Research project / 2026
An indoor-navigation prototype for Berlin Hauptbahnhof: estimate where a person is, understand which level they are on, and give them a route they can follow in the real station.
The problem
Five levels. Long transfers. No reliable way to tell which level you are on.
Berlin Hauptbahnhof is a complicated system of platforms, stairs, elevators, corridors, and interchanges. Moving from one train to another can take a long time because the route crosses several levels. Existing maps can show the geometry, but it is difficult to orient yourself inside the building—and especially difficult to know your vertical position.
The station’s open level data contains walls, platforms, stairs, elevators, and other objects. I convert those objects into a layered 3D representation. Semi-transparent geometry keeps the building understandable without blocking the route, so a person can see how the path moves through the station.
GPS is unreliable inside the station and is particularly poor at determining which floor a person is on. I measured Wi‑Fi signal strength at known locations across five levels. Each location has a characteristic pattern of signals from nearby access points—a fingerprint.
When the phone scans the surrounding networks, the system compares the observed signal strengths with the stored dataset. It selects the nearest matching measurement points and returns the most probable location and level. Repeated scans make the estimate more stable than a single observation.
The fingerprinting service estimates the user’s position and level. The rendering service places that position on the 3D station model. The routing service draws a path between platforms. As the user walks, the prediction changes and the marker moves through the rendered station with them.
Compare live Wi‑Fi observations with known measurement points.
Show the station, route, level, and current position in 3D.
Build a walkable line between platforms and vertical connections.
The prototype is tested by walking through the station with the phone. The first-person view is compared with the physical space: stairs, signs, platforms, and the direction of travel. The goal is not only to predict a coordinate, but to make the next action understandable.
Research question
This project explores whether Wi‑Fi can provide the missing vertical awareness inside a complex station, while a 3D representation turns an abstract route into something a traveller can match to the place around them.