An Approach to Determine a Human Joint Axis Using Force-Controlled Motion and the Power Iteration Method

摘要

The non-invasive determination of human joint axes is an important task in rehabilitation procedures and human musculoskeletal modeling. While most existing approaches apply for this purpose state-of-the-art infrared marker-tracking systems or inertial measurement units (IMU), this paper presents a novel approach to determine the flexion/extension axis of a human joint using force-controlled tactile motion and identifying the sough axis by the power iteration method. The basic idea is to iteratively impart a computed load to a limb until the ensuing motion aligns with the direction of the applied load. This is shown to correspond to the convergence of the power iteration or Krylov method to the eigenscrew of the corresponding impedance matrix of the limb, without the need of establishing the impedance matrix explicitly by measurements. The paper verifies the proposed procedure by a mimicry mechanical device of the upper limb, composed of a revolute joint representing the elbow joint and a set of parallel connected flexible suspensions imitating the local movements of the soft tissue (Fig. 1). In order to unveil the internal motion dependencies of the compliance model, the eigendecomposition of the compliance matrix is analyzed, showing that the power iteration method can be applied in force space to identify the position and direction of the sought revolute joint axis. The experimental results show a promising perspective of the proposed approach.