Section 5.6 Glossary

Localization

🔗

Answering the question "Where am I?"—estimating a robot’s own position and orientation within its environment.

🔗

Mapping

🔗

Answering the question "What does the world look like around me?"—building a representation of the environment from sensor data.

🔗

Odometry (Dead Reckoning)

🔗

Position estimation computed by integrating wheel or motion sensor data over time, without reference to external landmarks.

🔗

Drift

🔗

The gradual accumulation of position error over time from small, uncorrected measurement errors.

🔗

Wheel Encoder

🔗

A sensor that measures wheel rotation, used to estimate distance traveled.

🔗

Wheel Slip

🔗

Loss of traction between a wheel and the ground that causes encoder readings to overstate actual travel distance.

🔗

Landmark

🔗

A recognizable environmental feature used to correct a robot’s estimated position, either artificial (e.g., an AprilTag or QR code placed in the environment) or natural (e.g., a wall corner or pillar already present).

🔗

Feature-Based Mapping

🔗

A map representation that stores the environment as a list of discrete landmark coordinates, rather than a fixed grid of cells as in an occupancy grid map.

🔗

Simultaneous Localization and Mapping (SLAM)

🔗

The "chicken-and-egg" problem of building a map of an unknown environment while simultaneously tracking a robot’s position within it, since each depends on the other.

🔗

Uncertainty Cloud

🔗

A probability distribution over possible values for a robot’s position or a landmark’s location, rather than a single fixed estimate; it grows during motion and shrinks after a sensor observation.

🔗

Probabilistic Co-estimation

🔗

The SLAM strategy of repeating a prediction step (odometry grows uncertainty), an observation step (sensors scan landmarks), and a correction step (both pose and landmark estimates are updated together) to solve localization and mapping at the same time.

🔗

Sensor Fusion

🔗

Combining data from multiple sensors to produce a position or map estimate more accurate than any single sensor alone.

🔗