The performance of Automated Guided Vehicle systems is highly related to the implemented control strategies for vehicle fleet management. Especially the assignment of transport carriers to vehicles and the decision on the processing sequence have a high impact. So far, a dynamic transfer of transport carriers between the vehicles of an Automated Guided Vehicle fleet has not been sufficiently investigated. Nevertheless, applications from other areas like passenger transport or courier services show promising results in reduced vehicle movements and delivery times. However, the approaches to generate solutions of these problems cannot be applied to the control of Automated Guided Vehicle systems. These planning tasks differ significantly in modeling (e.g. the use of a single depot as start and end for all vehicles) and solution generation (e.g. no real-time requirements). Hence, there is no sufficient control approach for Automated Guided Vehicle systems considering dynamic carrier transfers. A heuristic approach adapted to the control of Automated Guided Vehicle fleets in intralogistics systems is presented in this article. The approach is called Modified Adaptive Large Neighborhood Search. The article describes the basic concepts of the approach and the adaptions to the field of application. Experiments based on generic test instances prove that the approach is sufficient to plan transfer operations for small vehicle fleets. Furthermore, potentials and limitations for the application in industrial systems are discussed.