CMC not tracking kinematics

[Mark Gordon]

I am working on a leg press simulation and cannot get the CMC to work properly. When I run the CMC it does not appear to be tracking the kinematics that I put in from the IK calculation. One possible reason is that I have the "Adjust kinematics" box checked, but it will not run if I do not have that checked. If it is not checked I receive the error "Tool execution failed. Check Messages window for details." and the Messages window says

"CMC.computeControls: t = 0.8
WARN- small force range for extobl_r (0.717673 to 0.864977)
WARN- small force range for extobl_l (0.632113 to 0.779296)
SimTK Exception thrown at InteriorPointOptimizer.cpp:249:
Optimizer failed: Ipopt: Unrecoverable exception (status -100)
OPTIMIZATION FAILED...


CMC.computeControls: WARN- The optimizer could not find a solution at time = 0.800000.
If using the fast target, try using the slow target.
Starting at a slightly different initial time may also help."

I already switched from the fast solver to the slow one and that didn't help.

I also received that same type of error when running the gait 2354 example when I took off the external loading if that information is useful to you at all.

[Mark Gordon]

I should have noted that the model does move (ankles start rotating wildly and the legs flop and come in towards the body).

[Mark Gordon]

I finally got the kinematics to track by re-working a lot of things (started again from the example so I'm not sure which thing was wrong before).

However now it is very picky about where it starts and stops. It regularly gives errors about not finding solutions at certain points that are around .1 and .2 seconds apart. I'm working with a 4 second simulation so that obviously makes for a lot of issues.

[Kang Li]

I have also found that the CMC, RRA is very picky with the starting and ending time.
I have also find that natCubicSpline is also very picky on the points you select.

I am wondering if the unstability of the CMC and RRA is caused by the unstability of the spinefitting algorithm?

[Mark Gordon]

I changed the kv and kp values and that got me past a few of the stops, but there are still many which I can get past.

[Mark Gordon]

I meant "cannot get past" in my previous message.

I am still at that point where I cannot get past multiple points, but I can run it for almost 2 seconds of my trial.

However now I am not sure how much of the data to trust. In the tutorial it says not to trust the first 0.03 seconds I believe, but when I changed the kp and kv values it drastically changes my muscle forces for the first second or so of my trial.

[Günter Schneider]

Hi,

I have the same problems with my CMC simulation.

CMC.computeControls: t = 0.12
WARN- small force range for patlig_r (0.026647 to 0.459658)
WARN- small force range for tfl_r (25.0078 to 25.5149)
WARN- small force range for patlig_l (0.0266466 to 0.459651)
WARN- small force range for pect_l (345.991 to 88.1665)
WARN- small force range for tfl_l (25.0716 to 25.5717)
SimTK Exception thrown at InteriorPointOptimizer.cpp:261:
Optimizer failed: Ipopt: Unrecoverable exception (status -100)
OPTIMIZATION FAILED...


CMC.computeControls: WARN- The optimizer could not find a solution at time = 0.120000.
If using the fast target, try using the slow target.
Starting at a slightly different initial time may also help.

I know it´s about one and a half year gone, but did you fix it?

THX a lot,

günter