Adaptive Gait for Dynamic Rotational Walking Motion on Unknown Non-Planar Terrain by Limb Mechanism Robot ASTERISK

Abstract

An adaptive gait is proposed for dynamic rotational walking motion of multi-legged mobile robots by utilizing body angle compensation and the center of mass height control. Posture control is used to further enhance the robustness and stability of the robot based on a posture optimization database. The database is created by using a genetic algorithms in order to find the most suitable posture for each virtual plane created during body compensation in adaptive gait control. Additionally, the stability of the robot is controlled by online zero-moment point compensation and compliance control. In order to test the robustness of motion, experiments are divided into three parts: inclined-plane (linear variation), step (step variation), and obstacle (impulse variation). As a result of this research, with the proposed method, the robot could walk up and down inclined-planes with angles of 6.8° and 5.6°, respectively, and walk up and down a step and over an obstacle with a height of 20 mm.