Ensuring the longevity and efficient performance of road infrastructures necessitates the creation of sound maintenance and rehabilitation investment strategies. These strategies are instrumental in ensuring that the infrastructure not only lasts longer but also continues to provide the expected level of service, even when there are limitations on available capital resources. This research paper presents an innovative approach to this challenge by leveraging the power of reinforcement learning. The primary objective is to establish an optimal policy that can guide decisions regarding the selection of appropriate maintenance, repair, and rehabilitation options for an interconnected system of road facilities. The foundation of this approach lies in the policy gradient method. Traditional methods often grapple with computational challenges, especially given the myriad of possible combinations of varying network conditions and the range of maintenance and rehabilitation alternatives. However, the method proposed in this paper adeptly circumvents these computational complexities. An essential feature of the devised policy is its ability to recognize and factor in the intricate interdependencies that exist among different facilities within a comprehensive road network. Such a holistic view ensures that decisions made for one facility take into account potential impacts and benefits for others. To validate the efficacy, applicability, and robustness of the proposed reinforcement learning approach, the paper delves into numerical studies. These studies specifically target concrete bridge decks, which are vital components in many road networks. Through these studies, the paper seeks to underline the significant advantages offered by the proposed method, while also highlighting its feasibility for practical implementation and its capacity to deliver superior results. This research contributes a novel method that promises to revolutionize the way road infrastructure maintenance and rehabilitation decisions are made, with potential implications for cost savings, improved infrastructure longevity, and enhanced service levels.