We build statistical models that predict behaviours based on the output of mobile sensor devices or similar systems. We can work with many types of sensor data including the output from the accelerometer, gyroscope and GPS systems.

What is sensor modelling?

Smart phones, tablets, cars, fridges, wearables… so many everyday devices now contain sensors and store or upload the resulting stream of data. A smartphone alone can contain:

  • a gyroscope
  • an accelerometer
  • a proximity sensor
  • an ambient light sensor
  • a moisture sensor
  • a GPS tracker
  • a compass
  • a fingerprint sensor

These signals can combined to detect specific events or complex movements. For example is a person anxious at a certain location, or are they driving dangerously, or do they follow a specific routine. We build statistical models and machine learning algorithms which will extract these patterns from the combined data. However we will only do this when the use case is ethical and the the device user has agreed to share their data.

Where have we used it?

We have used sensor modelling for clients in the telematics industry and have run feasibility simulations for sensor effectiveness for clients in the biotech industry.