About the project
A trainer for the Swiss practical driving exam, built on the actual ground the examiner drives you over.
Why it exists
Every digital product on the Swiss market prepares for the theory exam. Thirteen of them, all licensed, all good, all stopping at the written test. For the practical exam there is one Windows simulator and a driving instructor at CHF 90–105 an hour.
Meanwhile in canton Vaud only about half of candidates pass at the first attempt, and anyone exchanging a foreign licence gets a single attempt at the control drive — fail it and the whole Swiss licence starts again, at CHF 2 500–3 500.
What we actually do
We survey the examination area object by object, then let you drive it. Around the Cossonay examination centre that means 170 mapped objects on the routes — roundabouts, give-ways, pedestrian crossings, bus stops, locality boundaries — each in its real position, each with the article of federal law that governs it.
Four plausible routes are computed on the real road network with the real speed limits, coming out between 42 and 48 minutes, with the same mix of town, country road, motorway and parking the regulation requires. The expert's commands stay in French, because the expert will not translate.
Where the data comes from
| Source | What it gives | Licence |
|---|---|---|
| OpenStreetMap | road network, speed limits, the 355 surveyed objects | ODbL |
| swisstopo | the national map and aerial photography | free geodata, ©swisstopo |
| KartaView | photographs taken from the road | CC BY-SA 4.0 |
| Fedlex | OCR 741.11, LCR 741.01, OSR 741.21 | official texts, unprotected |
Photographs are shown unmodified and credited to the person who took them. Google Street View is not used: its terms forbid extraction and derivative works, and passing an image through a generative model does not change that.
How a rule gets in
No driving rule enters this product until the article has been read in the federal text. That is not a slogan — the build refuses to compile a scenario that cites a law reference which has not been entered by hand after reading it.
The discipline has already paid for itself. Three citations in our own source material turned out to be wrong: lateral distance to a cyclist was attributed to an article about motorcycles in general, and the amber-light rule was attributed to the wrong ordinance and the wrong paragraph.