Description of the frequency spectrum plots.

[moorepants]

@spinningplates Dan Simon reviewed the preprint and commented on this sentence:

Additionally, the amplitude of output signal in the 0.8~\si{\meter\per\second} walking speed begins to attenuate around 2~\si{\hertz}, which is a noticeably lower frequency than the other walking speeds.

He said: "I don't see this."

After you create the new frequency spectrum in issue #77, there needs to be a better description of what is actually seen.

[skhnat]

This has been addressed in pull request #98. I just removed the line since it made no sense with the new results.

[tvdbogert]

Sorry to reopen this, but in the previous paragraph, it says "The system introduces a delay and seems to act as a low pass filter". It definitely looks like that in the time domain, but not in the frequency analysis. Why don't we see more high frequencies in the commanded signal than in the measured signal? The spectra are almost identical. You could quickly replace "plot" by "semilogy" and if that shows the low pass filtering, I would be confident that it's OK. Maybe that high frequency stuff is large in the time domain, but spread out over a wide range of frequencies and too small to show up at all in the spectrum on the linear scale.

[moorepants]

That is a discrepancy. We double checked both doing a normal FFT on both signals and the spectrogram method and got the same results (identical spectra). But is odd. Worth checking.

[tvdbogert]

I wlll investigate to convince myself and then close the issue with a brief report of my findings. I trust it's correct. I just need to understand why the spectra look so similar.

@spinningplates I already have one of the output files. Can you also send me the input file? I don't have Simulink so I can't generate it.

[skhnat]

@tvdbogert I sent it to your email. Columns aren't labeled. They are Time, 0.8 m/s, 1.2 m/s, 1.6 m/s, and two running speeds that we didn't use.