Reduced IMU count

[Lena Uhlenberg]

Hello,

I have a working OpenSense model with 7 IMUs for the lower limbs (pelvis, left/right femur, left/right tibia, left/right foot). Now, I wanted to reduce the number of IMUs by omitting/deleting for example the femurIMU and see the simulation results. This way the whole femur does not move anymore (stays in calibration pose), which makes sense as it does not get any orientation estimation. However, I was thinking that in terms of a kinematic chain model, OpenSim/OpenSense should still be able to simulate the movement of the femur and calculate meaningful angles. Does OpenSim/OpenSense provide any tools for doing so?

Thanks and best regards,
Lena

[John Davis]

I don't think this is possible with OpenSense, but it might be doable using OpenSim Moco: you could set up a simulation that tracks the sparse set of accelerations and angular velocities from the IMU data you do have, plus minimizes some other effort or metabolic cost (maybe sum of squared muscle activation). I remember seeing a few slides about this idea this summer at NACOB...it might've been Reed Gurchiek's talk, but I can't remember for sure. I think Ton van den Bogert's group has published some work on this idea too. Might be worth asking about on the Moco forum.

[Lena Uhlenberg]

Thank you John, I will have a look into MOCO and also the provided references.