In teleoperation, a force feedback device is a medium to build a transparent interaction environment between a human and a remote robotic arm. Using force feedback devices, the users can operate the remote robotic arm intuitively and perceive remote interaction through the force channel, just as if they are in the remote environment. Compared with impedance devices, admittance devices have the advantages of large feedback output, high stiffness, high reverse driving performance, and flexible structure, which are more suitable for the teleoperation of heavy-duty and large-size robotic arms. However, the control of admittance devices is relatively complex and has some inherent limitations such as response delay, instability from high-frequency oscillation, difficulty in achieving constant speed control, etc. Errors in admittance model parameters and human physiological characteristics, such as force application fluctuations, are the root causes of these problems. In this study, we proposed a fuzzy variable damping admittance algorithm, which allows the device to identify the user's movement intention and give respond quickly and accurately. We also established a human-robot interaction (HRI) system model of an admittance master controller device and summarized the principles of the admittance parameter configuration of a stable system. For the device's high-frequency oscillation instability caused by human arm stiffness, we propose an oscillation observation and reduction algorithm. By observing the force signal change characteristics, the algorithm can quickly detect the unstable behavior caused by human hands and perform oscillation reduction. To reduce the influence on upper limb uniform motion caused by fluctuating force application, we proposed a constant velocity intention inference algorithm based on a velocity spherical cone to smooth out the device operating velocity to achieve smooth control. The method proposed in this study achieved stable control in a 6 DOF force feedback device as a master controller, and the effect has been verified by experiments.

在远程操作中，力反馈设备是在人与远程机械臂之间构建透明交互环境的媒介。通过力反馈装置，用户可以直观地操作远程机械臂，通过力通道感知远程交互，犹如置身于远程环境中。与阻抗器件相比，导纳器件具有反馈输出大、刚度高、反向驱动性能高、结构灵活等优点，更适用于重型、大型机械臂的遥控操作。然而，导纳装置的控制相对复杂，并且存在一些固有的局限性，如响应延迟、高频振荡不稳定、难以实现恒速控制等。导纳模型参数和人体生理特性的误差，如施力波动，是造成这些问题的根本原因。在这项研究中，我们提出了一种模糊可变阻尼导纳算法，使设备能够识别用户的运动意图并快速准确地做出响应。我们还建立了导纳主控装置的人机交互（HRI）系统模型，总结了稳定系统的导纳参数配置原则。针对人体手臂僵硬引起的设备高频振荡不稳定性，我们提出了一种振荡观察和减少算法。通过观察力信号变化特性，该算法可以快速检测到人手引起的不稳定行为并进行振荡减少。为降低脉动力施加对上肢匀速运动的影响，提出了一种基于速度球锥的等速意图推理算法，平滑设备运行速度，实现平滑控制。本研究提出的方法在作为主控制器的6自由度力反馈装置中实现了稳定控制，并通过实验验证了效果。