The lack of suitable commercial tactile sensors has limited developments in the robotic handling of fragile objects. An algorithm that can discriminate between types of contact surface and recognize objects at the contact stage is also proposed. A technique for recognizing objects using tactile sensor arrays, and a method based on the quadric surface parameter for classifying grasped objects is described. Tactile arrays can recognize surface types on contact, making it possible for a tactile system to recognize translation, rotation, and scaling of an object independently. In addition, the studies showing continuous adjustment of force to stabilize gripping, particularly during motion, have been reviewed. This prehension force adjustment occurs simultaneously with, or slightly ahead of, fluctuations in load forces. They may therefore be seen as anticipatory, and it is argued here that a key purpose of research in manipulation should be to understand the integration of sensory motor information in building an internal model of the object and the effector system in order to support such anticipation.