Add passive joint moments

[tvdbogert]

@moorepants I have coded this in the branch "add-passive-joint-moments". It seemed quite straightforward. Any feedback is welcome!

The code still works without passive moments, but code generation crashes when I turn them on:

(gait2d-dev) C:\Users\Ton\Documents\GitHub\gait2d>python examples/plot_forward_sim.py
loading modules...
deriving equations of motion...
Traceback (most recent call last):
  File "C:\Users\Ton\anaconda3\envs\gait2d-dev\Lib\site-packages\pydy\codegen\ode_function_generators.py", line 907, in generate_ode_function
    g = generator(*args, **kwargs)
TypeError: 'str' object is not callable

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "C:\Users\Ton\anaconda3\envs\gait2d-dev\Lib\site-packages\pydy\codegen\cython_code.py", line 222, in compile
    output = subprocess.check_output(cmd, stderr=subprocess.STDOUT)
  File "C:\Users\Ton\anaconda3\envs\gait2d-dev\Lib\subprocess.py", line 473, in check_output
    return run(*popenargs, stdout=PIPE, timeout=timeout, check=True,
           ~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
               **kwargs).stdout
               ^^^^^^^^^
  File "C:\Users\Ton\anaconda3\envs\gait2d-dev\Lib\subprocess.py", line 578, in run
    raise CalledProcessError(retcode, process.args,
                             output=stdout, stderr=stderr)
subprocess.CalledProcessError: Command '['C:\\Users\\Ton\\anaconda3\\envs\\gait2d-dev\\python.exe', 'pydy_codegen_0_setup.py', 'build_ext', '--inplace']' returned non-zero exit status 1.

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "C:\Users\Ton\Documents\GitHub\gait2d\examples\plot_forward_sim.py", line 46, in <module>
    rhs = generate_ode_function(
        symbolics.kanes_method.forcing,
    ...<9 lines>...
        specifieds_arg_type='array',
    )
  File "C:\Users\Ton\anaconda3\envs\gait2d-dev\Lib\site-packages\pydy\codegen\ode_function_generators.py", line 917, in generate_ode_function
    return g.generate()
           ~~~~~~~~~~^^
  File "C:\Users\Ton\anaconda3\envs\gait2d-dev\Lib\site-packages\pydy\codegen\ode_function_generators.py", line 624, in generate
    self.generate_min_mass_matrix_function()
    ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^
  File "C:\Users\Ton\anaconda3\envs\gait2d-dev\Lib\site-packages\pydy\codegen\ode_function_generators.py", line 714, in generate_min_mass_matrix_function
    self._set_eval_array(self._cythonize_symbolic_lusolve(outputs,
                         ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~^^^^^^^^^
                                                          self.inputs))
                                                          ^^^^^^^^^^^^
  File "C:\Users\Ton\anaconda3\envs\gait2d-dev\Lib\site-packages\pydy\codegen\ode_function_generators.py", line 666, in _cythonize_symbolic_lusolve
    return g.compile(tmp_dir=self._options['tmp_dir'],
           ~~~~~~~~~^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
                     verbose=self._options['verbose'])
                     ^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^^
  File "C:\Users\Ton\anaconda3\envs\gait2d-dev\Lib\site-packages\pydy\codegen\cython_code.py", line 227, in compile
    raise Exception('Failed to compile and import Cython module.')
Exception: Failed to compile and import Cython module.

[tvdbogert]

No need to reply quickly. I think I will put a breakpoint in ode_function_generators.py and see what I can find out. I might learn something.

The error message shows me the line number and where the file is located in the Conda environment.

[moorepants]

The script "plot_forward_sim.py" in this repo may need an update. Or maybe you've changed something that doesn't maintain prior behavior for that script.

[tvdbogert]

I fixed the error. The symbols qminb, qmaxb, etc. had not been appended to the segment constants. They ended up in the C code which then did not compile. It compiles and runs.

Now, the simulation somehow gets into a numerical problem:

C:\Users\Ton\Documents\GitHub\gait2d\examples\plot_forward_sim.py:93: ODEintWarning: Excess work done on this call (perhaps wrong Dfun type). Run with full_output = 1 to get quantitative information.
trajectories = odeint(rhs, initial_conditions, time_vector, args=args)

The model coordinates all become zero, suddenly.

Strangely, it works fine when I turn off the linear elastic term (segment.py, line 198):

            # hard-coded parameters, these are not critical
            k1 = 1.0     # linear stiffness (Nm/rad)
            k2 = 5000.0  # quadratic stiffness (Nm/rad^2) applied when outside the qmin to qmax range
            b =  1.0     # linear damping (Nms/rad)
            
            # turn the three terms off individually, for testing
            # currently, just k1=0.0 makes it work.
            k1 = 0.0;
            # k2 = 0.0;
            # b = 0.0;

Getting close!

[moorepants]

If you open a pull request from your branch. I can give inline comments (if you are ready for them).

[tvdbogert]

I just created the pull request

[tvdbogert]

I had specified the qmin and qmax parameters in degrees. Now corrected to radians in the latest commit. Simulation still fails.

The passive torques were coded with the assumption that a positive torque pushes the child counterclockwise and the parent clockwise. If I got that wrong, the movement would explode. Tried changing the sign of k1 (with k2 zero) but it still failed. Unless there are different sign conventions for different joint torques (I would not think so), this is not the issue.

[moorepants]

The passive torques were coded with the assumption that a positive torque pushes the child counterclockwise and the parent clockwise.

I need to make a diagram showing all this.

It is same convention as what your original autolev model is.

[moorepants]

Fixed in #32