In recent years, drones have become increasingly popular, and the market size has been expanding every year. Promising applications of drones include logistics, security, aerial photography, surveying, inspection of equipment, agriculture, and disaster relief. And small drones are highly maneuverable due to their compact size, making them suitable for flying in narrow environments such as indoors. However, small drones have significant limitations on the types and performance of sensors they can carry due to their weight and price constraints. For example, to achieve autonomous flight of a drone, a depth sensor is necessary to measure the distance between the drone and the obstacle/target in front of it. However, many small drones do not have a depth sensor, or the measurable distance is very short. In this study, we propose an obstacle avoidance program for the small drone Tello using depth estimation from a monocular camera image without using a depth sensor. The feasibility of the proposed algorithm was evaluated in both simulation and real-world environments.
