Even if the exhibit went fine, there is always something to learn, and that can be applied in a future.
Invite people to go in
First thing learned, especially in a collective and when the location is unfortunate (like the one we had, which was downstairs, next to toilets and on the way between two wings of the building), is that visitors have to be attracted inside: we noticed that when someone was inside the room and the sound was playing, people were more likely to go downstairs and wait for their turn. But when it was silent, the small sign with the arrow was not enough.
We could have, for example, played a sound every 10 seconds when nobody was inside, or detect when someone was approaching the beginning of the stairs and play a sound to attract him.
Make technology more transparent and people proof
Another thing we learned is that all the equipment must be totally transparent and setup in a way that people cannot, even by chance, mess with it. It happened that a few persons touched the sonar sonar sensors, and pushed them inside inside their box, thus making the distance detection going crazy. Also the speakers, and the computer was a bit too much in sight and they could have been easy to touch.
Next time we have make the sensors (and cables) blend more with the walls, and same for the computer and the board, we probably need to build a small cabinet for them. Or even better, try to get rid of the laptop, and build everything to run on the Arduino or on the new super-mini full fledged boards like Raspberry Pi.
Build the installation around the room
We had the confirmation of the room just a few weeks before opening of the exhibit, so we had to adapt our installation to a room that was different from the one we built our installation for: we thought of a rectangular 5m x 2m room, while the one we got was 2 rooms connected by an opening, so with a wall in the middle, and was also very strangely shaped, with lots of angles and niches, and very big door opening toward the inside. All of that was fooling the sensors. So instead of detecting the people position in the room we had to fall back to just tracking the distance from the side walls, and detecting when they entered in the second room. Probably if we had know of its strange shape upfront we could have either developed the sensing differently or tried to work around its shape.
So for next time, we have to be sure of the room in which we’ll do the installation, and build the installation around it, not the other way around.
As the number #00 implies, this was our first installation of that kind, meant both to see how people reacted to this experience, and also to test the technologies and tools involved. So for sure next time it would be a more advanced and more polished installation