Autonomous Intersection Management (AIM) is an intersection through which vehicles are independently directed according to a given policy. This is an innovative approach based on a positioning system and on the wireless negotiation of the right of way. However, AIM raises the issue of the feasibility that is strongly related to whether the vehicles are able to cross the intersection without collisions. In this paper, Instead of vehicles we used robots in order to prototype an elementary AIM. For that we present a Petri net model of an elementary AIM. The Petri net model is used to define a centralized protocol that allows the robots to cross the intersection without collisions. This protocol uses an event observer that controls the distribution of the rights of way to the robots. The event observer is based on a linear system of inequalities under constraints. An elementary AIM of robots is developed. Tests show that the protocol allows a safe intersection with two different scheduling algorithms. Although delayed and several losses of messages, collisions and deadlocks are avoided by means of the event observer.