View Full Version : Getting Rotation Angles From the VICON

06-25-2011, 07:22 PM
We are using VICON system for Kinematic analysis in our lab. I use my own model for motion capturing and I don't use any of VICON templates for humman motion capturing and evaluating the joint angles. What I do is that I use the marker position (x,y,z) obtained from the Vicon and I use the code that I have written to obtain the angles. My question is: If I use the Vicon joint angles instead of my code, are the angles (Euler, helical....) returned by the Vicon, when I check joint kinematics option to save in a Csv file, bio-mechanically meaningful to evaluate flexion/exten, external/internal rotation and etc of human joints? or they are just meaningful when I use the standard templates of the Vicon?

06-27-2011, 05:41 AM
If you run the "Export CSV (Comma Separaterd Variable)" pipeline operation in Vicon Nexus, and tick the "Kinematics" option, the software will export the kinematics calculated by the "Fit Subject Motion" pipeline operation, assuming this operation has been previously executed. The outputs in the exported CSV file depend on the hierarchical model defined in the Vicon Skeletal Template (VST) file and how this was calibrated to fit marker placement on the subject being captured. Normally, this template is used to define a labelling template for the automatic identification of markers, and NOT to define a biomechanically valid model. The tools in Nexus called "Labelling Template Builder" lets you set up an approximate biomechanical model suitable for automatic labelling, where the origin of the different segments are at the actual position of the marker, and not offset to approximate the anatomical position of the joint centre. In other words, if you set up your VST file using the "Labelling Template Builder", you will certainly NOT get a biomechanically valid model.

If, on the other hand, you hand-edit the VST file to apply relevant offsets (for example by using medial and lateral joint markers and specifying the joint centre half-way between), it is possible to set up a customized biomechanical model which outputs valid kinematics. This is certainly an expert task, though how difficult it is depends on what requirements you have. A single-joint model, for example, could be relatively easy, whereas a full-body model would, of course, be a lot more complicated.

The "Fit Subject Motion" operation uses a technique known as "kinematic fitting" to find the optimized solution given correctly identified markers. There are a couple of advantages - for example, the kinematic fitting technique finds an optimized solution where segment lenghts are fixed to their calibrated lengths, unlike several commonly used models (such as the Conventional Gait Model, known as Plug-in Gait in the Vicon software) which allow the bone lenghts to change when deriving the model's anatomical skeleton from the measured marker positions.

In summary: assuming you've created your labelling template (VST) file using the Labeling Template Builder, you will not get valid kinematics. The standard VST templates from Vicon (i.e. the Plug-In Gait ones) output kinematics based on the kinematic fitting technique, with joint centres offset according to the Conventional Gait Model. Note that these are not exactly the same as the ones output by the actual Conventional Gait Model, mainly because of the aforementioned difference in how kinematic fitting keeps the model's segment lenghts fixed and the Conventional Gait Model does not.

Hope this helps!

Best regards,

07-09-2011, 10:04 PM
Thanks for your comprehensive answer.