Walking pattern design and feedback control for humanoid robot

This paper proposes a smooth cooperated walking pattern design and feedback control based on tilt senor for biped humanoid robot. In this research, the smoothness and stability problem is the key work of the motion planning. We make the robot walk stably and naturally like a human using static Zero Moment Point (ZMP) theory and the inclination feedback control. Its basic idea is as follows: firstly, through the geometry and static ZMP analysis, we get separate smooth rotation trajectories of all joints of robot. Then a smooth cooperation action of all joints of robot will be calculated by interpolate breakpoints which are selected in each rotation trajectory. Lastly, using tilt feedback control, we can get a smooth and stable walking of humanoid robot. To make robot walk corporately, we use three methods of interpolation to get all trajectories of 18 joints. The design of walking contains single-support phase planning and double-support phase planning. The effectiveness of our proposed method was confirmed by Matlab simulation and walk experiments of an actual 18 DOF humanoid robot.