機械知覚&ロボティクスグループ
中部大学

Local Image Feature 口頭発表

ARToolKitを用いたヒューマノイドロボットのモーション推定

Author
清水彰一, 藤田征士, 草富省吾, 伊藤孝浩, 藤吉弘亘.
Publication
人工知能学会SIG-Challenge研究会, pp. 44–49, 2008

Download: PDF (Japanese)

The RoboCup SSL Japanese teams have proposed a new league that uses humanoid robots and a global vision system. The humanoid robot is able to execute motions by selecting them from a list implemented in advance. While the humanoid robot is moving, it cannot change the motion. Therefore, it is important to estimate the motion by observing the humanoid robot using global vision systems. We propose a method of motion estimation for humanoid robots using ARToolKit. Our method uses the ARToolKit to obtain this observed information, which includes the position and pose of the humanoid robot. We show that the proposed method can estimate the motion within 15 frames from the frame in which the current motion was started.

前の研究 次の研究