The International Research Fellowship Program enables U.S. scientists and engineers to conduct nine to twenty-four months of research abroad. The program's awards provide opportunities for joint research, and the use of unique or complementary facilities, expertise and experimental conditions abroad.
This award will support a twelve-month research fellowship by Dr. Jeremy J. Blum to work with Drs. Tom Mathew and Dr. M. C. Deo at the Indian Institute of Technology in Bombay, India.
The goal of this project is to improve the design of routes and schedules for bus transit systems, a problem with potential global application. The proposed research will utilize modeling and simulation techniques from the PI's dissertation research, and extend his research in the flow optimization for freight railroads. In collaboration with Union Switch and Signal, they integrated an effective intelligent agent optimization system with their train dispatch system. The intellectual merit of this research is the development of a deeper understanding of the application of intelligent agent systems to optimization problems with multiple, competing objective functions. On a theoretical level, this research will assess the ability of the system to be structured to manage the competing objectives in this multi-criteria optimization problem. Dissemination of both this theoretical research in intelligent agent optimization systems and its application to the bus transit system routing will produce the broader impacts of the research. They expect that this research will confirm the hypothesis that a team of intelligent agents, encompassing a multi-algorithmic approach to optimization, will more effectively and efficiently solve the Transit Route Network Design (TRND) problem than current approaches based on individual heuristics or generic optimization techniques. Due to the large size of transit systems in India, it is an ideal place to test the scalability of these optimization systems. The technical approach for this project begins with the creation of an intelligent agent system for transit optimization. Unlike previous approaches to this problem, they will identify a wide range of relevant domain-specific heuristics and generic optimization techniques and then encode these algorithms as agents. Then, the performance of the intelligent agent system will be compared with homogeneous approaches, consisting of individual heuristics or meta-heuristics. The expected results of this research are that the intelligent agent system will produce better solutions than homogeneous approaches and that the new mechanisms will produce a more efficient system.
