An extended statically equivalent serial chain—Identification of whole body center of mass with dynamic motion

Tracking the whole body center of mass (CoM) trajectory of balance-impaired individuals with a personalized model is useful in the development of customized fall prevention strategies. A personalized CoM estimate can be obtained using the statically equivalent serial chain (SESC) method, but the subject has to perform an identification procedure to determine the set of subject-specific SESC parameters. During this identification, the subject must hold a series of static poses, some of which are unsuitable for balanced-impaired individuals. Can non-static poses be used to replace the static poses during SESC parameter identification? A new method that extends the range of postures used to determine SESC parameters is presented. It takes advantage of CoM dynamics and can be executed by predominantly using dynamic motions with a few static frames. Furthermore, it is implemented using a Kalman filter to allow automatic switching between the dynamic and static models. The proposed method was tested with motion data obtained from seven healthy adults using a Vicon motion capture system and an AMTI force platform. We found that dynamic motions could be used to estimate the SESC parameter and even reproduce ground reaction forces; however a small number of static poses are still required to determine the subject’s CoM position. The SESC-based CoM estimate obtained with this new approach was similar to that obtained using conventional full-static identification, except that the subject did not have to assume and maintain static poses. Our proposed extension of the conventional SESC method would facilitate its application in the field of neuro-rehabilitation, especially in patients who need balance training. This personalized CoM method could be applicable for patients who are not able to maintain a static posture. In addition, this method helps minimize the total identification time by increasing the number of usable recorded frames. •An extended SESC model using mostly dynamic data is used to estimate human CoM position.•The static-dynamic parameter estimation can be automated using a Kalman Filter.•Some static poses are still required in order to decrease estimation error.