This paper deals with the conflict avoidance problem of an Automated Guided Vehicles (AGV) system, in a Flexible Manufacturing System (FMS). Regarding the complexity of this kind of problems, it has generated many works to find an optimal strategy for scheduling and routing AGVs. In this paper, we propose a new strategy based on a temporal logic, modelled using time Petri nets. It consists in imposing suitable delays on an AGV in order to avoid a critical situation with another AGV. We built an algorithm including three main stages. In the first one, we test the evolution of our system. In the second stage, we calculate our suitable delay, if it exists, by testing the different critical scenario cases. In the last stage, we assign the delays and update the system. The validity of the algorithm is proven mathematically, and its feasibility is shown using a simulation.