This paper deals with the ofï-line route planning of Automated Guided Vehicles (AGVs). We propose a conflict-free route planning approach for the AGVs' control system in a warehouse system. The candidate routes are determined by using the improved Dijkstra's algorithm. The potential conflicts are detected and classified upon the comparison of node coordinates and occupancy times. A self-adaptive strategy is implemented to resolve the potential conflicts based on the conflict and mission classes. Presented simulation results demonstrate the efficiency of the proposed conflict-free route planning approach.