Improvements in Measure-Based Simulation of the Human Lower Extremity

摘要

Measure-based simulation involves reproduction of the motion of a dynamic system based on measurements performed for certain parameters. In this paper, we regard a method for polishing human data obtained from existing gait motion reproduction systems such as the VICON motion capturing system. Such systems allow one to automatically reconstruct limb lenghts, joint center locations and limb motion from the measured data obtained by tracking, through infrared cameras, special reflective markers applied on the patient skin. However, due to generic assumptions on limb functionality, biomechanical models of the human lower extremity based on these measurements contain systematic errors on joint center location that lead to unrealistic bone-length variations in the simulation. The objective of this paper is to develop a method that uses a given erroneous data set and from this corrects prediction of joint-center location by analysing systematic errors using mechanism theory. To this end, a virtual mechanism is introduced with the unknown bone locations as one set of links, and the measured varying-length limbs as a second, parallel mechanism. By establishing the offset between real and assumed skeleton through optimization techniques, the location and geometry of the real bones can be determined. The model is extended to inverse dynamics to evaluate the effects of errors on the resultant net joint torques. It is shown that the presented method allows patient-specific adaption of the model. Moreover a more exact animation of gait motion is possible. Finally, due to the self-correcting algorithm, the method allows crude marker placement, simplifying the motion-capturing preparation procedure.