Level 7 — Navigation (Non-AI)
Position, movement, and decision rules without learning
Navigation defines how a robot moves through space and reaches goals. This level focuses on classical, rule-based navigation methods that rely on geometry, sensors, and deterministic logic — not artificial intelligence.
These methods are fully explainable, reproducible, and suitable for embedded systems with limited resources.
Coordinate systems
Navigation requires a reference frame. All positions and movements must be expressed in a consistent coordinate system.
- global coordinates — fixed reference frame
- local coordinates — robot-centered frame
- transformations between frames
Odometry
Odometry estimates robot position based on wheel rotation or joint motion. It is simple and widely used but accumulates error over time.
- wheel encoders track distance
- integration converts motion into position
- slip causes drift
Dead reckoning
Dead reckoning estimates position using motion and orientation data. It does not require external references.
- combines odometry and IMU data
- errors accumulate with time
- requires periodic correction
Line following
Line following is a fundamental navigation task. It demonstrates sensor-based feedback control in real environments.
- optical sensors detect contrast
- control adjusts steering continuously
- PID control is commonly used
Obstacle avoidance
Obstacle avoidance relies on predefined rules rather than prediction. The robot reacts to sensor thresholds.
- distance-based triggers
- simple steering decisions
- priority-based behavior selection
Navigation limitations
- accumulated position error
- sensor noise and delays
- limited environmental understanding
What you should know after Level 7
- how robots represent position
- why odometry is imperfect
- how rule-based navigation works
- why deterministic navigation is predictable and debuggable
Next: Level 8 — Communication & Debug