Elliptic trajectory generation for galloping quadruped robots

Abstract

Despite the fact that a running legged robot can be simply modeled by a spring loaded inverted pendulum and that the trajectory of its center of gravity is approximated as a ball on the rebound, the motions of its swinging leg and supporting leg are very important to run stably. In many previous works, trajectories of legs could easily deviate from their workspace of legs and thus make the robots unstable. In this paper, for stability in running, we propose a method to generate trajectories of legs based on an ellipse and to modify joint angles and angular velocities of the legs during touchdowns and take-offs depending on the running speed. To evaluate the performance and the effectiveness of this method, a series of computer simulation of a galloping 4-legged robot in the sagittal plane are performed. Their results show that the proposed method is effective in generating stable galloping locomotion.