simtk.org

Hi all,

I'm curious as to whether you would expect to see the excitation and activation signals be as instantaneous as seen below/attached?
The main reason I ask is that we're seeing significant delays between the solutions from moco and experimental EMG (see also below/attached). This would be partly explained if there were a more significant delay between excitation and activation signals - excitation is more or less indicative of EMG, correct?
Could the lack of delay between excitation and activation be a result of appending the optimisation-friendly DeGrooteFregly muscles?
P.s. This is hopping data, and these findings are consistent between hop heights and frequencies.

Thank you!
Luke

Hi Luke,

I think this looks pretty normal if you're using the default activation time constants for DGF (0.015 s for activation, 0.060 s for deactivation). They are on the faster side especially for "slower" muscles like soleus. They can be modified on a muscle-by-muscle basis (if you aren't doing that already). The XML tags are :

<activation_time_constant>0.015</activation_time_constant>
<deactivation_time_constant>0.060</deactivation_time_constant>

Winters & Stark (1985) I think uses the same activation dynamics as DGF and describes a procedure for specifying the time constants as functions of muscle mass and fiber type:

tau_act = 0.005 + 0.2*m*(1-FT)^2
tau_deact = 0.030 + 0.5*m*(1-FT)^2

m = muscle mass in kg
FT = fraction of fast-twitch fibers (from 0-1)
tau's will have units of seconds

https://doi.org/10.1109/TBME.1985.325498

Umberger et al. (2003) has a simpler approach (time constants are linear functions of FT only) but I think the activation dynamics formula in his muscle model is different from DGF; unsure if that is important here.

https://doi.org/10.1080/1025584031000091678
https://doi.org/10.1080/10255840612331393266

Hope this helps,
Ross

Much appreciated. Thank you, Ross!