A modified gait planning method for biped robot based on central pattern generators

The gait planning is essential in the sampling-based footstep planning of biped robot. However, conventional gait planning methods of Zero Moment Point and neural networks are much time-consuming, and existing central pattern generator (CPG) based method can only generate the gait of going forward. Based on the phase and trajectory analysis, this paper proposes a modified gait planning method based on CPG for the sampling-based footstep planning. By adjusting the parameter of the CPG, it is available to obtain different gaits of going forward, stepping side and swerving, and realize smooth transition between these gaits. Physical experiments on NAO robot verified the effectiveness of the proposed method.