Abstract
This paper proposes a motion pattern generator using dimensional reduction. A human motion pattern is measured by a motion capture system, and it is projected onto a low dimensional space based on principal component analysis. The low dimensional motion pattern is modified, so that the high dimensional motion pattern satisfies the motion condition, the dynamical consistency and the joint angle or torque limitation. The proposed method is applied to the motion pattern generation of the planar bipedal robot. The moon-walk performed by a human is transformed to the robot motion pattern. The ground contact states of foot links and a kinematic closed loop condition are given as the motion condition, and a floor reaction force condition is given for the dynamical consistency.
