Development of methodological framework for effective deployment and operational optimization of flexibly automated production and service systems

Georgia Tech Research Corporation, Atlanta GA

Investigators

Abstract

Development of a methodological framework for the effective deployment and the operational optimization of flexibly automated production and service systems The quest for an ever increasing level of automation is a prevailing trend in many sectors of the modern economy, ranging from the contemporary production systems and their supporting material handling systems, to the various intelligent transportation systems that are currently employed or contemplated for the future transport needs, to the internet-based workflow management systems that have been proposed for the mechanization of various routine operations taking place in service sectors like banking, insurance claiming, and the backend transactions of contemporary logistics systems. However, the existing theory for supporting the deployment and the real-time management of such automated operations in the aforementioned application contexts is challenged by the operational complexity and the large-scale nature of these applications. Hence, the proposed research program seeks to develop a novel methodological framework that will address the currently perceived needs with respect to the aforementioned task. More specifically, the successful implementation of the proposed program will enable a more streamlined and a more robust deployment of the sought automating capability and it will help in materializing the operational flexibility and efficiency that have been contemplated for the corresponding environments. From a methodological standpoint, the proposed research program will abstract the aforementioned operations as a class of resource allocation systems (RAS) that must be controlled for operational safety, behavioral correctness / logical consistency, and eventually performance. More specifically, the proposed research program will leverage the existing theory of RAS supervisory control developed by the PI and his collaborators, and it will complement this theory with the necessary performance-oriented control / scheduling theory that will lead to the complete development of a real-time control framework for the considered RAS. The proposed developments will also be based on an effective integration of qualitative and quantitative DES theory in the context of the existing RAS, and they will further provide pertinent approximating schemes that can support an effective and systematic trade off between the operational efficiency of the derived control policies and their computational tractability. In fact, it should be noticed that the pursued methodology for the aforementioned integration of DES qualitative and quantitative control in the context of the considered application is a substantial contribution to the field of DES theory itself, since, historically, and in an effort to effectively manage the underlying analytical complexities, these two major DES areas have evolved on parallel tracks. Also, when viewed from the standpoint of scheduling theory, the proposed research program seeks to develop effective and efficient scheduling policies for what can be perceived as stochastic networks with blocking, a topic that has not been extensively investigated by the current scheduling theory due to the involved complexity. On the other hand, it is expected that the aforementioned grounding of the proposed research on qualitative and quantitative DES analysis and the effective integration of these two areas, will provide the necessary analytical and computational tools for the effective management of this complexity.