Game theoretic multi-agent approach to traffic flow control

Abstract

In this paper, the researchers have made an attempt to address the problem of traffic network congestion to speed up the disaster response activity during a disaster. A dynamic environment is made up of static and moving obstacles. Each moving vehicle acts as a dynamic obstacle for other moving vehicle. There by increasing the network congestion. It is shown that the game theoretic model effectively helps to construct a navigation graph for each moving vehicle and computes collision free geometric path in a 2-D and 3-D work space. In this study (i) the problem of Network congestion is addressed by using the concept of Multi-Agents where, the transportation network is depicted as a non-deterministic, non-cooperative n - person game and (ii) the computation of collision free geometric path by using a 2 player non-deterministic game which detects the dynamic obstacles based on the observation made by the object sensor, computes all the possible geometric paths and selects the optimal path for the vehicle to travel in a 2-D and 3-D workspace (iii) Cooperative game theory model to resolve traffic congestion at a merge using the concepts of Time-Division Multiple Access(TDMA) data slot that propagates through the a transportation channel (iv) Construction of the multi-agent navigation graph online taking into consideration only the obstacles affecting the query.