Abstract Robot arms exhibit non-ignorable residual vibration during point-to-point motion. Fractional Delay Time-Varying Input Shaping Technology (FD-TVIST) has previously been shown to reduce residual vibrations in robots arms, but requires an accurate estimate of the configuration dependent vibrational behavior. This paper presents a practical set of new tools for identifying and mapping the natural frequencies and damping ratios of dominant vibration modes of a robot arm with the commercial UR5e as a case. The new dynamic map is based on a physical interpretation of the robot arm, followed by strategic simplification. The new map of the vibration modes can be described by only 10 identified coefficients. Additionally, FD-TVIST is extended to the multi-mode case. The combination of new methods yields an efficient vibration suppression featuring short time delay, great robustness, simple implementation, low-cost computation, and insensitivity to trajectory generation method. Experimental implementation and validation in a commercial UR robot show that the residual vibration of the robot manipulator is reduced up to 90% with the developed vibration control method.

中文翻译：

---

基于多模时变输入整形的工业机械臂振动控制

摘要 机械臂在点对点运动过程中表现出不可忽视的残余振动。分数延迟时变输入整形技术 (FD-TVIST) 先前已被证明可以减少机器人手臂中的残余振动，但需要准确估计依赖于配置的振动行为。本文以商用 UR5e 为例，介绍了一套实用的新工具，用于识别和映射机器人手臂的主要振动模式的固有频率和阻尼比。新的动态地图基于对机械臂的物理解释，然后是战略简化。新的振动模式图只能用 10 个识别的系数来描述。此外，FD-TVIST 扩展到多模情况。新方法的结合产生了一种有效的振动抑制，具有时间延迟短、鲁棒性强、实现简单、计算成本低、对轨迹生成方法不敏感等特点。在商用 UR 机器人中的实验实施和验证表明，使用开发的振动控制方法，机器人机械手的残余振动降低了 90%。