Study of similar motion imitation of stepping upstairs for humanoid robot

Aimed at the situation of fitting variable environments in similarity imitation of stepping upstairs for humanoid robot, the process of designing similarity imitation was analysed as well as the rhythm control of key posture in similar locomotion. The simplified locomotion model and retargeting, connection of motion phases, and ZMP stability constraints were shown in similarity locomotion. The hierarchical reinforcement learning MaxQ was used to adjust the movement tracks of similarity to fit the variable stairs. The experiments showed the validity of the method.