Reorienting Objects with a Robot Hand Using Grasp Gaits
Reorienting Objects with a Robot Hand Using Grasp Gaits
复制标题
使用抓取步态用机器人手重新定向物体
DOI：
--
复制
发表时间：
1996
期刊：
影响因子：
--
通讯作者：
K. Salisbury
中科院分区：
文献类型：
--
作者：
Susanna Leveroni;K. Salisbury
本文讨论了规划的机器人手指运动，使连续和稳定的重定向把握对象。在物体上给定抓握的情况下，手指可以移动物体的重新定向的范围受到手指的工作空间和抓握稳定性的限制。然而，如果对象可以被适当地重新抓握而不使其掉落，则对象在期望方向上的进一步运动是可能的。如果我们使用的手的手指数超过了稳定抓握所需的最小数量，那么就有可能从抓握中取出一个（或多个）手指而不使其掉落，然后可以将这个手指（或这些手指）重新放置在表面上，形成一个新的抓握，从而进一步增加重新定向。我们称手指/物体运动和重握的结果序列为抓握步态。为了开发一种方法来规划这样的步态，我们已经解决了这个问题的重新定向对象被约束躺在平面上。由于在这种环境中仅用两个手指就可以实现稳定的抓握，因此具有三个手指的手可以能够执行一系列重新抓握和物体运动，使得连续的重新定向成为可能。我们的研究结果表明，对于某些类型的对象，简单的刻板步态可能会发现，实现这一目标。对于更严格的情况，例如接触处的低摩擦或形状奇怪的物体，我们已经开发了使我们能够找到更复杂的步态模式（如果存在）的方法。
This paper addresses the planning of robot finger motions to enable continuous and stable reorientation of grasped objects. With a given grasp on an object, the range of reorientation through which the fingers may move the object is limited by the workspace of the fingers and grasp stability. If however the object can be properly regrasped without dropping it, further motion of the object in the desired direction may be possible. If we use a hand with more than the minimum number of fingers required for a stable grasp, then it is possible to remove one (or more) of the fingers from the grasp without dropping it. This finger (or these fingers) may be then be replaced on the surface to form a new grasp with which a further increment of reorientation may be achieved. We call the resulting sequences of finger/object motions and regrasps grasp gaits. To develop an approach to planning such gaits we have addressed the problem of reorienting objects which are constrained to lie in the plane. Since a stable grasp in this environment can be achieved with only two fingers, a hand with three fingers may be able to execute a series of regrasps and object motions such that continuous reorientation is possible. Our results show that for certain types of objects, simple stereotypic gaits may be found which achieve this goal. For more restrictive cases, such as low friction at the contacts or oddly shaped objects, we have developed methods which enable us to find more complex gait patterns if they exist.