build a new model with joints defined by myself

[Le Zhou]

Hi all, I meet a problem, I use depth sensor of ORBBEC to get some data of joints on the human body like left hip,right hip, torso.... and put them into a cvs file. What I want to do is to convert data I detected into opensim file and build a new model in the opensim, and then use them to calculate the hip flexion angle and hip abduction angle using the tools in the opensim. Because the depth sensor I used only can get 19 joints , I have to build a new model. Did anyone do the projects like I described above before? Please give me some tutoring ? The joints I get from the depth sensor is in the attachment file.

[Dimitar Stanev]

Hi,

I have done something similar with Kinect (https://github.com/mitkof6/KinectMotionCapture). The approach that I followed is as follows:

1) Use a pre-existing full body model (e.g. https://simtk.org/projects/runningsim) and do not change the joint and body definitions. The notion of a joint in your device has nothing to do with the actual joint, it just reports the position of a point on the body that it calls a joint.
2) Add 19 markers on the OpenSim model corresponding to the locations of the "joints" as provided by your device.
3) Export the trajectory of the "joints" to a .trc file format (see the example files in the OpenSim vX.X/Models folder) so that it can be recognized by OpenSim inverse kinematics.
4) Scale the model segments using a static motion from your device.
5) Perform inverse kinematics on the main motion.

https://simtk-confluence.stanford.edu:8 ... %27s+Guide

[Le Zhou]

Hi,

I have done something similar with Kinect (https://github.com/mitkof6/KinectMotionCapture). The approach that I followed is as follows:

1) Use a pre-existing full body model (e.g. https://simtk.org/projects/runningsim) and do not change the joint and body definitions. The notion of a joint in your device has nothing to do with the actual joint, it just reports the position of a point on the body that it calls a joint.
2) Add 19 markers on the OpenSim model corresponding to the locations of the "joints" as provided by your device.
3) Export the trajectory of the "joints" to a .trc file format (see the example files in the OpenSim vX.X/Models folder) so that it can be recognized by OpenSim inverse kinematics.
4) Scale the model segments using a static motion from your device.
5) Perform inverse kinematics on the main motion.

https://simtk-confluence.stanford.edu:8 ... %27s+Guide

Hi

Thank you for your help.I have a question : if I use the model built by others , how can I find the correct points on the model corresponding to my detected points ? I mean how can I define the joints on the model correctly but not just estimate the position?

[Dimitar Stanev]

Visually using markers.

https://simtk-confluence.stanford.edu/d ... ker+Editor

[Le Zhou]

Hi Dimitar

Are you willing to do project to help me? I have solved the convert data from depth sensor into trc file. But the depth sensor what I used is orbbec depth sensor. Can you help me to connect the opensim and orbbec sensor, I can offer what you want, Please.