AI In Advanced Traffic Management Systems

Traffic management systems have historically been limited to addressing the control of street signal lights. Algorithmic solutions to this problem have proved to be very restrictive, while expert system solutions have only shown valid results with small signal networks. None of these approaches has addressed the need for management of the overall transportation system of surface streets, interstate highways, public transportation, and emergency vehicle response. Georgia Tech has developed a distributed blackboard system designed for advance traffic management in large urban areas. Knowledge sources in the system address problems in traffic control, monitoring, congestion prediction, adaptive communication, and incident management. The knowledge sources exploit rule, frame, script, and neural network representations to solve individual traffic management problems that appear on the blackboard data structure. The resulting traffic management decisions are then implemented and evaluated through simulation.