The control architecture RoViDiAsS — A robotic visual disassembly and assembly system

Abstract

In highly automated and dynamic production systems the plant availability plays a crucial role, especially in terms of economic goals. In order to ensure a high system availability various maintenance strategies like preventive or predictive approaches are studied and applied. Through the increasing complexity of production systems, more and more methods for (semi-)automation of maintenance tasks are in use, especially for fault-diagnosis. Today, service or repairing tasks are usually performed by a human operator. To enable a flexible automation of different maintenance tasks this paper introduces the modular robot control architecture of RoViDiAsS (Robotic Visual Disassembly and Assembly System). RoViDiAsS combines CAD data from assembly models with vision system data for planning possible manipulations for an autonomous robot. The use of offline data from CAD and online data from the vision system enables the compensation of uncertainties as well as a suitable planning approach. RoViDiAsS is therefore divided into five function modules which handle different responsibilities from the management of the input data to controlling the different manipulations. The architecture is designed in a modular way, making it possible to integrate new functions and algorithms. This contribution describes the structure of the system architecture and closes with the preparation of experiments.