Architecture of multi-agent system for traffic signal control

The essential matter of urban traffic control is to timely evacuate the most potential congesting movement in an intersection. Distributed multi-agent system could efficiently disassemble the globe control of regional signal system into the local and single traffic controller with autonomously intelligent actions. The paper discusses a flexible structure for multi-agent-based traffic control system by adopting technologies of object-oriented case representation, case-based reasoning (CBR) with case retrieval nets (CRN), and cooperative distributed problem solving (CDPS). And an example of traffic signal control in Guangzhou CBD shows the obvious advantages: distinct architecture, flexible assembly and efficient control impact. It will be suitable to the situations with dynamical and complex traffic flow characters.