Multi-Modal Analysis of Human Motion From External Measurements

Abstract

The “analysis” or “inverse dynamics” problem in human motion studies assumes knowledge of the motion of the dynamical system in various forms and/or measurements of ground reaction forces to determine the applied forces and moments at the joints. Conceptually, methods of attacking such problems are well developed and satisfactory solutions have been obtained if the input signals are noise free and the dynamic model is perfect. In this ideal case, an inverse solution exists, is unique, and depends continuously on the initial data. However, the inverse solution may require the calculation of higher order derivatives of experimental observations contaminated by noise—a notoriously difficult problem. The byproduct of errors due to numerical differentiation is grossly erroneous joint force and moment calculations. This paper provides a framework for analyzing human motion for different sensing conditions in a manner that avoids or minimizes the number of derivative computations. In particular, two sensing modalities are considered: 1) image based and 2) multi-modal sensing: combining imaging, force plate, and accelerometery.