A Robust Algorithm for a System Identification Approach to Digital Human Modeling: An Application to Multi-fingered Hand Movement

摘要

A recent study [2] proposed a forward bio-dynamic model of multi-fingered hand movement. The model employed a physics-based heuristic algorithm for system identification of the model parameters, and succeeded in replicating measured multi-fingered flexion-extension movement. However, while the model itself is general and readily applicable to other bodily movements, the heuristic algorithm required empirical adjustments to initial setups, and was therefore difficult to generalize. This paper introduces a rigorous and more robust parameter estimation algorithm to enhance the intended general modeling approach for digital human movement simulation. The algorithm is demonstrated by solving the same modeling problem posed in [2].