AnyBody-Xsens workflow (webinar): kinetic modeling on the field

On November 30th 2017 Anybody and Xsens hosted a webinar together. This webinar showed both new software releases of Xsens (MVN 2018) and AnyBody (AnyBody v7.1.0) and discussed the pipeline of importing Xsens data into AnyBody. 

Please watch the webinar here:


At the end of the webinar some great questions were asked. See the Q&A here:



The recommended operating temperature of the MVN System is between -10°C and +50°C ambient temperature. Is it OK to use this system to record skiing sports motion?

Xsens: This is not a problem. The trackers are mounted on the body, causing the trackers to be in a somewhat warmer environment compared to the surrounding. Next to that, normally with skiing there is another layer of clothing on top, protecting them a bit more from the surroundings. We have multiple customers involved in jumping/alpine skiing and when I work with those customers, we tend to put the motion tracker of the foot on the skiing boot (on the foot-part, not the moving ankle/shank-part of the boot) with some duct tape and this never causes any problems in terms of temperature.

Do you need new Xsens IMU Sensors for the new Software version? Ours has been bought in 2010.

Xsens: In 2010 there was an older hardware generation. This old hardware generation has been declared end-of-life 3 years ago, and the current software development is only compatible with the new hardware generation (released 3 years ago and still current). For upgrading an MVN system, we have special offers due to the release of MVN 2018. Please click here.

Where comes the contact information from in the Xsens system? Xsens: Contacts are extracted from the motion data.
I was wondering if the "magnetic immunity" means that the magnetometers are no longer used? If they are still being used, how is immunity achieved?

Xsens: The magnetometers are still being used whenever possible. The new MVN software has been validated, including magnetically challenging environments, among which was Angelos’ study by walking repeatedly over force plates (see his article: Estimation of Ground Reaction Forces and Moments During Gait Using Only Inertial Motion Capture). In a few weeks, we will release the new MVN whitepaper with more information and results on the accuracy, consistency and repeatability of MVN 2018. Please monitor your email for the Xsens newsletter, or visit the MVN Analyze 2018 download page.

Is the accuracy of the kinematic reconstruction sensible to the placement of the XSens IMU sensors on the bony segments? Do they have to be placed in specific anatomical landmarks?

Xsens: The placement of Motion Trackers of the MVN system are quite flexible in terms of orientation, because the calibration procedure will calculate the segment orientation based on any orientation of the trackers. It is best to follow the recommendations in the tutorial video ( about placing the hardware of the MVN Link or MVN Awinda system for obtaining the most accurate motion capture data. The recommendations are based on locations on the body with the least amount of loose skin and muscle mass to minimize the effects of skin artifacts (skin vibrations, for example).

Can we use Xsens MVN 2018 to analyze the motion data recorded by using the older version of the software? If yes, is the performance better than old version? Xsens: MVN 2018 requires an Npose (or Tpose) +Walk calibration in order to get the enhanced data accuracy provided by the new MVN engine. Therefore, processing old MVN files in the new software is not possible. You can open your old recordings in MVN 2018 and look at them, but they will be read-only.
How accurate are predicted GRFs compared to measured ones? Any limitations in using predicted GRFs compared to measured ones? Can you predict GRFs accurately for any movement (e.g. movement under perturbations) or is this successful mostly for normal typical movements?

AnyBody: We present the results for walking in the presentation. Overall strong to excellent agreement was found based on Pearson correlation coefficients, while RMS differences range from 7% to 34% with respect to the magnitude of the respective GRF&M component. Depending on the requirements of your application this may impact your analysis, but among other ambulatory solutions it provides you with motion freedom and promising accuracy. Whether that is acceptable highly depends on the parameters you are looking into. For instance, overall the method should perform better in tracking relative changes in kinetics with respect to a subject’s baseline rather than absolute values of the variables.A paper that validates the GRF&M prediction method in various sports-related movements was published by Skals et al. 2017. We anticipate that the performance of the GRF&M prediction will be similar if Xsens MVN input is used for such movements, given the findings of the study using Xsens input for gait analysis.In addition to that paper, there is this study by Eltoukhy et al. which uses ground reaction force prediction with Parkinson’s patients.

Do you have different models for non standard bodies (fat, slim, long legs, short legs, long arms, etc.?

Xsens: In MVN you can scale the biomechanical model to the body dimensions of your subject, related to total body height, arm span, hip width/height and a few more, all assuming symmetrical conditions. More realistic scaling can be done in AnyBody.

AnyBody: You can also customize the human body model used by the AnyBody Modeling System. There are many readymade scaling laws, which scale the models according to size (e.g., short legs, long arms etc.) and distribute masses according to body fat percentage. You can also define your own customized scaling laws, if you’d like to deviate from the default laws. A full list of such scaling laws can be found here.

Is there any difference in the AnyBody model used in the study presented by Angelos to the Anybody model released today, that was mentioned by Ananth?

AnyBody: The model presented by Angelos in the webcast and the demo model being released in the new AnyBody v7.1 are essentially based on the same simulation principles and workflow. There are however some minor differences. For example, both models are based on different datasets. Therefore, they may also differ in the ground reaction force prediction settings, which include the floor level, friction coefficient etc. More on the GRF prediction settings here.

We are currently having problems with running the kinematics analysis, due to virtual markers on the two foot segments obstructing each other as these segments seems to travel through each other during the movement. Could this be due to errors in the Xsens MVN Link calibration?

Xsens: When setting up the Xsens MVN system (either MVN Link or MVN Awinda), it is important to pay very good attention to the body posture of your subject, especially during the Npose/Tpose. Any deviations from the assumed posture during the static part of the calibration will result in an offset in the data. Please see the tutorial video on about how to perform an Npose(or Tpose) + Walk calibration for more information on how to accurately calibrate. After calibration, remember to walk around for 20 seconds for a filter warm up and to feed some additional motion data into the system so that it will give you accurate and consistent data.

AnyBody: With the very latest version of the Xsens MVN, the position of the 2 markers placed on the AnyBody model’s feet (Medial and lateral toes) need to be changed a little. We are currently testing the demo model with these new settings and will release them soon. If you are using a BVH file generated from the latest MVN version, this change could affect the values of your ankle angles. I of course cannot say for sure if this is the cause of your issue. If you reckon that it is, please move the model (red) markers on the feet a couple of centimeters vertically upwards so that it’s closer to the plane forming the base of the foot.

Is there any link between Catia and Anybody instead of solidworks ?

AnyBody: There isn’t a one-click solution for Catia-AnyBody like we have for Solidworks. You can however export individual Catia parts as STL files and read them into AnyBody. The advantage of the AnyBody Exporter for SOLIDWORKS is that it doesn’t just read in individual parts but automatically interprets the entire assembled mechanism in the AnyBody format. With individual STL parts imported from CATIA, you can manually construct joints & assembly constraints in AnyBody.


Was this article helpful?
0 out of 0 found this helpful
Do you have a question? Please post your question in our Community Forum