simtk.org

Hello,

I am looking into using tri-axial accelerometer data as input to the motion of a segment in my model. I expect there to be some higher frequency vibration that I would like to capture to some degree.

I am hesitant to integrate, since I am not sure how well the vibration will be reflected in the velocity and position after integration. The other idea I was considering was to break the signal down into amplitude and frequency domain and then reconstruct it to build the position of the segment.

I was wondering if anyone had encountered a similar scenario, or has used any tools that may be helpful.

Any advice would be terrific.

Thanks

Alex

Hi Alex-

Its hard to tell what you would be getting from the high frequency vibration? Wouldn't that be dominated by noise from the localized skin motion artifact?

I think the common method is to integrate but you will get drift without using additional methods.

Not sure that helped
-james

Hi Alex,

As you mentioned double integration of the acceleration has an second order error with time due to noisy measurements. So you can't extract the exact position. Additionally, you need the initial conditions. Maybe you can try to use orientation measurements to constrain your motion.

BR,
Jim