基于虚拟模型控制的四足机器人缓冲策略  

A Buffering Strategy for Quadrupedal Robots Based on Virtual Model Control

作　　者：刘斌[1] 荣学文[1] 柴汇[1] 

机构地区：[1]山东大学控制科学与工程学院,山东济南250061

出　　处：《机器人》2016年第6期659-669,共11页Robot

基　　金：国家863计划(2015AA42201);国家自然科学基金(61233014;61305130);山东省自然科学基金(ZR2013FQ003;ZR2013EEM027);中国博士后科学基金(2013M541912)

摘　　要：提出了一种基于虚拟模型控制的四足机器人缓冲策略.根据机器人落地过程中的触地状态和躯干纵向速度,将机器人的落地过程分为下落阶段、缓冲阶段和恢复阶段.在下落阶段,通过虚拟"弹簧-阻尼"元件驱动足端沿期望轨迹运动.在缓冲阶段和恢复阶段,控制器根据机器人下落过程中落地腿数目的不同,分别建立支撑腿与躯干质心虚拟力之间的数学关系,并通过施加在躯干质心的虚拟力的大小来控制躯干的位姿.躯干质心所施加的虚拟力通过主动变刚度控制实现,控制器根据机器人所处的落地阶段,给出合理的虚拟刚度和阻尼,从而减小机器人落地的冲击.由仿真实验可以看出,此缓冲策略是有效的.A buffering strategy for quadrupedal robots based on virtual model control is presented. According to the contact status of legs and the vertical velocity of the torso, the whole landing process is divided into three phases, that is, the falling phase, the buffering phase and the recovering phase. During falling phase, virtual spring-damper sections are implemented for flight toes to track the planned trajectories. For the buffering phase and the recovering phase, the mathematical relation between the virtual forces applied on the COM （center of mass） of the torso and the supporting legs is established according to the number of the supporting legs. And the torso＇s COM is controlled by the virtual forces through active variable stiffness control. To reduce the impact in landing process, reasonable stiffness and damping are determined according to which phase the robot is in. The simulation shows that the buffering strategy is effective.

关 键 词：四足机器人 虚拟模型控制 力控制 主动变刚度控制 

分 类 号：TP242[自动化与计算机技术—检测技术与自动化装置]