Contents 1. Introduction
Biped robot is a complex dynamic system, which is non-linear, multi-variable, strongly coupled. The problem of controlling the walking motion of a biped robot has intrigued and challenged researchers for many years [1] [2] [3] [4] [5] [6]. It is very difficult to establish an accurate mathematical model of kinematics and dynamics for it. So far, the mainly mathematical model of biped robot be established by researchers were decoupling model and reduced-order model, reduced-order model mainly referring to the inverted pendulum model [7] [8]. The holistic structure of biped robot walking model similar to people's lower limbs. Each leg equipped with the joint of hip, knee and ankle. These joints are indispensable for steady and effective walking. Hip is to swing legs to move on and make the body leaning forward or layback position. Knee mainly used to adjust the height of centre of mass and adjust the swing leg's height of touchdown, make it compatible with landform. And ankle used to cooperate with hip to realize the support leg and upper limbs' motion, moreover it can adjust the tangent state between sole and ground. In a certain sense, its structure is similar to the structure of the inverted pendulum [9]. In the structure, humanoid robot can use two double inverted pendulum which are associated with a two-axis to simulate. The biped robot's walking mode can divided into static walking and dynamic walking [10] [11] [12]. Static walking is the mode of less center of gravity move and slow-footed. Dynamic walking is the mode of self destruction balance and forward tipping, human's walk by dynamic walk primarily. The move of inverted pendulum is belong to typically dynamic walking, the pole of the pendulum leaning forward with inertia when the cart back up, the cart move forward to maintain balance to achieve the purpose of moving forward. To the robots, may make a forward slant of the upper body through the motor in hip joint. The concrete value of upper body and legs movement, using the simulation data of pendulum to gain the relations between them.
