抄録
This paper describes task estimation in upper-limb exoskeleton power assist robot with perception assist. In the perception assist by using exoskeleton robot, it is important to recognize objects which user has, because robot's assist depends on objects which user has. Although object recognition by using image processing has proposed, the method is susceptible to environmental factor like a luminance. Therefore, in this paper, the method in order to recognize an object which user has is proposed. The method use image processing in combination with task estimation by motion of user s arm. To estimate the motion of user's arm, the feature quantity which is calculated by hand vector is used. The effectiveness of the proposed method was evaluated by performing the experiments.
