Hi everyone,
I am tyring to predict the torque my prosthetic knee design will require to replicate able bodied kinematics.
I wish to ensure the hip torque is the same as able-bodied and reach the same able-bodied kinematics.
So I have my hip torque to apply to segment model, I have the kinematics I want to achieve, how would I predict the knee torque I require to achieve these kinematics under the hip torque environment.
The mass properties of the limb have changed which is why I need to simulate the movement rather than assume able-bodied knee torque is appropriate.
Can someone please help with the best way forward as I have been unable to find a similar example in the documentation.
Also if you think there is a simpler way to achieve this, please do let me know.
Many thanks,
Caitlin
Predict prosthetic knee torque required
- Caitlin Edgar
- Posts: 12
- Joined: Thu Mar 17, 2022 2:23 am
- Nicholas Bianco
- Posts: 1003
- Joined: Thu Oct 04, 2012 8:09 pm
Re: Predict prosthetic knee torque required
Hi Caitlin,
Happy to help, and I have a couple of questions. Is this a powered prosthetic or passive knee device? If it's a passive device, you could use inverse dynamics (assuming you're working with experimental gait data) to see what the knee joint moment is for your motion, and then see if the passive device produces the appropriate forces based on the knee joint angle and stiffness properties. This is probably the simplest approach.
If you wanted to predict kinematics rather than joint moments, you would need a predictive simulation approach to see how close you can track the desired knee kinematics based on the prosthetic model.
-Nick
Happy to help, and I have a couple of questions. Is this a powered prosthetic or passive knee device? If it's a passive device, you could use inverse dynamics (assuming you're working with experimental gait data) to see what the knee joint moment is for your motion, and then see if the passive device produces the appropriate forces based on the knee joint angle and stiffness properties. This is probably the simplest approach.
If you wanted to predict kinematics rather than joint moments, you would need a predictive simulation approach to see how close you can track the desired knee kinematics based on the prosthetic model.
-Nick
- Caitlin Edgar
- Posts: 12
- Joined: Thu Mar 17, 2022 2:23 am
Re: Predict prosthetic knee torque required
Hi Nick,
Thanks so much for your reply!
Yes the device will be passive.
I agree inverse would be the simplest approach, however I don't have experimental gait data of amputee gait, only able-bodied. As such I was worried when doing inverse dynamics the effects of a passive prosthetic foot vs able-bodied foot would not be taken into account in GRF and joint angles etc. Maybe I could try both and see how different they are or even start with inverse before moving up complexity?
Would you be able to explain the predictive approach that would best suit this problem as I am really interested in how this would work?
Thanks for your help!
Caitlin
Thanks so much for your reply!
Yes the device will be passive.
I agree inverse would be the simplest approach, however I don't have experimental gait data of amputee gait, only able-bodied. As such I was worried when doing inverse dynamics the effects of a passive prosthetic foot vs able-bodied foot would not be taken into account in GRF and joint angles etc. Maybe I could try both and see how different they are or even start with inverse before moving up complexity?
Would you be able to explain the predictive approach that would best suit this problem as I am really interested in how this would work?
Thanks for your help!
Caitlin
- Nicholas Bianco
- Posts: 1003
- Joined: Thu Oct 04, 2012 8:09 pm
Re: Predict prosthetic knee torque required
Hi Caitlin,
If your end goal for this design is for the prosthetic wearer to reproduce typical/healthy joint moments and kinematics then I think you could use the data that you have for now. For example, you could use the healthy data to optimize the passive force parameters of your device to get as close to the ankle joint moment as possible based on the ankle angle.
You could then test this design in an experiment or in a follow-up simulation. Since you probably won't be able to fully replicate an ankle joint moment with a passive device, you will likely get different kinematics and GRFs when testing the device, so the design iteration will begin there. There's lots of ways continue from that point; if you want to chat about simulation-based ideas, I'm happy to continue discussion here.
-Nick
If your end goal for this design is for the prosthetic wearer to reproduce typical/healthy joint moments and kinematics then I think you could use the data that you have for now. For example, you could use the healthy data to optimize the passive force parameters of your device to get as close to the ankle joint moment as possible based on the ankle angle.
You could then test this design in an experiment or in a follow-up simulation. Since you probably won't be able to fully replicate an ankle joint moment with a passive device, you will likely get different kinematics and GRFs when testing the device, so the design iteration will begin there. There's lots of ways continue from that point; if you want to chat about simulation-based ideas, I'm happy to continue discussion here.
-Nick
- Ross Miller
- Posts: 372
- Joined: Tue Sep 22, 2009 2:02 pm
Re: Predict prosthetic knee torque required
Hi Caitlin,
If what you're looking to do is something like this study (have a model with a prosthetic limb "track" able-bodied walking data as closely as possible):
https://www.sciencedirect.com/science/a ... 9006002880
...I've used MocoTrack or MocoStateTrackingGoal for that and it works quite while. The setup is conceptually the same as the example2Dwalk code that comes with Moco. Modeling the prosthetic knee is more complicated than a prosthetic ankle but there are some reasonable options in Moco for that.
I'm not sure it will be possible to have identical hip mechanics vs. able-bodied mechanics since the ankle is a net energy generator in able-bodied gait but a net energy loser (or energy-neutral at best) with a passive prosthetic leg. This is one thing the "optimal tracking" approach can help with (minimize such deviations, rather than require them to be zero).
Ross
If what you're looking to do is something like this study (have a model with a prosthetic limb "track" able-bodied walking data as closely as possible):
https://www.sciencedirect.com/science/a ... 9006002880
...I've used MocoTrack or MocoStateTrackingGoal for that and it works quite while. The setup is conceptually the same as the example2Dwalk code that comes with Moco. Modeling the prosthetic knee is more complicated than a prosthetic ankle but there are some reasonable options in Moco for that.
I'm not sure it will be possible to have identical hip mechanics vs. able-bodied mechanics since the ankle is a net energy generator in able-bodied gait but a net energy loser (or energy-neutral at best) with a passive prosthetic leg. This is one thing the "optimal tracking" approach can help with (minimize such deviations, rather than require them to be zero).
Ross