Inverse Kinematics Errors

Provide easy-to-use, extensible software for modeling, simulating, controlling, and analyzing the neuromusculoskeletal system.
POST REPLY
User avatar
Jeffrey Min
Posts: 3
Joined: Wed Mar 01, 2017 8:30 pm

Inverse Kinematics Errors

Post by Jeffrey Min » Wed Nov 15, 2017 4:49 pm

Hi everyone,

I am currently on a project using Gait2392 to study knee angles and moments. Something i am trying to wrao my head around is how IK is done and its errors. With scaling, I have managed to get errors under 0.015m but then IK errors end up being 0.40+ m. Is this simply due to soft tissue artefacts and the markers not staying as rigid bodies? If so, what would you recommend trying? I've tried playing around with marker weightings in IK with no success and the only other option I can think of is re-doing scaling. Thanks in advance!

User avatar
Thomas Uchida
Posts: 1794
Joined: Wed May 16, 2012 11:40 am

Re: Inverse Kinematics Errors

Post by Thomas Uchida » Wed Nov 15, 2017 9:06 pm

You might try adjusting the locations of the markers on the model (marker registration). It is common to iterate between Scaling and IK. There are some tips at the bottom of the Scaling and IK pages in the Confluence documentation:

- Getting Started with Scaling – Best Practices and Troubleshooting: https://simtk-confluence.stanford.edu:8 ... leshooting
- Getting Started with Inverse Kinematics – Best Practices and Troubleshooting: https://simtk-confluence.stanford.edu:8 ... leshooting

The following webinar might be helpful as well: "OpenSim Webinar: Tips and Tricks for Data Collection, Scaling and Inverse Kinematics in OpenSim", on the OpenSimVideos YouTube channel: https://www.youtube.com/watch?v=ZG7wzvQC6eU

User avatar
Jeffrey Min
Posts: 3
Joined: Wed Mar 01, 2017 8:30 pm

Re: Inverse Kinematics Errors

Post by Jeffrey Min » Thu Nov 16, 2017 3:56 pm

Thanks for that! I've been through the documentation and watched that video previously, so I'll continue to iterate. I think marker placement could also improve on our end and that should translate into better scaling and IK errors. :)

POST REPLY