Ellipsoid Wrapping Issue

[Connor Stapp]

I've been messing around with ellipsoid wrapping algorithms in order to understand how they work. I've been specifically working with the Tug of War model that's already included with OpenSim and been trying to add my own ellipsoid wrapping algorithms. In this simulation, I edited the model so that the block would fall when simulated. Before simulating the model, I placed an object beneath one of the muscles to see how the muscle would wrap around the object as the block fell. What I noticed is that the muscle begins to wrap around the muscle, as seen in the first picture, and as the simulation progresses, the muscle "jumps" to beneath the object as seen in the second picture. Is there any particular reason for why the muscle would "jump" to beneath the object?

[Thomas Uchida]

Is there any particular reason for why the muscle would "jump" to beneath the object?

Yes, I think the reason is that the algorithm wasn't designed to avoid discontinuities over time. I think discontinuities of this sort can usually be resolved by specifying the appropriate quadrant over which the muscle should wrap (which, it seems, might also require rotating the obstacle so that the muscle wraps over the same quadrant for the entire simulation). Several models use wrapping surfaces (e.g., arm26); it might be helpful to see how other model developers have used them.