Goal Based Traffic Control

One problem of urban centers is the large amount of traffic which commonly leads to congestions as well as noise and air pollution. Especially, during rush hours. The idea of this Master Team Project is to develop an adaptive traffic control system that controls the traffic lights on the basis of a Goal Recognition approach. Goal Recognition is the problem of predicting/recognizing an observed agents goal on the basis of a domain model and a sequence of observed actions, this agent already executed. Many traffic lights still follow a static control pattern, although traffic flows change dynamically during the day. Hence, controlling the traffic lights on the basis of the recognized destinations of the currently observed traffic could lead to less congestions.

In this Master Team Project, a traffic control system will be developed, that controls traffic lights in a simulated environment (SUMO traffic simulator) on the basis of the recognized goals of observed traffic participants. Participants will get familiar with the traffic simulation tool SUMO as well as recent goal recognition goal recognition approaches that are based on concepts from the area of classical planning. Four students, with a strong interest in artificial intelligence, especially classical planning, will work together with four software engineering students from Cluj-Napoca.