Skip to Main Content
Agent or multiagent systems have evolved and diversified rapidly since their inception around the mid 1980s as the key concept and method in distributed artificial intelligence. They have become an established, promising research and application field drawing on and bringing together results and concepts from many disciplines, including AI, computer science, sociology, economics, organization and management science, and philosophy. However, multiagent systems have yet to achieve widespread use for controlling traffic management systems. Most research focuses on developing hierarchical structures, analytical modeling, and optimized algorithms that are effective for real-time traffic applications, as you can see from well-known traffic control systems such as CRONOS, OPAC, SCOOT, SCAT, PRODYN, and RHODES. Although those functional-decomposition-based systems are useful and successful for many traffic management problems, costs and difficulties associated with their development, operation, maintenance, expansion, and upgrading are often prohibitive and sometimes unnecessary, especially in the rapidly arriving age of connectivity. We need to rethink control systems and reinvestigate the use of simple task-oriented agents for traffic control and management of transportation systems.
Date of Publication: Sept.-Oct. 2005