits such as range, resolution, noise, bias, latency and environmental sensitivity.
• camera — A camera records light intensity or color across an image. It provides rich terrain information but depends on illumination, visibility, calibration and interpretation.
• lidar — Lidar estimates distance by emitting laser light and measuring its return. It can build precise geometric maps but performance can degrade with dust, reflectivity and range.
• radar — Radar uses radio waves to detect distance, motion or structure. It can operate in darkness and some obscurants but provides different resolution and interpretation from optical sensors.
• wheel odometry — Wheel odometry estimates rover motion from wheel rotation. It is simple and high-rate but accumulates error when wheels slip, sink or deform the terrain.
• inertial sensor — An inertial sensor measures angular motion or specific force with gyroscopes and accelerometers. Its estimates are available continuously but drift when small errors are integrated over time.
• sensor fusion — Sensor fusion combines complementary measurements so weaknesses in one source can be constrained by others while uncertainty and disagreement remain explicit.
• confidence — Confidence is a quantified or classified indication of how trustworthy a perception or estimate is under current conditions. Low confidence should change robot behavior rather than be ignored.
• path planning — Path planning selects a feasible route from the current state to a goal while accounting for obstacles, terrain cost, vehicle capability, energy and safety constraints.
• obstacle avoidance — Obstacle avoidance detects and steers around hazards during motion, usually on a shorter time scale than global path planning.
• safe state — A safe state is a predefined configuration that minimizes immediate hazard, stops risky motion, preserves critical energy and keeps enough capability for diagnosis or recovery.
• actuator — An actuator converts a command into physical action, such as wheel torque, joint motion, valve movement or tool force.
• motor — A motor converts electrical energy into mechanical rotation or force. Its usable output is limited by torque, speed, current, temperature and gearing.
• torque — Torque is a turning effect of force about an axis, measured in newton-metres. Robotic joints and wheels must provide enough torque without exceeding structural or thermal limits.
• saturation — Saturation occurs when an actuator reaches a physical or commanded limit and cannot produce the additional force, torque, speed or rate requested by the controller.
• robot arm — A robot arm is a chain of powered joints and links used to position an end effector for manipulation, inspection, repair or sample handling.
• condition-based maintenance — Condition-based maintenance triggers inspection or intervention from measured equipment health indicators rather than relying only on fixed calendar intervals.
• diagnostics — Diagnostics is the process of detecting, isolating and identifying faults from symptoms, sensor data, tests and system models.
• vibration — Vibration is repeated mechanical motion around an equilibrium. Changes in amplitude or frequency content can reveal imbalance, looseness, bearing wear or structural problems.
• temperature trend — A temperature trend is the change of temperature over time rather than a single reading. A rising trend can reveal growing friction, cooling loss or electrical overload before a limit is crossed.
• predictive maintenance — Predictive maintenance uses condition trends and models to estimate future degradation or remaining useful life so maintenance can be scheduled before failure.
• spare — A spare is a replacement component or consumable reserved to restore function after wear, damage or failure. Its usefulness depends on compatibility, storage life and replacement capability.
• modularity — Modularity divides a system into replaceable units with controlled interfaces so repair, upgrades and reconfiguration can be performed without redesigning the whole robot.
• fleet — A fleet is a group of robots managed as a coordinated operational resource. Fleet performance depends on task allocation, availability, charging, maintenance and shared infrastructure.
• throughput — Robotic throughput is the useful work completed per unit time, such as tonnes moved, inspections completed or parts delivered, after delays and downtime are included.
• queueing — Queueing occurs when tasks or robots wait for a limited resource such as a charger, airlock, workstation or communications link. Waiting time can dominate fleet productivity.
• charger — A charger transfers electrical energy into a robot's battery within voltage, current, temperature and state-of-charge limits. Charger availability can become a fleet bottleneck.
• common-cause failure — A common-cause failure defeats multiple supposedly independent units through one shared cause, such as the same software defect, dust environment, power bus or maintenance error.