Hi all, thank you so much for your work on this model. I have been reviewing some of the forum posts and wanted to seek any input on potential implementation of this model for forward dynamics simulations.
I am hoping to run some forward dynamics/contact-based simulations with a focus on the spinal model and spinal loading. Would you have any advice on which muscles or components I might be able to remove/restrict/modify to be more appropriate for a forward dynamics-based approach? I saw in some of the forum posts that this model hasn't been used for forward sims but figured i'd ask any and all for advice.
-Brandon
Recommended setup/modifications to optimize forward dynamics?
-
Brandon Fugger
- Posts: 5
- Joined: Tue Apr 26, 2022 3:09 pm
-
Jacob J. Banks
- Posts: 97
- Joined: Tue Jul 15, 2014 5:17 am
Re: Recommended setup/modifications to optimize forward dynamics?
Brandon,
Thank you for your interest in this model and kind words.
As for forward simulations, you are correct, we are unaware of anything published using this model. Since I left the Anderson Lab I do know that he worked with another student on something in MOCO, but unsure how far that got or if they ever got anything working. So, I really don't know what to tell you.
I have a little experience with the basics of forward simulations etc. (from my time at UMass and taking courses from Dr. Brian Umberger) and appreciate the challenges this many MTAs presents (and I have toyed with the idea of trying get it to work with this model). My inclination would be (in no order):
- try throwing CPU hours at it. If you have confidence in your abilities and have access to a Super Computer, see what it can do.
- Consider trying to consolidate segregated MTAs of similar function (~what you suggest)
- Consider just making a torque driven model then applying the kinematics to a model with MTAs. Knowing that you would ideally model the joint torques based off from their potential. This would ignore biarticular functions, but still should be a ~decent representation of things.
- Use a more basic gait model with a single rigid body for the trunk (or one with 1-2 joints), and then apply the kinematics of that model to this one. Or maybe try the Beaucage-Gaveraue lumbar model (or my SOpt model, which is basically the same thing), as it is 1/3 the complexity.
Clearly all these suggestions have limitations, but just spit balling. But I think it should be possible if you know what you are doing. I just wouldn't start out with this model, get confident doing forward sims with a simpler model and try adding levels of complexity. Otherwise, I'll still a quote from Einstein:
"... models should be as simple as possible, but not simpler than that".
Let me know if you have any ideas you want to bounce off etc. Sorry I can't be of more assistance than that.
Jake Banks
Thank you for your interest in this model and kind words.
As for forward simulations, you are correct, we are unaware of anything published using this model. Since I left the Anderson Lab I do know that he worked with another student on something in MOCO, but unsure how far that got or if they ever got anything working. So, I really don't know what to tell you.
I have a little experience with the basics of forward simulations etc. (from my time at UMass and taking courses from Dr. Brian Umberger) and appreciate the challenges this many MTAs presents (and I have toyed with the idea of trying get it to work with this model). My inclination would be (in no order):
- try throwing CPU hours at it. If you have confidence in your abilities and have access to a Super Computer, see what it can do.
- Consider trying to consolidate segregated MTAs of similar function (~what you suggest)
- Consider just making a torque driven model then applying the kinematics to a model with MTAs. Knowing that you would ideally model the joint torques based off from their potential. This would ignore biarticular functions, but still should be a ~decent representation of things.
- Use a more basic gait model with a single rigid body for the trunk (or one with 1-2 joints), and then apply the kinematics of that model to this one. Or maybe try the Beaucage-Gaveraue lumbar model (or my SOpt model, which is basically the same thing), as it is 1/3 the complexity.
Clearly all these suggestions have limitations, but just spit balling. But I think it should be possible if you know what you are doing. I just wouldn't start out with this model, get confident doing forward sims with a simpler model and try adding levels of complexity. Otherwise, I'll still a quote from Einstein:
"... models should be as simple as possible, but not simpler than that".
Let me know if you have any ideas you want to bounce off etc. Sorry I can't be of more assistance than that.
Jake Banks