Step Counter: MetaMotion R vs MetaWear RG

Both MetaMotion R and MetaWear RG have the same 3-axis accel and gyro. However, the step counter gives very different results. 

Trying to understand why. (R is accurate and RG is off ) 

Also, is there more documentation around step counter algorithm ? 

Thanks, 
Tarun

Comments

  • edited June 2017
    Please quantify what you mean by "very different results".  How are you placing the sensors and configuring them?

    The step counter comes as part of the BMI160 sensor; you can read more about it in the datasheet.
  • Eric, 

    I hold the sensor with y axis of accelerometer pointing upwards and move the sensor up and down. 

    R gives consistent (and accurate) results.  I have repeated the experiment several times with the same result on R. 
    However, RG updates the step counter by 2 instead of 1 and does so somewhat inconsistently. Will try to record a video and send it. 

    Also, thanks for the pdf link. 
  • What code are you using to configure the devices?
  • Configure the device? 
    Same code base (iOS app code on your Github) for counting steps. Doing that along with capturing the accel data. Firmware up to date for both. 
  • Have you tried placing the boards in your pocket (or wrist) and walked around instead of doing the motions you previously mentioned?
  • I can try that. Although do you see any reason for the difference in the test results - for the simple test i did ?  

    Also, could you pls try it at your end too - maybe I have a faulty device ? 

     


  • It's possible you aren't exactly replicating the same motion for both boards.  

    Also, streaming accelerometer while running the step detector is not recommended.  Disable the stream and see if that resolves the issue on the RG.  Lastly, try using the stepCounter property instead of the stepEvent property.
  • Ok thanks, will try the suggestions. 

    Also - I have repeated the experiment several times with both R and RG - with the same results. 
This discussion has been closed.