In order to address the decision planning problem in multi-AMR scheduling systems, this paper proposes a hierarchical strategy-based multi-AMR scheduling algorithm under spatiotemporal conflicts and nonholonomic constraints. Firstly, this paper proposes the weighted spatiotemporal hybrid A* algorithm as the lower-level of this algorithm to conduct single AMR path planning, ensuring that the planned paths conform to the kinematic model of the Ackermann-type AMR. Then, this paper proposes a conflict detection tree combined with the separating axis theorem as the upper-level algorithm to conduct precise spatiotemporal conflict detection on the paths planned by the lower-level of this algorithm. The detected spatiotemporal conflicts are utilized as the spatiotemporal constraints for the lower-level of this algorithm. Finally, the lower-level of this algorithm is utilized once again, taking into account the spatiotemporal constraints for path planning, to achieve efficient AMR conflict resolution. Simulation results demonstrate that the proposed approach efficiently generates conflict-free paths satisfying nonholonomic constraints for multi-AMR scheduling systems.