论文部分内容阅读
提出一种基于虚拟模型控制和加速度规划的腿部缓冲方法。通过建立机器人腿部的虚拟模型,设定落地过程中躯干加速度从而可减小地面对机器人的冲击力,保护机器人的机械结构。该方法将机器人的落地过程分为下落、缓冲、恢复3个阶段。在下落阶段,通过在足端与期望位置之间添加虚拟“弹簧—阻尼”系统来控制足端位置。在缓冲阶段和恢复阶段,通过规划躯干质心加速度,从而减小落地过程中躯干受到的冲击。该方法可避免在激烈的足地交互过程中调节系统的刚度和阻尼,控制过程更简单精确。基于Webots的仿真试验表明,该方法在机器人落地过程中的保护是有效的。
A method of leg cushioning based on virtual model control and acceleration planning is proposed. By establishing the virtual model of the robot leg, the torso acceleration during landing can be set so as to reduce the impact of the ground on the robot and protect the mechanical structure of the robot. The method divides the landing process of the robot into three stages: falling, buffering and restoring. During the fall phase, the foot end position is controlled by adding a virtual “spring-damped” system between the foot end and the desired position. During the cushioning phase and the recovery phase, the torso center of gravity is minimized by planning the acceleration of the center of mass of the torso. The method can avoid the stiffness and damping of the system during intense foot-to-hand interaction, and the control process is more simple and accurate. Simulation experiments based on Webots show that this method is effective in robot landing.