We are looking for a game-changing and impactful AI implementation that has the potential to change the way we are able to locate objects utilizing radio propagation and triangle measurements.
The solutions should be able to learn and improve the current capabilities of location accuracy, by creating a learning model that utilizes raw data from current best in class implementations and adds AI enabled improvements via a developed learning machine that utilized the raw data from current sensors as well as provided high accuracy data provided manually.
The basic borders for the AI innovation are:
- Combine available low accuracy (>5m accuracy) location data collected by Bluetooth enabled sensors and beacons with manually measured exact location data (accuracy <10 cm)
- Create an AI powered solution that utilizes the low accuracy on-line data with the high accuracy data in order to create a learning machine that improves the accuracy of the sensor network to a 1m level
- Solutions should be radio agnostic, e.g. suitable to be used for other radio based location solutions (like WiFi) by just providing new raw data as well as accurate data to repeat the learning process
All propositions can be stored in opensource libraries, following the BSD licensing guides.