Predictive Simulation: Who are engaged in this area?

[Jakob Welner]

Hi,

Does anyone here know about groups or people around the world who are currently working with Predictive Simulation in biomechanics? - Mainly regarding gait but not necessarily.

I know of:
- Hartmut Geyer who has had a few students continuing on his and Hugh Herr's Muscle-reflex Model from 2010.
- Using OpenSim I know of Scott Delp along with Tim Dorn, Jack Wang and Jennifer Hicks, also based on the reflex model by Geyer and Herr and whose PredictiveSim system I am currently using.
- Then there is Yoonsang Lee et al. with their Many-Muscle project which appears to be half predictive control and half motioncapture based
- And finally Thomas Geijtenbeek with his Flexible Muscle-Based Locomotion, but that's slightly different again.

Are there anyone else?

Cheers and thanks
- Jakob Welner

PS: I am sorry if this is a little off topic for the forum as it is not OpenSim specific but I am unaware of better places to ask. Are there any public forums for general discussion on biomechanics simulation or maybe mailing lists or do people usually discuss internally with their group? I am new in this field and would love a place to discuss stuff in general terms.

[Christopher Dembia]

Carmichael Ong in Scott Delp's lab is also working on predictive simulation.

Are you only considering single-shooting style predictive simulation (as opposed to direct collocation)?

Why do you consider Thomas' work to be different?

[Jakob Welner]

Hi Chris,

Thank you for your reply. I have looked up Carmichael Ong but he does not appear to have any publications on the matter?

As for single-shooting stype as opposed to direct collocation I am actually not aware of those terms' relation to Predictive Controllers? Google didn't seem cooporative in explaining it either.
I must admit that I am very new to this topic and while my supervisors have general experience in biomechanical simulation, none of them have insight into Predictive Controllers so I am on my own in that regard and has to rely on what I find in articles and questions I ask here.

As for Thomas' work, I read his article a while ago but just re-read it and I see that I must've confused the SIMBICON target angles for using motion capture data so you're right, it should not be considered different and I'm glad you pointed it out

Thanks

[Christopher Dembia]

Thomas' review paper probably talks about direct collocation. http://www.goatstream.com/research/publ ... GF2012.pdf

Here's one of now a handful of papers using direct collocation: http://www.ncbi.nlm.nih.gov/pmc/articles/PMC2849893/.

Direct collocation removes the need for developing "predictive controllers" (reflexes and such), but would still be classified as "predictive simulation."