阻抗控制

阻抗控制区别

Compared with other regular robotic control schemes, such as position control, force control, and
hybrid position/force control, impedance control has three distinguished features. First, the core idea of impedance control is to control the dynamic interaction between motion and contact force as desired instead of controlling these variables separately. Second, impedance control can be utilized in all manipulation phases consisting of free motion, constrained motion, and the transient process between them, without the need to switch different control modes. Last, impedance control provides a possibility to control motions and contact forces simultaneously by designing a proper interaction between a manipulator and its environment. As a result, the performance of robotic manipulation can be improved, and the safety of human–robot interaction can be guaranteed. With impedance control, the application domain of robotic manipulation can be significantly broadened.

与其他常规机器人控制方案（如位置控制，力控制和混合位置/力控制）相比，阻抗控制具有三个显着特征。首先，阻抗控制的核心思想是根据需要控制运动和接触力之间的动态相互作用，而不是分别控制这些变量。其次，阻抗控制可用于所有操纵阶段，包括自由运动，约束运动和它们之间的瞬态过程，而无需切换不同的控制模式。最后，阻抗控制提供了通过设计操纵器与其环境之间的适当交互来同时控制运动和接触力的可能性。结果，可以提高机器人操纵的性能，并且可以保证人机交互的安全性。通过阻抗控制，可以显着扩展机器人操作的应用领域。

Comparisons among different control schemes

In order to reveal the features of impedance control more explicitly, the comparisons among impedance control and other common control modes, such as position control, force control, and hybrid position/force control, are detailed in this subsection. Position control is the most frequently used control scheme in robotics, whose concern is to track
a motion trajectory as closely as possible. This scheme can work well in the free space. The typical applications include robotic welding, spray-painting, and other non-contact tasks. However, once an unexpected contact occurs, the position controller will treat this contact as a disturbance to be rejected, which may lead to position tracking errors and excessive contact force. This high resistance to the external contact reflects that the position controller actually possesses extremely high impedance. By contrast, the pure force control aims to keep the contact force as the reference without positioning requirements. To achieve the reference contact force, the robot in pure force control tends to move with the environment passively, which means the position of the manipulator cannot be guaranteed. This is because the pure force controller has very low impedance. In fact, both position control and force control can be viewed as two extreme situations of impedance control. In addition, the force control is effective only in the constrained space. If the robot has no contact with the environment initially, perhaps the position control mode has to be applied first to drive the robot to the contact state. Then the force control mode can be switched on successfully. However, this switch between control modes is likely to bring about unstable responses. When the moment to switch control mode mismatches the instant of the actual environment change, these unstable responses may be caused. For example, when the position-controlled manipulator has already touched the environment but the force control mode has still not been invoked, the excessive contact force may be produced. Similarly, when the force-controlled manipulator has just left the contact surface of environment but the position control mode has not been switched on, the manipulator may move disorderly at this moment. The excessive contact force and the unexpected motion that occur in the transient stage of the control mode switch are defined as the unstable responses. The other “unstable responses” in this paper refer to the same situations. The hybrid position/force control is one type of compliant control methods combining the position
control and force control together. In this method, the entire task space is divided into two subspaces, called the position-controlled subspace and the force-controlled subspace. During manipulation, the position control law and the force control law are employed in the corresponding subspaces, respectively. The most important step of this method is to determine a proper division of the subspaces, which relies on the prior knowledge of the structure and geometry of the environment. However, for those tasks performed in the unstructured and dynamically changing environments,7 it is difficult to obtain the precise geometry information about the environment in advance. This drawback limits its application range. In addition, the different control modes are switched according to the division of subspaces, which may also lead to some unstable responses. The detailed comparison results are summarized in Table I.7, 12, 50, 54, 132, 133 Due to their extreme impedance properties, the position control and the pure force control are only suitable for the specific work conditions. The hybrid position/force control cannot work well in the unstructured or dynamically changing environments. By contrast, impedance control performs better in terms of “flexibility” and “adaptability.” The “flexibility” means that the parameters of impedance model can be designed flexibly to change the compliance of robot as needed to meet the different demands of manipulation. The “adaptability” here has two implications. First, impedance control is adaptive to the different stages of one manipulation task. It means one unified impedance model can be employed in all manipulation phases, including free motion, constrained motion, and the transient stage between them, without the need to switch control modes. Second, impedance control is adaptive to many kinds of environments. Even in the unknown environment, a well-designed impedance model can work well. This character reflects the intrinsic robustness of impedance control.

为了更明确地揭示阻抗控制的特征，本小节详细介绍了阻抗控制与其他常用控制模式之间的比较，如位置控制，力控制和混合位置/力控制。位置控制是机器人中最常用的控制方案，其关注点是尽可能地跟踪运动轨迹。这种方案可以在自由空间中很好地工作。典型应用包括机器人焊接，喷涂和其他非接触任务。但是，一旦发生意外接触，位置控制器会将此接触视为要拒绝的干扰，这可能导致位置跟踪错误和过度接触力。这种对外部触点的高阻抗反映出位置控制器实际上具有极高的阻抗。相比之下，纯力控制旨在将接触力保持为参考而没有定位要求。为了获得参考接触力，机械手在纯力控制下会随着环境被动地移动，这意味着机械手的位置无法保证。这是因为纯力控制器具有非常低的阻抗。事实上，位置控制和力控制都可视为阻抗控制的两种极端情况。此外，力控制仅在受约束的空间中有效。如果机器人最初没有与环境接触，则可能首先应用位置控制模式以将机器人驱动到接触状态。然后可以成功打开力控制模式。但是，这种控制模式之间的切换可能会带来不稳定的响应。当切换控制模式的时刻与实际环境变化的瞬间不匹配时，可能导致这些不稳定的响应。例如，当位置控制的操纵器已经触及环境但仍未调用力控制模式时，可能产生过大的接触力。类似地，当力控制的操纵器刚刚离开环境的接触表面但是没有接通位置控制模式时，此时操纵器可能无序地移动。在控制模式开关的瞬态阶段中发生的过度接触力和意外运动被定义为不稳定响应。本文中的其他“不稳定反应”指的是相同的情况。混合位置/力控制是一种将位置控制和力控制结合在一起的顺应控制方法。在该方法中，整个任务空间被划分为两个子空间，称为位置控制子空间和力控制子空间。在操纵过程中，位置控制律和力控制律分别用于相应的子空间。该方法最重要的步骤是确定子空间的适当划分，其依赖于对环境的结构和几何形状的先验知识。但是，对于在非结构化和动态变化的环境中执行的那些任务，7很难提前获得有关环境的精确几何信息。这个缺点限制了它的应用范围。另外，根据子空间的划分切换不同的控制模式，这也可能导致一些不稳定的响应。详细的比较结果总结在表I.7,12,50,54,132,133中。由于它们的极端阻抗特性，位置控制和纯力控制仅适用于特定的工作条件。混合位置/力控制在非结构化或动态变化的环境中不能很好地工作。相比之下，阻抗控制在“灵活性”和“适应性”方面表现更好。“灵活性”意味着可以灵活地设计阻抗模型的参数以根据需要改变机器人的顺应性以满足不同的操纵要求。这里的“适应性”有两个含义。首先，阻抗控制适应于一个操作任务的不同阶段。这意味着可以在所有操纵阶段采用一个统一的阻抗模型，包括自由运动，约束运动以及它们之间的瞬态阶段，而无需切换控制模式。其次，阻抗控制适用于多种环境。即使在未知环境中，精心设计的阻抗模型也能很好地工作。该特性反映了阻抗控制的固有鲁棒性。

三个模型

position-based method.

Torque-based method

model-based method