Abstract

This paper reports an essential part of a wider research activity, which entails the development of a procedure for the Determination of the Optimal Prosthesis Architecture (DOPA) for a given upper limb amputee. A fundamental algorithm of the DOPA procedure performs the kinematic analysis of several prosthetic arm models (also with less than the six degrees of freedom normally required to correctly execute a generic manipulation task). The algorithm must simulate the execution of important daily living activities performed by a prosthesis and thus it requires reference trajectories of the hand. By means of experimental analysis, 59 trajectories of the hand of an able-bodied subject were acquired to identify a modality to correctly perform the corresponding tasks. This paper illustrates in detail the stages of task analysis, experimental acquisition and data processing in order to define the required reference trajectories. The obtained reference trajectories are a temporal succession of the hand pose (position and orientation). A customized algorithm automatically selects the most relevant poses to be considered for the definition of the reference trajectory. The hand pose is reported in the Cartesian Space by means of Natural Coordinates. In order to correctly execute a given task the pose error admitted for the end-effector of the different architectures is associated to each trajectory. In particular, the critical problem to express the orientation error is solved by means of the use of Spherical Rotation Coordinates.