Abstract
This paper studies the trajectory planning of a biped locomotive robot from the following standpoints.
* Only the position and velocity of the center of gravity of robot in the forward direction for each step are given a priori.
* It is supposed that the center of gravity of the robot moves like an inverted pendulum.
* The neural network like Hopfield's type solves the inverse kinematics so as to obtain the joint positions in the world coodinate from the positions of the supporting toe and of the center of gravity calculated from an equation of inverted pedulum.
* The method was examined by simulation studies. They showed satisfactory results in stationary walking and also the robustness for impulsive disturbances. The proposed method therefore provides a simplified version for the autonomous trajectory planning.
