Researchers have devised a classifier that can tell when a mobile phone user is walking, running or resting. This opens up the opportunities for many new monitoring apps.

The accelerometer inside most mobile phones doesn't have much use outside of controlling games by shaking the phone but now a group of researchers based at a range of universities have come up with something new.

By analysing the output of a three-axis accelerometer the researchers can work out what the user is doing. At the moment they can only distinguish between three states - rest, walk and run - but with a very high accuracy of around 95%. The basic idea is that this could be used as part of a monitoring system suitable for ensuring that users got the exercise they needed or that the elderly were safe and functioning. A fall detection category could easily be added for the monitoring of the infirm. The study used a special accelerometer module complete with Bluetooth connection but the same principles would apply to a mobile phone with accelerometer.

Three classification procedures were used - a neural network, frequency only data and frequency plus other characteristics. The neural network managed a respectable 84% correct classification with the handcrafted  frequency and data characteristics classifier running a close second with 81% correct. The frequency method wasn't tested as extensively,??but multiple data from two subjects was used, and produced very good results of over 90%.

So it looks as if detecting simple activities such as resting, walking and running is fairly easy. Can this be extended to more complex activities? If you think about it from a signal processing point of view it is clear that different repetitive limb movements are likely to set the body into oscillation with a characteristic spectrum. Can you for example, work out from the accelerometer spectrum that a user is typing, drinking, reading a book and so on? It is also reasonable to guess that the accelerometer could be used to pick up traumatic events such as falling, being hit, being involved in a car crash and so on - the next problem is working out how to turn any of these ideas into marketable apps.

It seems there are ever more uses for the mobile phone than were first imagined - does anyone use them to make phone calls any more?

More Information

arXiv:1107.4414v1

arXiv:1107.4417v1

arXiv:1107.4429v1

Developer required for pothole identification

If you would like to be informed about new articles on I Programmer you can either follow us on Twitter or Facebook or you can subscribe to our weekly newsletter.