关键词: 双足机器人, 步态控制, 支持向量机（SVM）, 人类控制策略

Abstract: To deal with the dynamically walking control problem in the complex environments for biped robot, a method of gait control based on learning human control strategy is put forward. State equation of biped walking system based on 3D linear inverted pendulum is constructed. The parametric model of learning HCS is built. A learning controller based on SVM is designed. The proposed method guarantees that torso is always perpendicular to the ground. Robustness of the gait control is enhanced. The dynamically stability of biped walking is improved in complex environments. Experimental results demonstrate the effectiveness of the proposed methods.

Key words: biped robot, gait control, support vector machine, human control strategy