Stability Analysis And Walking Pattern Design Of The Humanoid Robot

The stability problem is the prerequsite of the motion planning. The zero meoment point(ZMP) position is the criterion for the dynamic stability. Motion planning is one of the key techniques for the research of humanoid robot. stable and natural walk is the main sign differentiating humanoid robot from other robots, and it is the base of robots to realize its essential manipulation.Firstly, the popular methods of walking pattern about the humanoid robot are stated. It can be give the reference for the further research.Secondly, the balance condition of the single-support pahse is analyzed. The part of the mechanism above the ankle of the support foot is negelected and replace its influence by the force and the moment during this analyzation.and then, the ZMP equations are derived from the D'Alembert's pricinple.In addition, the sagittal and the lateral decoupled dynamic model of nine-link mechanism are derived based on the dynamic model of the seven-link mechanism, because its high order, nonlinear and strong coupling, valid mathematical model's construction is very difficult. The moment of the joints can be gotten because of the explicit character of the decoupled dynamic model.Then, the emphasis is put on algorithm design and simulation for walking pattern planning.The static walking pattern is based on the position of CoG whose key problem is how to avoid the shock of the steer-motor. The dynamic motin planning method based on ZMP(Zero Memont Point) is been discussed in detail. According to this way, the constriants of foot during walking are gotten firstly and are taken them as constrain points, then three order spline interpolation funciton through constrain points is used to get the smooth trajectory . ZMP stable margin is token as evaluation function for searching the stable pattern in the feasible scope, optimal hip trajectory has been gotten. Finally, the joint angles have been calculated by kinematics. Simulations have been done according to this two algorihtm. The simulation results testify the feasibility and validation for the methods.