Better error and exception messages. Formal mechanism for to make sure all enabled degrees of freedom are actuated, have no tasks, etc.; and to generate template actuator and task sets.

[chrisdembia]

initial title: CMC, etc. should give warning or meaningful error when there are active CMC_Tasks for locked coordinates.

Right now, users get a DLASD. It'd be nice in general to have a formal error checking mechanism that makes sure all enabled degrees of freedom are actuated.

[jimmyDunne]

Is there a way of extending this issue to include the removal of cmc/rra actuator and task files?

I feel that there should be an automated way to either generate these files, or internally create an object if no file is provided.

Your problem would be part of this solution

[chrisdembia]

Oh yea it'd be great if we automatically generate template actuator and task files.

[chrisdembia]

James has a good idea to have the taskset automatically generate a taskset for a model.

[chrisdembia]

Static optimization gives fairly good information. It tells you the joints for which there is a constraint violation, and which actuators are maxing out.

[chrisdembia]

In our OpenSim meeting yesterday, it sounds like this issue is really important for our opensim 4.0 release. I think we should prioritize it and really approach it systematically. (1) Catalog all the common exceptions people hit, (2) address each "failure mode".

@jimmyDunne

I feel that there should be an automated way to either generate these files, or internally create an object if no file is provided.

Right now, RRA and CMC produce template setup files for you; but should also be able to generate template tasks files and force set files. To get the template setup file, you can right now do cmc -PS (i think). Imagine being able to do cmc -template-tasks my_model.osim.

Also, as for a mechanism to DETECT unactuated degrees of freedom: I'm thinking of a method in Model like Model::checkForUnactuatedDegreesOfFreedom. It takes a state, and sets the control signal for all actuators to be its min and max value, and records which coordinates are affected by the actuator. Then, after looping through all actuators, we are left with which coordinates are not affected by the actuators in the current state.

I don't think we could use this diagnostic at the beginning of CMC; we'd have to wait until an exception is thrown, then use the diagnostic to give a more informative exception message. That's because the underactuation can be state dependent; so a preliminary check might show underactuation, but in practice, there might not be any underactuation.

[chrisdembia]

Here's an exception that I just got:

Exception:
  Object: ERR- Empty filename encountered.
  file= /home/fitze/repos/opensim/Branches/OpenSim32/OpenSim/Common/Object.cpp
  line= 109

Ideally, the exception could also report the name of the property or field whose value was empty.

We should also make a distinction between user-facing errors and developer-facing errors, like Simbody does with its exceptions. In this case, the file listed will confuse users who may think that Object.cpp is the file that the code is looking to open.

[chrisdembia]

Here's a log from the forward tool when it fails to simulate a model that has coordinates that are prescribed AND locked.

Constructing tool from setup file BC001_NW_avg_Setup_ForwardDynamics_Arnold.xml.

AbstractTool BC001_NW_Forward loading model 'BC001_NW_Arnold.osim'
Updating Model file from 20200 to latest format...
Loaded model GenericModel from file BC001_NW_Arnold.osim
Adding force object set from AFOPS_Actuators.xml
Storage: file=BC001_NW_controls.sto (nr=300 nc=5)
-----------------------------------------------------------------------
Running tool BC001_NW_Forward.
Old version storage/motion file encountered
Storage: file=BC001_NW_avg_grf.mot (nr=300 nc=22)
.. assuming rotations in Degrees.
ExternalForce::right
Data source being set to BC001_NW_avg_grf.mot
ExternalForce::AFO_foot
Data source being set to BC001_NW_avg_grf.mot
ExternalForce::AFO_shank
Data source being set to BC001_NW_avg_grf.mot
Storage: file=BC001_NW_avg_ik_filt.mot (nr=300 nc=14)
Storage: file=BC001_NW_controls.sto (nr=300 nc=5)
Model unable to assemble: AssemblySolver::assemble() Failed: SimTK Exception thrown at Assembler.cpp:843:
  Method Assembler::assemble() failed because:
Optimizer failed with message: SimTK Exception thrown at InteriorPointOptimizer.cpp:261:
  Optimizer failed: Ipopt: Not enough degrees of freedom (status -10)
Assembly error tolerance achieved: 1.01435785923922 required: 1e-09.
Model relaxing constraints and trying again.
Model::formCompleteStorages(): WARNING- Did not find column knee_angle_r_beta in storage object.
Coordinate.setValue: WARN- coordinate subtalar_angle_r is locked. Unable to change its value.
AssemblySolver::track() attempt Failed: SimTK Exception thrown at Assembler.cpp:897:
  Method Assembler::track() failed because:
Optimizer failed with message: SimTK Exception thrown at LBFGSBOptimizer.cpp:119:
  Optimizer failed: ABNORMAL_TERMINATION_IN_LNSRCH
Assembly error tolerance achieved: 0 required: 1e-09.
ExternalForce::right_transformedP
Data source being set to BC001_NW_avg_grf.mot_transformedP
ExternalLoads: WARNING ExternalForce 'AFO_foot' does not specify a point of application.
ExternalLoads: WARNING ExternalForce 'AFO_shank' does not specify a point of application.
Storage: file=BC001_NW_controls.sto (nr=300 nc=5)
Model unable to assemble: AssemblySolver::assemble() Failed: SimTK Exception thrown at Assembler.cpp:843:
  Method Assembler::assemble() failed because:
Optimizer failed with message: SimTK Exception thrown at InteriorPointOptimizer.cpp:261:
  Optimizer failed: Ipopt: Not enough degrees of freedom (status -10)
Assembly error tolerance achieved: 1.01435785923922 required: 1e-09.
Model relaxing constraints and trying again.
MODEL: GenericModel
...
Integrating from 0.0055 to 3.3816
ForwardTool::run() caught exception 
SimTK Exception thrown at Integrator.cpp:431:
  Integrator initialization failed apparently because:
SimTK Exception thrown at SimbodyMatterSubsystemRep.cpp:4075:
  Error detected by Simbody method SimbodyMatterSubsystem::projectQ(): Failed to achieve required accuracy 1e-06. Norm on entry  was 0.059235; norm on exit 0.0569599. You might need a better starting configuration, or if there are prescribed or  locked q's you might have to free some of them.
  (Required condition 'perrNormAchieved <= consAccuracy' was not met.)

Printing results of investigation BC001_NW_Forward to ResultsForward/NW_Arnold.
Forward simulation time = 320.971ms

I don't think there should have been a soft warning when assembly failed. The application should have terminated with a message saying "Could not assemble model because coordinate 'name' is both locked and prescribed. Edit this Coordinate by editing the 'locked' property to false or set the 'prescribed' property to false."

[chrisdembia]

A frequent error is ArrayPtrs.get() when a string field in an xml file was left empty. The ArrayPtrs.get() error is not very informative.

[chrisdembia]

Here's another exception: https://github.com/opensim-org/opensim32/issues/16

[chrisdembia]

#449