Model Scaling for OpenSense

[Mohammad S. Shourijeh]

Dear OpenSim and OpenSense developers and users,

I hope this message finds you well. I am reaching out to address a query from a collaborator regarding the necessity of scaling models for use in OpenSense, specifically concerning inverse kinematics and inverse dynamics analyses.

Our current understanding suggests that for inverse kinematics, which involves deriving joint angles, scaling is not required since IMUs provide orientation data directly. This perspective is supported by the documentation available https://opensimconfluence.atlassian.net ... h+IMU+Data, which indicates that model scaling is not necessary for calculating joint angles or conducting other kinematic analyses.

However, there appears to be a contrasting viewpoint presented in a recent video discussion, particularly concerning scaling for CP children, accessible https://youtu.be/ewuvpB2QUf8?feature=shared. It seems there may be some confusion, as the presenters might be referring to the necessity of scaling for analyses that extend beyond kinematic analyses.

Given these discrepancies, it would be greatly beneficial to our community to clarify the following points:

1) Does OpenSense require or perform any scaling for sensor placement or the calculation of joint angles? Additionally, what about analyses that go beyond kinematics, such as inverse dynamics?

2) How does scaling impact the animation of the model using inverse kinematics results?

Your insights and clarifications on these matters would be greatly appreciated to enhance our collective understanding and application of OpenSense.

Thank you.

Kind regards,
Mohammad Shourijeh

[Carmichael Ong]

Thanks for these questions. Some responses provided below:

1) Does OpenSense require or perform any scaling for sensor placement or the calculation of joint angles? Additionally, what about analyses that go beyond kinematics, such as inverse dynamics?

For kinematics, only the orientations of the sensors are used in calculations (see the IMU Inverse Kinematics documentation for more information). This means that model scaling will not affect the results, and while the sensor placement with respect to its location on a body will not affect results, the orientation of the IMU on the body will affect the results for inverse kinematics (see the IMU Placer tool documentation).

For inverse dynamics (and other analyses beyond inverse kinematics), model scaling will affect results. This is because, for instance, it will be important to know the relative locations of the joints to the external forces being applied, as well as the mass and inertial parameters of the bodies. See the Inverse Dynamics documentation page for more info.

2) How does scaling impact the animation of the model using inverse kinematics results?

Scaling will impact the animation since the bodies will not be of the correct lengths. However, the joint angles will still be correct even if the exact location of the bones in space are not correct.