Skip to content

Tracking and API review

Knowledge check · self-test

Tracking and API review

Revisit the chapter’s numbers with the separate calculation audit and full working.

Back to the chapter →

Revisit the chapter’s numbers with the separate calculation audit and full working.

An outlier is a single reading far from the rest. Persistent bias shifts readings in the same direction, so averaging will not remove it. A real change in motion, such as boarding a train, is neither kind of error: the filter must adapt to it. Use the examples below to test these distinctions.

- **Predict:** Three location fixes read 10 m, 30 m, and 11 m from the platform sign. Will a median position follow the 30 m jump? **Check:** No. The median is 11 m, so one large mobile outlier is held back.
- **Predict:** Every Wi-Fi position is 8 m north of the true platform. Will a longer moving average remove that error? **Check:** No. Position filtering can reduce scatter, but a fixed location bias stays.
- **Predict:** The phone boards a train and speed rises fast. Should a walking-only filter keep the old station point? **Check:** No. The mobile motion model must allow a real change of travel mode.

Back to Mobile Sensor APIs: Positioning and Filtering: Make the API Decision Reviewable