Computer Vision is essentially concerned with emulating the process of seeing, naturally manifested in various higher biological systems,1–4 on a computational apparatus, and is consequently part of the Artificial Intelligence field within the sub-category of Machine perception. Seeing has to do with making sense of image data acquired through an optical system and subsequently dealt with at increasing levels of abstraction and association with known facts about the world. The spectrum of interest in Computer Vision ranges from attempting to answer basic questions concerning the functionality of biological vision systems, particularly human, at one end, all the way to enhancing the reliability, speed and cost effectiveness of specific industrial operations, particularly component inspection and vision driven robotic manipulation. The main bulk of interest is in the middle, where the quest for generality pushes interest towards biological vision systems with their demonstrated effectiveness in a wide range of environments, some hostile, whilst the need for economic viability and timeliness in relation to particular application pushes interest towards finding workable algorithms which function reliably at high speed on affordable apparatus.

This paper is addressed, in somewhat tutorial style, at clarifying, by examples of work in the area, the issues surrounding application oriented robotic vision systems, their assumptions, strengths, weaknesses and degree of generality, and at the same time putting them in the context of the overall field of Computer Vision. In addition, the paper points to directions of development which promise to provide powerful industrial vision tools at an acceptable price.

A hierarchical robot sensory system being developed for industrial robotics is described. At each level of the hierarchy, sensory interpretative processes are guided by expectancy-generating modeling processes. The modeling processes are driven by a priori knowledge (object prototypes), by knowledge of the robot's movements (feed-forward from the control system), and by feedback from the interpretative processes (prior state of the sensory world). At the lowest level, the senses (vision, proximity, tactile, force, joint angle, etc.) are handled separately; above this level, they are integrated into a multi-model world model. At successively higher levels, the interpretative and modeling processes describe the world with successively higher order constructs, and over successively longer time periods. Each level of the modeling hierarchy provides output, in parallel, to guide the corresponding levels of a hierarchical robot control system.

With the recent advances of VLSI technology, it has become possible to realize parallel architectures for image processing which are more efficient than the customary sequential processors. The paper discusses general architectural requirements for parallel processing and then the existing systems architectures. This is followed by a description of VLSI devices for image processing in three examples. The paper emphasizes VLSI architectures for image feature extraction, classification and structure (syntax) analysis.

Machine vision systems incorporating highly parallel processor architectures are reviewed. A new processor architecture, the image flow computer, is presented in detail. An interactive image processing programming language based on mathematical morphology is then presented. A detailed example of the use of the system for the inspection of a particular industrial part concludes the presentation.

A new robot system has been introduced which was designed to seal the seams of car body panels. The system has nine degrees of freedom, and includes the following features: (1) A seam tracking servo using a solid-state TV camera is mounted at the robot hand to compensate for the seam deviations. (2) The robot arm is equipped with a flexible mechanism at the wrist to provide a wide working range for the seal nozzle. (3) A two axis orthogonal robot carrier is provided to make the robot follow the work on an indexed conveyor during the sealing operation. This paper deals with the structure and operation of the system and presents test results obtained on a sealing line.

This paper describes a primitive environment recognition system for advanced robots. The system can recognize the location, configuration and distribution of 3-D objects in an object space (4m×4m×2.5m) using an ultrasonic imaging method. As an experiment, we have tried to develop the recognition system for the distribution of standing persons. First, this system was tested by mechanical scanning in order to obtain a satisfactory result. Next, we have investigated the electronic scanning method. A transmitter built for testing was composed of twenty-one ultrasonic transducers (40 KHz) built into a 2-D array. The characteristics supported by measurements have agreed well with the theoretical results. Therefore, it has become clear that the ultrasonic imaging system using an electronic scanning method can be applied to environment recognition.