Cooperative systems allow the exchange of large amounts of information either vehicle to vehicle or vehicle to infrastructure. New traffic control systems can take advantage of this information, thereby making more efficient traffic regulation. This is the underlying idea of the urban traffic control strategy presented in this paper. It assumes that the real time paths of all vehicles on the network are known. Coalitions of vehicles sharing the same parts of routes are built. Each coalition is monitored by a Coordinator Agent who communicates with intersections in order to get the best traffic light configuration for his vehicles. A genetic algorithm is used to find the optimal set of traffic light configurations which satisfy the Coordinator Agents' requests. The optimal solution is obtained within the set of the proposed solutions by negotiation between the Coordinator Agents in order to reduce the total travelling time in the network.