The design of public transport system must take into account different dimensions to solve two main problems of optimization: the vehicles scheduling and driver scheduling. In our work, we focused on bus driver scheduling. Its objective is to minimize the number of drivers in accordance with social and environmental constraints. By its combinatorial nature, bus driver scheduling is considered a complex task in the design process of network transport. We have proposed an approach based on hyper-heuristics whose main advantage lies in their ability to adapt to different problems. We are particularly interested in a cooperative approach, which is able to take into account changes in the resolution process. We have extended the functionality and improved performance of the traditional framework of hyper- heuristics by proposing a pattern based on an organizational model. The proposed algorithm consists of a combination of several phases and several levels. The metaphor of the coalition is used to make cooperate several hyper-heuristics. The coalition is intended to favor diversified solutions and expand search capacity with decentralized control where each hyper-heuristic has certain autonomy. It is thus possible to consider different ways of cooperation between the hyper-heuristics: sharing solutions, learning by mimetism or carrying out different competitive search strategies. The experiment was carried out both on real-world instances and benchmarks. It gave good results on both quality of solution and execution time
