I noticed something happening in the Hunt Crossley Model that I can't understand completely. I have an adapted gait9musc18 model with 2 spheres in both feet and a half space in the ground as contact geometries. I am running a forward simulation with this model giving it some (random) activations for the muscles. At some point the integration starts going very slow/virtually stops. I tried to investigate the cause of this and I think it originates from a huge contact force - slowing down integration. When I look at my contactforces I noticed that they sometimes fall to zero, while the contactbodies are still in contact (visual control). I can't understand why this happens. Because there are no forces, the contact bodies sink into eachother. At a certain point a very large peak in the contact force shows, probably hampering the forward integration.
In the attachment I added my model, and a 2 files containing the states and force during the simulation.
A way to replicate the issue is to simulate a drop of the model in the GUI - If you let it "fall" from the maximum height (setting pelvis_y maximal) then the simulation gets stuck at some point.
Can someone explain the cause of this to me? I think I can avoid the issue by increasing the stiffness, but still I would like to understand the cause
.Regards
Tom