Abstract

Many smartphones contain a complete set of inertial sensors (i.e. a three axis gyro and accelerometer), so in principle it should be possible to track the 3D position and orientation of the device over short periods of time. Obtaining enough accuracy to to support interesting applications is difficult, but not impossible – this seminar will give a technical overview of a new iOS app which allows users to measure and visualise room layouts using an inertial dead-reckoning approach, and a related biomechanics app designed to assess human gait.

Both apps track the position of the phone in 3D space, with accuracy of a few centimetres. In these apps, the key to getting accurate position tracking is to use application-specific motion constraints. Both apps share an inertial tracking system based around a multiplicative extended Kalman filter, which gives low-level control over the data fusion approach, and is flexible enough that it can be easily adapted to exploit the different constraints that are available in each application. I’ll review the key design choices for this filter, and how it fits with in the overall architecture of each application. Finally, I’ll discuss the potential to extend the system to include data from other sensors, e.g. video feature tracking.