A distributed testbed designed to support the development of a multi-sensory (vision and tactile) system for investigations in "active perception" of three dimensional objects is presented. Active perception means being able to not only see and feel objects but also manipulate and probe them. The nucleus of the testbed is a network of heterogeneous computers designed to support low-level real-time control processes as well as high-level knowledge-based systems. The programming environment of the testbed facilitates the construction and execution of a distributed multi-sensory system from sequential programs written in different programming languages.

The Integrated Tactile Network Architecture or ITNA is a hierarchical system for managing the interaction of tactile sensing and motor control in the 3-D active sensory environment. The overall ITNA includes custom dedicated hybrid front end tactile arrays incorporating electronics and microprocessors for sensor linearization, tactile information preprocessing, and local feature extraction approaches to the distributed motor control of manipulator fingers for grasping; interconnection networks for guarded movement and reflex arcs; and special purpose hardware for model generation derived from tactile information. This paper primarily addresses the overall ITNA structure and, in particular, the design of an intelligent sensor array and its associated communications subsystem.

The two-channel system previously reported has been implemented in hardware using system parameters appropriate to consumer television. The basic system divides the spectrum into lowand high-frequency spatial components. The lows are coarsely sampled and finely quantized and the highs finely sampled and coarsely quantized using a companded, randomized quantizer. The purpose of this experiment was to demonstrate the potential of the system for low-cost practical application, to study the effect of the character of the randomizing noise, and to ascertain that there were no deleterious effects due to interlace, motion, or input noise. Theoretical noise calculations were qualitatively confirmed.

A computer-based system for automated engraving of gravure cylinders has been developed and is now used in a normal production environment. Either fully composed pages or individual page components are scanned and stored on a large disk. In the case of fully composed pages, an operator uses a TV display to segment the page image into line and tone areas. The image is then coded by a software process and is ready for subsequent engraving. Prior to the scanning of page components, the operator uses a tablet in order to demark both cropping locations and to specify the location of the components on the final page image. The scanned components are then assembled and coded by a single software process. The encoding process reduces the data storage requirement by a factor of two without any apparent loss of quality. Data are retrieved from disk storage, buffered, decoded, transmitted to a special formatter, and used to drive a Helio Klischograph,® which engraves the cylinder in approximately one hour. Completely arbitrary imposition (the arrangement of pages on the gravure cylinder) is accomplished at the time of engraving. Provision is made for the arbitrary intermixture of computer-processed pages and conventional engraving by means of Cronapaquesgi mounted on the companion scanning machine. The motivation for this development was to reduce the cost and time required for the production of gravure cylinders.