A Robot Arm Motion Generation Based on Learning of Objects Shape and Human Motion
Sho Tajima, Tokuo Tsuji, Yosuke Suzuki, Tetsuyou Watanabe, Ken’ichi Morooka, Kensuke Harada, Masatoshi Hikizu, Hiroaki Seki · The Proceedings of JSME annual Conference on Robotics and Mechatronics (Robomec) · 2018
We propose the method for planning a robot motion of daily tasks by learning the relationship between objects shape and human motion. Since objects have different shapes even in the same category, it is difficult for a robot to perform tasks automatically. Motion generation of a robot consists of object recognition, shape estimation, and estimation of a motion. The motions are estimated by learning the relationship between objects shape and human motion using linear regression analysis in advance. We focus on the task of pouring water with a plastic bottle. Finally we evaluate a motion position posture estimated from objects shape and show the effectiveness of our proposed method.