|author:||Jae S. Son|
|adviser:||Robert D. Howe|
The main contribution of this thesis is the advancement of robot hand dexterity in manipulating objects by using tactile sensing in robot hand control. Many issues in manipulation such as transient events, determining the contact location on a finger, grasping and grasp refinement, and object-to-object interactions are examined in detail. The first two chapters provide useful insights for selecting tactile sensors and algorithms. The following grasp refinement chapter combines the use of visual and tactile feedback for controlling a robot. Finally, Cartesian Object Stiffness Controller is used to study fundamental performance limits of a robot hand to modulate the stiffness of a grasped object. The consequences of these limitations and errors are illustrated by examining insertion forces for a high tolerance peg insertion task. The results of this thesis show that tactile sensing is able to increase robot hand dexterity by providing information about the hand object-relation and the mechanical state of the manipulation process.