In this paper, three solutions for scheduling problem of the Single-Load and Multi-Load Automated Guided Vehicles (AGVs) in Container Terminals are proposed. The problem is formulated as Constraint Satisfaction and Optimization. When capacity of the vehicles is one container, the problem is a minimum cost flow model. This model is solved by the highest performance Algorithm, i.e. Network Simplex Algorithm (NSA). If the capacity of the AGVs increases, the problem is a NP-hard problem. This problem has a huge search space and is tackled by the Simulated Annealing Method (SAM). Three approaches for its initial solution and a neighborhood function to the search method are implemented. The third solution is a hybrid of SAM and NSA. This hybrid is applied to the Heterogeneous AGVs scheduling problem in container terminals. Several the same random problems are generated, solved by SAM with the proposed approaches and the simulation results are compared. The experimental results show that NSA provides a good initial solution for SAM when the capacity of AGVs is heterogeneous.