The focus of this project is on automated production and manufacturing processes that require on-line assessment and sorting. European SMEs working in the field of manufacturing are often characterized by a low technological level. Moreover, repetitive and unskilled tasks are still demanding to operators. A clear disadvantage with respect to non-European industries, who have much lower production costs (e.g.,man power), must be pointed out. This proposal concerns the study and development of innovative techniques for automating and optimizing industrial processes. The system comprises two main components which will communicate directly to achieve automated sorting of disordered work-pieces such as fur skins, fabric or other flexible components. First a machine vision system comprising a computer controlled lighting and digital camera and associated image processing boards and software will be used to recognize specific work-piece objects. The location and orientation of individual work- pieces will be determined. The second system component, a manipulator arm will be to pick up the work-piece, re-orient it and place it down in a prescribed ordered fashion ready for transport to the next manufacturing or finishing process. The computer controlled robot arm, which can be a standard light-weight industrial robot, will be fitted with special purpose gripper systems which will enable reliable handling of these delicate flexible materials without damage. Machine vision systems are actually more economic thanks to advanced in sensor, lighting techniques and computing power. In particular, specific project objectives are: a) to increase the competitiveness of European SMEs and reach a satisfactory market situation; b) to increase the safety of the working environment according to EEC directives; c) to prove the feasibility of a new high-performance processing systems in this application. The main innovative aspect of this project lies in the integration of new artificial vision techniques (e.g., statistical morphology for feature extraction, artificial neural networks for surface recognition, etc.) for a specific visual inspection task whose goal is to extract reliable 2D and 3D information aimed at recognizing either fur skins or textile items according to sizes and shape. In particular: 1) the optimization and the regulation of the functioning of each visual sensor by regulating its own characteristic parameters aimed at improving the interpretation/recognition processes. 2) control and robotic techniques will be used for the automation and tele-operation of unskilled or repetitive tasks. 3) new artificial vision techniques will be developed to improve the visual inspection task. 4) a man-machine interface will be studied for on-line displaying to the operators each particular of the automated process. At the completion of the project, the final system will have to be tested and validated, also using a pilot process equipment.
Funding SchemeCRS - Cooperative research contracts
WC1E 7JE London