Modeling manufacturing grips and correlations with the design of robotic hands

Abstract

This paper represents the first part of an effort to codify the knowledge required for manipulation tasks in a small-batch manufacturing cell. The motivation for this work is to pave the way for robots that can independently determine how to grasp and manipulate parts in a limited environment and to facilitate the design of advanced, but cost-effective manufacturing hands. We begin with an examination of grasps used by humans working with tools and metal parts. The grips are compared in terms of power, contact area, friction, damping and tactile sensitivity. The comparison leads to a grip taxonomy in which grasps are mapped against task-related quantities (such as power) and object-related quantities (such as slenderness). The examinations of the task requirements and grasps suggest a number of general principles for the design and control of manufacturing hands.