The kinematics of multi-fingered manipulation

Abstract

Previously, general models of the kinematics of multi-fingered manipulation have only treated instantaneous motion (i.e., velocities). However, such models, which ignore the underlying configuration (or state) space, are inherently incapable of capturing certain properties of the fingers-plus-object system important to manipulation. In this paper, we derive a configuration-space description of the kinematics of the fingers-plus-object system. To do this, we first formulate contact kinematics as a "virtual" kinematic chain. Then, the system can be viewed as one large closed kinematic chain composed of smaller chains, one for each finger and one for each contact point. We examine the underlying configuration space and two ways of moving through this space. The first, kinematics-based velocity control, is a generalization of some previous velocity-based approaches. The second, hyperspace jumps, is a purely configuration-space concept. We conclude with a discussion of how these concepts can be used to understand the task of twirling a baton [6].

Multi-Fingered Grasping and Manipulation in Virtual Environments Using an Isometric Finger Device

“…The force applied at the contact point is defined by the contact wrench F Ci ʦ ℜ 6 consisting of the force and torque vectors of the corresponding fingertip. Two contact types were implemented in our model: point contact with friction and soft finger contact (Montana, 1995). The type of contact is described by the corresponding wrench basis B Ci ʦ ℜ 6ϫp which defines the number of degrees of freedom p in which the object is fully constrained by the finger.…”

“…Realistic grasping in a VE is achieved through accurate modeling of forces and torques resulting from the fingertips in contact with the surface of the virtual object. The model of multi-fingered grasping can be adopted from the analysis of multi-fingered robotic hands (Murray, Li, & Sastry, 1994;Montana, 1995) to appropriately describe the effects of whole-hand manipulation. Accurate dynamics modeling of multifingered grasping can greatly increase the realism of the interaction within the VE.…”

Regrasp planning in the grasp space using independent regions

“…The regrasping approach (or multi-fingered manipulation) solves the manipulation problem by using all the available fingers on the hand; the positions of the fingers can only be changed by rolling or sliding them along the surface of the object [2]. The finger gaiting and regrasping movements can also be combined to plan a desired manipulation of the object; for instance, the manipulation problem may be represented as a switching graph, where each node represents a grasp, and the finger gaiting planning is reduced to a graph search problem [3]- [5].…”

Dexterous manipulation planning using probabilistic roadmaps in continuous grasp subspaces

“…The early works on DMP concerned the problem formulation without introducing any resolution scheme ( [1], [2]). The first method to be presented was the one by Trinkle and Hunter [3].…”