When we first opened VW configurator we were surprised how hard it was to configure the car. Constant configuration conflicts and loops are real annoying things that we saw. Almost similar models are completely separated from each other from the start. We made
What it does
It allows user to get his perfect car from the set of pictures that he picked and use VR for it's configuration.
How we built it
We took a set of different pictures (cars, nature, city etc.), labeled them with feature words. We created website that shows this pictures as the 1st step of the configuration. After user selects images he liked, we suggest him the car that should fit his desires. Also he can configure with VR the interior of the car.
Challenges we ran into
Unstructured data, Models hierarchy is different across CH and UK, missing data, unusual Option choice behaviour.
Accomplishments that we're proud of
Real time colour change, VR interior configurator, smart picture to car mapping.
What we learned
Car configuration is hard and really depends on the backend. Existing backend almost fully defines the process of the car selection
What's next for selectogram
Improve the mapping by using larger datasets. Make the VR interior configurator for more car models.