A METHOD OF SKELETAL FINGER MODEL GENERATION CONSIDERING PHALANGE LENGTH AND JOINT ROTATION AXIS OF INDIVIDUALS

Anthropomorphic robot hands are increasingly being utilized to replace the human hand in remote operations or in dangerous places. For most robot hands the skeletal finger models have rotation axes for distal interphalangeal (DIP) joint and proximal interphalangeal (PIP) joint that are usually orthogonal to the finger bones. However, different people may have a different joint rotation axis. Therefore, the aim of this study is to construct a generation method for a finger skeleton model that considers the joint rotation center and axis of individuals, which can significantly influence the accuracy of a finger endpoint when controlling robot hands. This paper introduces generation methods for rotation axis and phalange length from the motion capture data of individuals. Furthermore, we compared the fingertip position generated by the proposed skeletal finger model to a finger model with rotation axes orthogonal to the finger bone. The result was a mean (standard deviation) distance between the fingertip and to all target positions of 2.6 mm (0.9 mm). Compared to a finger model with rotation axes orthogonal to the finger bone, the mean error of the finger model generated by the proposed skeletal finger model was reduced by 45%, and the standard deviation of a finger model generated by the proposed skeletal finger model was reduced by 47%. Additionally, the model generated by the new method had better accuracy and reproduced a more realistic finger motion