Teleoperation of a quadruped walking robot using an aperiodic gait that converges to a periodic gait

Abstract

Presents a gait control algorithm for teleoperation of a quadruped walking robot using gait convergence. A leg trajectory of the walking robot is generated in real-time according to the 3 DOF planar moving velocity command. Initially, the aperiodic gait is applied to follow the unpredictable command, since the periodic gait is known to be optimal and it requires an initial foot position. However, given an arbitrary initial foot position, the proposed algorithm leads the support pattern of the legs to converge from the aperiodic gait to a wave-crab gait or a wave-spinning gait. Further it avoids a deadlock after the directional change of the velocity command. The proposed algorithm has been applied to the walking robot, CENTAUR of KIST that has heavy legs and a large moving area of the gravity center. To compensate the moving gravity center, a sway motion is proposed to transfer the leg of which the kinematic margin is almost zero and the leg trajectory for the sway motion is overlapped with the operator command.