simtk.org

Hi There

I'm working on applying motion capture of a side-step maneuver to an edited version of the 2392 model. I've just got a question regarding the starting point of the RRA Tool. I've noticed that, when I start at a particular point in time (close to or during foot contact with the force plate) the algorithm completes in about 20-30 seconds. However, if I set the starting point to about 100ms before contact with the plate, the algorithm takes almost 8 minutes.

Can someone please explain why this happens?

Hi Shaun,

While it's a little hard to anticipate your specific scenario/configuration specific issues, I can say that the time it takes to run RRA would be proportional to the number of time frames for which you have data. Although the difference in start time may be small, I'd assume that there're many more data points that get collected during foot contact so if you end up include 2 full foot-strikes instead of one, that the time would roughly double. The time to solve one frame will also depend on how long it takes to solve the optimization problem (to compute torques and controls for reserves).

Hope this helps,
-Ayman

Hi Ayman

Thanks for your help, that makes a little more sense to me. Just one last question: if I only have force plate data for one foot's contact with the ground, would the optimization algorithm struggle to compute torques and controls for the reserve actuators at points in the simulation for the time periods before this contact? I.e without GRF data the system struggles to compute these values, hence my exponential increase in processing time the further I go prior to ground contact

Shaun