Indoor/Outdoor DetectorAndroid | Machine Learning | Aware Framework
It is a plugin for the Aware Framework that use the smartphones sensors to detect if the user is indoors or outdoors.
This application was developed for the course Mobile and Social Computing of the Ubiquitous Computing Master's program of the University of Oulu. |
|
Sensors:
Implementation:
Good results (81% accuracy) were achieved although no extensive testing was done.
- Barometer: Buildings have controlled pressure
- GPS: Number of satellites visible and speed
- Magnetometer: Magnetic intensity readings change due to electronics
- Light sensor: Artificial light vs natural light
- Proximity sensor: Tweak light sensor
- Network: Number of cell towers, signal strength, and internet connection type
- WiFi: Number of access points
- Clock: Moment of the day
Implementation:
- Random Forest algorithm using WEKA: Trained with 100 data points
- Weather information
- Part of the day information (Night, Twilight or Day)
- Limit the use of the sensors for less battery consumption
Good results (81% accuracy) were achieved although no extensive testing was done.