You drive into a tunnel. Above the car is rock and concrete. The receiver's view of the sky has vanished, yet the navigation arrow sometimes glides forward as though nothing happened. Has your phone developed X-ray vision? No. It is estimating movement using clues left behind.
1. Position and movement are different measurements
A satellite fix answers approximately where the receiver is at a point in time. An accelerometer and gyroscope reveal information about changes in motion and orientation. Navigation software can combine those readings with the last known position, a map and reasonable assumptions about the route. This is a form of sensor fusion.
The approach is useful precisely because satellite positioning is intermittent. If a phone knew where a car entered the tunnel and can estimate how its motion continued, it may be able to present a plausible path for a short distance. Some vehicle systems use additional information from the car, but what is available varies between devices and integrations.
Did you know? A smoothly moving navigation arrow in a tunnel is often a prediction. It does not prove that GPS signals are penetrating the roof.
2. Why the estimate drifts
Inertial sensing has a weakness: small measurement errors accumulate. If the system begins with a wrong speed or heading, its estimated path can increasingly diverge from reality. A map can constrain the possibilities, but tunnels can branch, and map data can be imperfect. Long stretches without a reliable reference become more difficult.
When satellite reception returns, the software can reconcile its prediction with new measurements. The resulting jump or correction may look like an error in the map, when in fact it is the visible sign of a hidden estimation process catching up.
3. A lesson about every location display
Navigation interfaces prefer smoothness because it is easier for humans to follow. Smoothness is not the same as certainty. A phone can show a confident-looking arrow while its underlying system is managing growing uncertainty.
For the related challenge of concrete walls, see why GPS struggles indoors. For reflections in busy streets, read the urban-canyon story.
Keep exploring the science of location.
Real-world tracking is rarely one signal or one sensor. Follow the evidence, understand the limitations, and use technology responsibly.
Browse all GPSNear stories →Sources and reporting notes
Technical explanations are editorial synthesis. Simulated images do not represent live scanning or a guaranteed detector capability.
Learn how we prepare and review GPSNear explainers: our editorial policy.