Unmanned Aerial Vehicles (UAVs) are widely used in Internet of Things (IoT) scenarios, but factors like circuit malfunctions and network issues can lead to data loss, disconnections, and service interruptions, causing UAVs to go offline and fail to meet user needs. This problem can be effectively solved using UAV resource migration techniques. However, traditional methods lead to resource redundancy and cannot ensure effective migration decisions or the continuity of services. Therefore, we proposed a new UAV framework with a high-availability migration strategy to improve UAV swarm resource utilization. By combining Software Defined Network and Mobile Edge Computing technologies, we designed a UAV collaboration architecture model and introduced large language model agents in the edge layer for better task execution. Additionally, we created a two-tier collaborative UAV high-availability mechanism for handling failures in edge scenarios during UAV collaborative tasks. Our experiments on the container cloud platform showed a 42% reduction in failure recovery time and higher overall resource utilization compared to baseline algorithms. While our focus is on UAVs, our findings can potentially be applied to broader IoT and mobile edge computing scenarios such as traffic management systems and healthcare IoT systems.