RRA failure

[Anne Schmitz]

Hello,

I am having a hard time getting RRA to to complete for a 3.3 m/s running simulation. I am using the gait2392 model with no arms. I used the Visual3D-->OpenSim pipeline to process the c3d data to mot data (inverse kinematics) and get manual scaling factors to apply in the scale analysis. When I run RRA, I get the error message:

SimTK Exception thrown at InteriorPointOptimizer.cpp:261:
Optimizer failed: Ipopt: Maximum iterations exceeded (status -1)
OPTIMIZATION FAILED...

CMC.computeControls: WARN- The optimizer could not find a solution at time = 0.435000.
If using the fast target, try using the slow target.
Starting at a slightly different initial time may also help.

I tried increasing the maximum excitation and optimal force of the residuals but this did not help, it fails at the same time. I also tried changing the starting time, no help. I have also made sure to alter the actuator file to apply the force at the scaled pelvis origin. Any other ideas? Your time is very much appreciated.

Thanks,

Anne Schmitz

[Ryan Bakker]

Anne,

Have you tried running inverse dynamics? Do the values look reasonable for the motion you are studying? Inverse dynamics solutions can be a simple way to check if you GRF is being applied correctly without the confusion of RRA. If you have run inverse dynamics in V3D, do the results match those from opensim?

Edit: Also is RRA failing in the first frame, or partway through you simulation?

-Ryan

[Anne Schmitz]

Hi Ryan,

Thanks for the suggestions. I re-looked at my GRFs and the shape of the waveform didn't look right. Good lesson learned: If Visual3D has parts where inverse kinematics are NaN, it will linearly interpolate the GRF even though the force plate data isn't NaN. I ended up with a 'large' portion of time were the GRF was linear at push-off.

Thanks,

Anne