Rapidly locating and accurately tracking the center of mass using statically equivalent serial chains

Abstract

This paper presents a center of mass (CoM) estimation technique that uses the statically equivalent serial chain (SESC). A SESC is a representation of any multilink branched chain whose end-effector locates the CoM. Locating the CoM is useful for modeling humans and humanoid robots. The method starts with an experimental phase involving a force plate and a motion capture system (MoCap) to construct a model to predict the CoM location. Subsequent motion of the subject updates the CoM model based on the MoCap node positions without need of a force plate, overcoming disadvantages of some other CoM estimation methods. The results show that the SESC methodology allows rapid and accurate real time estimation of the CoM. This paper includes the CoM estimation for a human from low-end and high-end MoCap and force plate sensors. The CoM in the plane of the force plate is shown to have an average error of less than 5 mm. A discussion of the accuracy of the vertical component of the CoM is also presented. Additionally, the presence of a static body in the workspace (a walker or chair, for example) to create stability in test subjects is presented.