error in inverse dynamics

Provide easy-to-use, extensible software for modeling, simulating, controlling, and analyzing the neuromusculoskeletal system.
POST REPLY
User avatar
Diana Toderita
Posts: 5
Joined: Tue Oct 10, 2023 2:19 am

error in inverse dynamics

Post by Diana Toderita » Fri Jan 12, 2024 6:37 am

Hello!

I have an issue with the inverse dynamics process using a lower limb model for gait.

The ground reaction forces and inverse kinematics look alright. However, the inverse dynamics shows a very large peak in the ankle torque towards the end of stance, that should not occur. This artefact then appears in all other joints for this limb. The contralateral limb doesn't have this issue. I have attached the force plate data file, inverse kinematics and inverse dynamics outputs for the ankle joint. The force vector looks to be acting in the correct place on the calcaneus.

The problem persists with other trials for this participant, and with other participants as well.

Please can someone help to understand where the issue is?

Thank you!
Diana
Attachments
ID_output_right.txt
(2.86 KiB) Downloaded 250 times
IK_results_right.mot
(2.86 KiB) Downloaded 273 times
GRF_right.mot
(66.83 KiB) Downloaded 242 times

User avatar
Ton van den Bogert
Posts: 166
Joined: Thu Apr 27, 2006 11:37 am

Re: error in inverse dynamics

Post by Ton van den Bogert » Fri Jan 12, 2024 9:00 am

I plotted your data.
Untitled.png
Untitled.png (11.14 KiB) Viewed 2676 times
The joint moment is close to 400 Nm at the peak, which is far too large indeed. The GRF at that time is 1000 N (which is normal), so this means that the center of pressure is about 0.4 meters from the ankle joint. Unless your subject is wearing skis, that is not possible.

The ankle joint angle looks normal, although normally it will go a little above zero. This depends somewhat on how the local coordinate systems are defined.

The high joint moment suggests that the spatial registration between motion capture and force plate is incorrect. I can possibly say more if you post the raw marker data for this trial.

Ton van den Bogert

Code: Select all

q = importdata('IK_results_right.mot','\t',11);
F = importdata('GRF_right.mot');
M = importdata('ID_output_right.txt');

subplot(3,1,1)
plot(M.data(:,1),M.data(:,2));
title('ankle moment');

subplot(3,1,2);
plot(F.data(:,1),F.data(:,[3 4]));
title('GRF')

subplot(3,1,3);
plot(q.data(:,1),q.data(:,2));
title('ankle angle')

User avatar
Diana Toderita
Posts: 5
Joined: Tue Oct 10, 2023 2:19 am

Re: error in inverse dynamics

Post by Diana Toderita » Fri Jan 12, 2024 9:55 am

Many thanks for your reply.

I have attached the raw marker data for the trial.

Any insights would be much appreciated!
Attachments
motion_data_right.txt
(120.79 KiB) Downloaded 266 times

User avatar
Ton van den Bogert
Posts: 166
Joined: Thu Apr 27, 2006 11:37 am

Re: error in inverse dynamics

Post by Ton van den Bogert » Fri Jan 12, 2024 1:15 pm

The spatial registration between raw motion capture data and force plate is correct, the center of pressure is exactly were it is expected and not 0.4 meters off. I plotted the horizontal position of COP and a few markers in the lower right leg and foot.
Untitled.png
Untitled.png (16.29 KiB) Viewed 2583 times
The next thing to check would be that the IK is done correctly. For ankle moment, what matters is the position of the ankle joint relative to the center of pressure (COP) of the GRF. Play back the result of the IK and show the GRF vectors in the same view. Does the skeleton track the markers? Is the vector located in the right foot? Is the ankle joint within 0-20 cm of the center of pressure?

It is also possible that the center of pressure information was not imported correctly. Search the forum for "center of pressure", it has been discussed a few times.

Ton van den Bogert

User avatar
Mohammadreza Rezaie
Posts: 408
Joined: Fri Nov 24, 2017 12:48 am

Re: error in inverse dynamics

Post by Mohammadreza Rezaie » Fri Jan 12, 2024 9:19 pm

Hi,
Inspecting your markers data, I could see early heel rise in the right leg (a plantarflexion peak is evident between 3.2 and 3.4 s, the mid-stance phase), which could increase the ankle joint moment. Is he/she a pathological subject, a prosthesis/orthosis user? This could explain why the ankle angle didn't come to dorsiflexion at all. When syncing the markers and GRF, everything looks good; however, you may need to transform your experimental data to match OpenSim coordinate system (XYZ). It's also a good practice to associate IK and GRF motion data before running ID tool. If this looks good, check the force, point, and torque identifiers in the external loads XML file. Please see:
https://simtk-confluence.stanford.edu:8 ... ap%29+Data
https://simtk-confluence.stanford.edu:8 ... h+a+Motion
https://simtk-confluence.stanford.edu:8 ... ernalLoads

Hope this helps,
-Mohammadreza

User avatar
Diana Toderita
Posts: 5
Joined: Tue Oct 10, 2023 2:19 am

Re: error in inverse dynamics

Post by Diana Toderita » Mon Jan 15, 2024 8:18 am

Hello both,

Many thanks for your help.

I have now used the matlab code OpenSim provided to export the .trc and .mot files (attached) from the .crd and they should be aligned with the OpenSim coordinate system. Previously I was exporting them from Vicon.

I played back the IK. The skeleton tracks the markers and the ground reaction force is located in the right foot (pictures attached).

I have also checked the identifiers in the external loads XML file, and it all looks fine (hoping I am not making a silly mistake...). I have also attached the .xml GRF setup file.

Also, this artefact occurs for the right limb only, the moments for the left limb look normal and do not have this large peak. Could this be a modelling issue in the right joint?

Any further suggestions would be welcome!

Thank you,
Diana
Attachments
Walk_Peak_front.PNG
Walk_Peak_front.PNG (71.42 KiB) Viewed 2510 times
Peak_lateral.PNG
Peak_lateral.PNG (98 KiB) Viewed 2510 times
P14_walk2_test3.txt
(207.15 KiB) Downloaded 280 times
P14_walk2_GRF_test3.mot
(164.33 KiB) Downloaded 242 times
P14_SelfSelected_walk_2_grf.xml
(2.74 KiB) Downloaded 274 times

User avatar
Mohammadreza Rezaie
Posts: 408
Joined: Fri Nov 24, 2017 12:48 am

Re: error in inverse dynamics

Post by Mohammadreza Rezaie » Mon Jan 15, 2024 8:33 am

Hi, according to the provided files, I could see that the identifiers are not set correctly. The right leg requires the following identifiers:

ground_force_1_v
ground_force_1_p
ground_moment_1_m

Please double check this and let me know if it can fix the issue.

User avatar
Diana Toderita
Posts: 5
Joined: Tue Oct 10, 2023 2:19 am

Re: error in inverse dynamics

Post by Diana Toderita » Mon Jan 15, 2024 8:54 am

Hi, the left limb is the first one to step on the force plate 1, and the right limb uses force plate 2. This is why I associated the right limb with identifiers 2 and the left limb with identifiers 1.

User avatar
Mohammadreza Rezaie
Posts: 408
Joined: Fri Nov 24, 2017 12:48 am

Re: error in inverse dynamics

Post by Mohammadreza Rezaie » Mon Jan 15, 2024 9:07 am

Oh, sorry. I didn't check the rest of the force file and thought you sent one FP data only, similar to the former files. If you share the scaled model and IK output of this subject, troubleshooting would be easier.

So, markers and GRF of the right leg are well-aligned and the XML force file is correct. The associated IK output and GRF (your image) seems fine as well. One other thing to check is the mass properties of the scaled model and the segments.

User avatar
Diana Toderita
Posts: 5
Joined: Tue Oct 10, 2023 2:19 am

Re: error in inverse dynamics

Post by Diana Toderita » Fri Jan 19, 2024 6:03 am

Hello both,

Thank you very much for your advice and help, much appreciated. I managed to fix the model, it turns out there was a modelling issue in the definition of the joints.

All the best,
Diana

POST REPLY