Connected and automated vehicles will soon disrupt the logistics and freight transport system. While AI advancements offer potential solutions, challenges like high development costs, safety requirements, and legal issues persist. This paper presents an optimization algorithm for scheduling automated guided vehicles (AGV) in yard logistics. In the problem setting, a fleet of AGVs needs to fulfill orders in an industry compound. The problem is modeled as a time-dependent pickup and delivery problem with time windows and solved with a large neighborhood search approach with destroy and repair neighborhoods and integrated conflict detection. Computational experiments show that the algorithm is able to handle instances with up to 100 orders over 8 h and a fleet of up to four AGVs. Furthermore, they show the potential and scalability of integrating an optimization algorithm with an effective fleet management system in such a logistics system.