Vision systems as cyber-physical sensors

Abstract

Machine Vision Systems are key components for information retrieval in modern production systems. Their unrestricted programmability in terms of image analysis makes these systems ideal resources for users to realize the required flexibility in production. In the decentralised, product-controlled manufacturing process envisaged in Industry 4.0, various products set tasks to image processing systems, which raises new demands in terms of self-description, intelligence and parameterisation. This article presents an approach for intelligent image processing systems which are able to perform measuring and inspection tasks independently of product shapes. These systems are capable of M2M communications by offering services in a network. Based on a classification of common industrial image processing functions, this article introduces a list of services for vision systems as well as necessary aspects of self-configuration. An implementation with industrial hardware and software is demonstrated.