Citation

Xiang Y, Rahmatalla S, Arora JS, Abdel-Malek K. Int. J. Hum. Factrors Model. Simul. 2011; 2(1-2): 111-126.

Copyright

(Copyright © 2011, Inderscience Publishers)

DOI

10.1504/IJHFMS.2011.041640

PMID

unavailable

Abstract

This paper presents an optimisation-based inverse kinematics (IK) method to compute realistic joint angles from marker-based positional data of a highly redundant three-dimensional digital human model. The effects of skin movement artefact and spine-shoulder joint coupling are alleviated by augmenting the optimisation problem with a skeletal discomfort function which is applied to the spine-shoulder-neck region only. The discomfort function ensures natural movement and avoids extreme joint motion. A key feature of the proposed error-minimisation-based method is that the error of the joint centre positions are not only minimised in the objective function, but also considered in the constraints so that the motion capture motion can be tracked more accurately. The methodology is demonstrated by three numerical examples, including vehicle-braking, throwing, and sprinting. The results showed that the proposed method is approved to be numerically stable and efficient to convert the motion capture data into joint space for a large-scale digital human model. Finally, the IK results are verified with the Visual3d software from C-motion.