仿人机器人的可变ZMP步态规划

摘要

为使仿人机器人能够稳定地行走,基于三维倒立摆模型,采用了可变ZMP(Zero-Moment Point)来进行仿人机器人的步态规划.在单脚支撑期依据可变ZMP采规划COG(Center of Gravity)的轨迹,在双脚支撑期内则让ZMP保持低速匀速运动,由此来规划此时间段的COG轨迹.仿真结果验证了规划的合理性,保证了更强的稳定性,并且使得采用此规划的仿人机器人的行走和人类的行走更加相似.%Based on the 3D Inverted Pendulum model,a new walking pattern synthesis was proposed,which can make a humanoid robot walk smoothly by variable ZMP(Zero-Moment Point). By the new way,in the single leg support phase, the COG ( Center of Gravity) trajectory equation of the humanoid robot can be educed through the variable ZMP;while in the double leg support phase,the COG trajectory can be educed by the ZMP moving at a lower constant velocity. Simulation results verifies the proposed method,which can guarantee a relatively high stability. And the humanoid robot in this walking pattern can walk more smoothly, and in a more similar way to human walking.