Inspiration
What if you could walk into a building and instantly know where to go? What if your doctors could get information on their phones as they walk in to meet you, without any extra hardware installation?
What it does
It uses WiFi to map buildings into unique spaces, and then triangulates the user and access points for each space. From there, it builds up maps of the rooms by fitting linear models to interference in WiFi signals. Topological mapping, Simultaneous Localization and Mapping, and Structure Mapping.
How I built it
A Raspberry Pi Zero with dual WiFi antennae measures the signal strength of local access points in real time. The sets of access points observed maps to known spaces in a topological map of the building. From there, the user is mapped within each spaces, and nearby spaces are networked together. The final step, once the user's location is known, is to look for anomalies in the WiFi strength and build up models of the wall structure.
Challenges I ran into
Continual WiFi interference, and building a model that related signal strength to distance. To solve the WiFi-performance model problem, I used 4 known wireless access point locations in the basement to measure the signal drop off over distance. I then used a least squares fit of a logarithmic model.
Accomplishments that I'm proud of
Implementing non-metric mapping, simultaneous localization and mapping, and WiFi-based metric mapping of a building
What's next for WiMapFi
3D and more sophisticated models for mapping walls
Log in or sign up for Devpost to join the conversation.