This paper presents a global strategy for object manipulation with the fingertips with an anthropomorphic dexterous hand: the LMS Hand of the ROBIOSS team from PPRIME Institute in Poitiers (France). Fine manipulation with the fingertips requires to compute on one hand, finger motions able to produce the desired object motion and on the other hand, it is necessary to ensure object stability with a real time scheme for the fingertip force computation. In the literature, lot of works propose to solve the stability problem, but most of these works are grasp oriented; it means that the use of the proposed methods are not easy to implement for online computation while the grasped object is moving inside the hand. Also simple real time schemes and experimental results with full-actuated mechanical hands using three fingers were not proposed or are extremely rare. Thus we wish to propose in a same strategy, a robust and simple way to solve the fingertip path planning and the fingertip force computation. First, finger path planning is based on a geometric approach, and on a contact modelling between the grasped object and the finger. And as force sensing is required for force control, a new original approach based on neural networks and on the use of tendon-driven joints is also used to evaluate the normal force acting on the finger distal phalanx. And an efficient algorithm that computes fingertip forces involved is presented in the case of three dimensional object grasps. Based on previous works, those forces are computed by using a robust optimization scheme. In order to validate this strategy, different grasps and different manipulation tasks are presented and detailed with a simulation software, SMAR, developed by the PPRIME Institute. And finally experimental results with the real hand illustrate the efficiency of the whole approach.

Reach and grasp are the two key functions of human prehension. The Central Nervous System controls these two functions in a separate but interdependent way. The choice between different solutions to reach and grasp an object-provided by multiple and redundant degrees of freedom (dof)-depends both on the properties and on the use (affordance) of the object to be manipulated. This same control paradigm, i.e. subdivision of prehension into reach and grasp as well as the corresponding multimodal (sensory/motor) information fusion schemes, can also be applied to a mechanical hand carried by a robotic arm. The robotic arm will then be responsible for positioning the hand with respect to the object, and the hand will then grasp and manipulate the object. In this article, we present a biomimetic sensory-motor control scheme in the aim of providing an object-dependent and intelligent reach and grasp ability to such systems. The proposed model is based on a multi-network architecture which incorporates multiple Matching Units trained by a statistical learning algorithm (LWPR). Matching Units perform a multimodal signal integration by correlating sensory and motor information analogous to that observed in cerebral neuronal networks. The simulated network of multiple Matching Units provided estimations of object-dependent 5-finger grasp configurations with endpoint positional errors in the order of a few millimeters. For validation, these estimations were then applied to the control of movement kinematics on an experimental robot composed of a 6 dof robot arm carrying a 16 dof mechanical 4-finger hand. Precision of the kinematics control was such that successful reach, grasp and lift was obtained in all the tests.

This paper concerns object manipulation with robotic dexterous hands using the fingertips. Two important aspects in the manipulation scheme are developed; the first one concerns grasp synthesis; it means that the choice of the initial grasp is done before the manipulation begins; and the second one concerns fingertips motion planning. Thus the LMS Hand of the ROBIOSS team from PPRIME Institute of Poitiers (France) is used as a demonstrator. The proposed strategy is first based on the use of an expert system for solving the grasp synthesis problem. As we wish to manipulate with the fingertips, we do not consider power grasp involving whole parts of fingers; so we propose to synthesize precision grasp. Human grasp analysis is used to teach the expert system. Thus the experimental protocol using motion capture is detailed and results are discussed. To complete this strategy a geometrical approach for motion planning based on a contact modeling between the grasped object and the fingertip is given. It offers a robust solution for solving the fingertips motion aiming for object manipulation. In order to validate the whole strategy, different manipulation tasks are presented and detailed with a three-dimensional CAD software dedicated to robotic simulation : SMAR developed by the PPRIME Institute.

This paper presents a new method for solving the grasp optimization problem by a multi-finger robotic hand; this method allows gripping an object using three articulated fingers, in order to manipulate it later. Because of the large number of operations and the high computation time, online grasp has not yet been reported. In this study, we propose a method that is able to provide an optimized initial grasp in a short time before online manipulation.

When handling an object by a mechanical hand, and w hen a joint limit or a collision occurs, it becomes necessary to reposition the fingers on the object b efore continuing the task at hand. We therefore pro pose a method that solves this problem of reconfiguration and allows the realization of given amplitude movem ents without any pose / recovery of the seized object.