Optimizing road infrastructures, like traffic lights, is a cheap and effective way to face urban road congestions. Using an individual-centered approach, we model each intersection as an agent, and we focus on their control strategies. Multiple strategies have already been developed, using heuristic rules or reinforcement learning, but they do not offer enough confidence for a real deployment, as they may not be interpretable or efficient. Moreover, most of these strategies need powerful and expensive sensors. In this paper, we present a new road intersection management strategy, interpretable and sparing in sensors. By modelling the intersection management as a scheduling problem, we obtain a strategy that outperforms equivalent algorithms, while guaranteeing properties such as a bounded waiting time or the prioritization of the lanes with the most vehicles.