As part of a research project, I have been tasked with determining gait speed (walking speed) for a subject using a handheld iPhone. The main constraint is that walking distances are very short (10-30 meters), and accuracy is critical (data is used for research.) Using a pedometer and measuring step distance is not an option (too time consuming.) The only plausible way I foresee this could be work is using accelerometer data. However, I'm not sure whether this will succeed. I was wondering if anyone has experience with these sensors and could answer the following two questions:

  1. Will an iPhone accelerometer plausibly be able to accurately measure walking distance for such short distances? Are there any figures on accelerometer accuracy for this kind of sensor? (I couldn't find any.)
  2. Are there any established protocols for converting accelerometer data into distance measurements (calibration, noise correction, sampling rates, etc.) that have been shown to have good accuracy for such short distances? I have been able to find this, http://www.freescale.com/files/sensors/doc/app_note/AN3397.pdf but I'm not sure if this will apply well to the iPhone sensors.
  • $\begingroup$ If you're going to use an iPhone, try using it for video recording of the walk at 30 or 60 FPS, and put markers on the ground along the walking path, either before or after. $\endgroup$ – hotpaw2 Mar 21 '15 at 3:34

I don't have any idea about Accelerometer in iPhone, but converting Accelerometer data into distance is fairly difficult because the offset in double integration causes drift as already told. Check this interesting presentation sensor fusion for further details.
As far as I know this can be avoided by using Digital Accelerometers( with high sampling rates), double integration has to be performed using electronic integrators with appropriate filters at every stage to remove the offset as well as the low frequency noise.
Also in that presentation he mentions the possibility of using more sophisticated algorithms like use of Kalman Filter for object tracking or Pedometer Algorithm for pedometer design.

Hope this gives little more insight into your problem:)

PS: This does not accurately address your question still I thought I will share my knowledge, I want to comment but I don't have enough reputation !


Using an accelerometer means you are doing integration and you will need as a tarter an accurate starting point (are you always starting from 0?) and than velocity. So I really think that you ask if you could use the distance measurement capability of iPhone.

One thing you could easily do, is testing the "system" - get several known points with the distance and use them for testing/calibration. Than you could repeat it.

If you control the distances, you may be able to have a good measure by combining camera (calibrated) video analysis.

Could you use the Kinect?

  • $\begingroup$ If you also stop at 0 then it might also ne possible to remove some of the drift of the integration. $\endgroup$ – fibonatic Jun 20 '15 at 8:38

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.