Human-motion analysis of grasping/manipulating motion including time-variable function using principal component analysis

Recently, motion control technology is more and more developed. Robots are used not only in industry but also in human society. Hereafter, in order to extend the range of work and kinds of motion, it is needed to think about what human is. As is the case with voice recognition, for human-motion streaming of data also have important meaning. However, motion trajectory and force adjustment are different in each case of motion. In other words, it is difficult to estimate the characteristics from motion data. Hence, what component dominant is for works is needed to make clear. This component is called function of motion, and it is thought that human beings work by using some functions skillfully. In the conventional method in function of human motion, it is presupposed that functional mode is predefined such as grasping mode and manipulating mode. This paper proposes estimation method using principal component analysis (PCA) of functional mode for human motion including time-shifted function. By using proposal, the dominant function is directly estimated from motion information. Validity of the proposal is confirmed by the experiment. Experimental results in this paper are compared with the theoretical value.