インバースキネマティクスを用いた現実感のある掌握姿勢の生成

概要

論文の詳細を見る
Research on human grasping has been a focus of interest in many fields ranging from basic sciences to those more interested in practical applications. The most significant efforts have been made in medicine and robotics. In computer graphics, the need for natural looking grasping has become urgent due to the emergence of synthetic actors. Most of the proposed algorithms have concentrated on the positioning of the hand onto the object without giving an explicit solution which takes the current posture of the whole body into account. Such a solution has been proposed by the authers. In this paper, an advanced form of the algorithm is proposed that is efficient in time and yields quite realistic results. A database is formed once at the beginning and used repeatedly as a guide. To obtain the posture of the body during the grasping of any given object, the related data is retrieved from the database and used first to estimate the posture of the hand. After the estimation of the hand posture, the whole body posture is estimated through the inverse kinematics. The weight of the object and the approximate torque applied to the wrist are also considered in the estimation of the hand posture.
社団法人電子情報通信学会の論文
1997-03-06