TABLE OF CONTENTS Marek Teichmann, Courant Institute, New York University, NY, USA
Bud Mishra, Courant Institute, New York University, NY, USA
This paper attempts to quantify the trade-offs that exist among the number of fingers, coefficient of friction and the goodness of a grasp. In particular, we give a general framework for defining a grasp metric that takes friction into account. We contrast our approach to an older metric that was proposed by Ferrari and Canny [FC92] for the frictional case. In addition, we discuss several algorithmic solutions related to frictional grasps.
A standard approach to grasp theory takes the point of view that it suffices to consider the situation where the fingers make only frictionless point contacts with the object to be grasped. Under such a model, referred to as a positive grip model, one then considers various grasp existence and synthesis problems. (See Mishra, Schwartz and Sharir [MSS87] for several key results under this model. For some related results, also see [MNP90, Ngu87, PSS+95].) This point of view is justified by the argument that presence of friction only improves the grasp and hence positive grips represent in some sense the most pathological situation. Another argument favoring the positive grip model is that even when there is uncertainty about how much friction is available, the grasps synthesized under this restricted model remain immune to such uncertainties. However, there have been relatively inadequate attempts to answer many questions that the above viewpoint raises.
For instance, given a...
