CPG-based gait planning of a quadruped robot for crossing obstacles

Abstract

Ability of quadruped robots for adjusting posture for crossing natural obstacles is an important issue. These robots should change their footholds to avoid any obstacles. In this paper, a gait planning algorithm based on Cartesian CPG-based controller for a quadruped robot in order to cross different planar obstacles will be developed. In proposed approach, the rhythmic patterns in central pattern generator are modulated by feedback signals which encode foothold positions in respect to the planar obstacles. By using these feedback signals the frequency of oscillation will be adjusted so that desired gait avoids the obstacles. The obtained gait results in the swing and stance sequences and also step length for each foot. Then, the trajectory of the swing legs in Cartesian space will be computed based on the CPG output data. So, the desired ZMP (Zero moment Point) trajectory will be obtained by using the footholds. Then, COG trajectory will be obtained by using a preview controller on the basis of the desired ZMP trajectory. Finally, proposed algorithm will be tested for crossing planar obstacles in different size and distance. Obtained results show the effectiveness of the proposed approach in crossing obstacles with different size and distance to each other.