Center of mass estimation during non-cyclic activities: Comparison of marker-based methods and their fusion with ground reaction forces

Accurate estimation of the whole-body center of mass (CoM) is essential for assessing human stability and postural control. However, selecting a suitable estimation method considering the nature of the activity, availability of equipment such as force plates measuring ground reaction forces (GRFs), and the time needed for data collection and processing is challenging. This study compares four methods for estimation of the 3D CoM position and velocity - "Pelvis Markerset" (PM), "Pelvis Markerset & GRFs" (PMG), "Whole-Body Markerset" (WM), and "Whole-Body Markerset & GRFs" (WMG) - across 4 activities classified as Static (e.g., standing with eyes closed), and 10 as Dynamic (e.g., picking up an object from the ground). Using the root mean square (RMS) of "external force residual" as a performance metric, we found that in the Static group, all the methods performed similarly for both position and velocity estimation. During the Dynamic activities, for position estimation, the pelvis-based estimates showed higher residuals compared to the whole-body methods (p<0.001, Cohen's d=3.03). For velocity estimation, the residual of WMG was similar to WM, and both outperformed PM (p<0.001, d=3.17); meanwhile, PMG achieved lower residuals than PM (p<0.001, d=2.26). Given our results, we recommend the WM method as it performed well and did not require fusion with the GRFs. The PM method can be used in activities similar to the Static group, during which markers on the pelvis reflect whole-body kinematics. When the GRFs are also measured, it is possible to improve the velocities estimates of this method using the Kalman filter.