Abstract:
An important goal of road traffic engineering is to sustain a smooth flow of traffic on a given road network. This can be achieved by optimizing the working of traffic signals or by imposing traffic rules. Our hypothesis is that traffic signals can improve traffic flow by sharing information that allows them to react to the changing traffic conditions in a peer-to-peer fashion. As part of our proposed architecture, a traffic crossing containing a set of signals will be under the control of an Intersection Agent. Each Intersection Agent communicates with its peer Intersection Agents to let them know about the load of traffic coming towards them. Furthermore, on the receipt of a message, each Intersection Agent adjusts the cycle time of traffic signals at their respective intersection, to maximize throughput of the traffic through the intersection. Intersection Agents also have the capability to learn from previous traffic flows. This scheme permits an optimized working of traffic signals in order to minimize the delay occurring in a traffic network. The efficacy of this approach is being tested through a simulator named Intelligent Traffic Simulator (ITSIM), which has been developed to evaluate the proposed architecture under varying traffic conditions. The blend of machine learning and agent technology integrated with the features of peer-to peer model provides an adaptive flow of traffic by using traffic signals.
