Robot Hands

Human hands have great potentiality not only for grasping objects of various shapes and dimensions, but also for manipulating them in a dexterous manner. It is common experience that, by training, one can perform acrobatic manipulation of stick-shaped objects, manipulate a pencil by using rolling or sliding motions, perform precise operations requiring fine control of small tools or objects. It is obvious that this kind of dexterity cannot be achieved by a simple gripper capable of open/close motion only. A multifingered robot hand can therefore provide a great opportunity for achieving such a dexterous manipulation in a robotic system. Moreover, we have also to consider that human beings do not use hands only for grasping or manipulating objects. Exploration, touch, perception of physical properties (roughness, temperature, weight, just to mention a few) are other fundamental tasks that we usually are able to perform thanks to our hands. We expect this type of capabilities also from robotic end-effectors and therefore, by adding quite advanced sensing equipments and proper control strategies, we may improve the interaction capabilities with the environment, achieving for example active exploration, detection of sensing surface properties (local friction, impedance, and so on), tasks that are usually very hard or impossible for simple grippers. For these and other reasons the study of multifingered robot hands has stronlgy interested the research community since the early days of Robotics. It was in late '70s that Okada developed a multifingered robot hand with a tendon driving system and demonstrated a nut opening motion cite{Ch16-Okada}. In early '80s, two major projects on multifingered robot hands have been launched: the Stanford/JPL hand and the MIT/Utah hand, cite{Ch16-Salisbury83,Ch16-Jacobsen86}. These two robot hands still represent a milestone and a term of comparison for the design of new devices. Since then, several multifingered hands have been designed and developed in a number research institutes all over the world. Among the most known, one can mention the DLR-Hand, MEL-Hand, ETL-hand, Darmstadt-Hand, Karlsruhe-Hand, Bologna-Hand, Barrett-Hand, Yasukawa-Hand, Gifu-Hand, U-Tokyo-Hand, Hiroshima-Hand, and many others, cite{Ch16-Butterfass99,Ch16-Melchiorri,Ch16-Townsend00,Ch16-Kawasaki,Ch16-Doll}. When designing a multifingered hand, on the basis of its utilization, one should first define the following key issues: number and kinematic configuration of the fingers, anthropomorphic or non-anthropomorphic aspect, built-in or remote actuation, transmission system (in case of remote actuation), sensor assignment, integration with a carrying device (robot arm), control. All these aspects are considered in the following Sections of this Chapter.