Abstract: Unmanned aerial vehicles (UAVs) have been widely used to greatly improve production efficiency in the open-field farming. However, the UAVs are still rarely used in greenhouses at present. Among them, the film-covered multi-span greenhouses are semi-enclosed and limited space, which can make the UAV operation more difficult. It is often required for the UAVs in the film-covered multi-span greenhouses, in order to promote their use in facility agriculture. In this study, a set of feasible solutions were provided to realize the demonstrative application of UAV inspection and plant protection in the film-covered multi-span greenhouses. Autonomous flight quadcopter was also used in the UAV system. In the accurate positioning module, the RTK-GNSS was combined with a dual Extended Kalman Filter (EKF) switching strategy, in order to achieve the accurate positioning under the film-covered multi-span greenhouse environments. The primary EKF was used the RTK-GNSS to provide the high-precision horizontal and altitude information. The secondary EKF was also used RTK-GNSS for the horizontal positioning, while the altitude information was provided by LiDAR. This dual EKF configurations were ensured the stable positioning during autonomous inspection and plant protection flights. Additionally, a high-speed shutter-controlled camera was employed to accurately capture the image location data during inspections. Experimental results showed that the drone system was achieved in the safe, stable autonomous inspection flights within the film-covered multi-span greenhouse environment, with a horizontal position error of ±7 cm for the returned inspection images transmission. In the obstacle sensing and avoidance module, a perception and obstacle avoidance solution was designed to combine a forward-facing stereo vision with the multi-directional ultrasonic sensors. A reactive obstacle avoidance algorithm was employed to realize the intelligent obstacle sensing and avoidance. The avoidance distance was selected as an experimental indicator, in order to assess the performance of the obstacle avoidance. A simple and effective approach was introduced to measure the upper bound of the minimum avoidance distance for the real obstacle. The accuracy of the sensing and avoidance module was evaluated as well. The results showed that the drone system was achieved in the minimum distance of obstacle avoidance less than 38.5 cm, indicating the safe and reliable close-range obstacle avoidance. In the crop protection spraying module, a variable control system was adopted for a speed-adaptive UAV pesticide variable spraying. The initial flow rate and the initial speed were calculated, according to the plant protection target and the operation area. The nozzle flow rate was then adjusted adaptively in the real-time speed of the UAV during flight. The UAV spraying of the crop protection was achieved more than 80% of the crop protection control effect on the leafy greens with the striped flea beetle, which was better than the manual prevention of the pesticides. The feasibility and validity were fully verified to adopt the autonomous flying UAVs. The plant protection was also carried out in the film-covered multi-span greenhouse. Autonomous drones can be expected to serve as the viable solution to the greenhouse crop monitoring and protection, effectively replacing conventional labor-intensive tasks. The finding can provide a foundational framework for the application of the autonomous drones in facility agriculture. The operational efficiency can be promoted for the pest and disease control in the sustainable agricultures under controlled environments in smart agriculture.