Importance has been placed on improving productivity at construction sites in a recent year, reducing bitterness and repetitive work, and improving the e?ciency of inspec- tion and management work. Since buildings usually have a unique shape, e?ciency improvement methods used in mass production plants had not been able to be applied, however, due to the recent progress in machine learning and robotics technology, complex works have also been becoming possible gradually. Especially at a construction site where the gross floor area is large, it is necessary to spend a lot of time for checking the status of daily situations and inspection work, and then it is often discussed as an object of e?ciency improvement. There are also attempts to autonomously run the rover at a construction site, photograph the surroundings with a camera, automatically check a situation with Building Information Model (BIM), and identify the unusual part. However, the floor surface of the construction site is generally complex shape in many cases, for example reinforcing bars are exposed, it is not easy to run around. Accordingly, we focused on Unmanned Aerial Vehicle (UAV) as a means to move freely without concern the shape of the floor. There are many merits because it is possible to move in any structure as long as it is a small UAV, and it is also possible to grasp states of a high altitude that cannot be reached by a rover. Although autonomous flight of UAV is almost established outdoors, the introduction of indoor autonomous flight has not advanced yet because localization technology and object recognition technology are insu?cient, and the risk of damage to people and structures due to collision and fall. There are some UAVs equipped with collision avoidance sensors, and safety is not guaranteed completely. Therefore, in this research, we developed a high safety UAV against collision and falling and conducted experiments on verifying the e?ciency of inspection work and the possibility of visualization of acquired data. To ensure strong safety, a balloon type fuselage was manufactured, and collision tests and drop tests were carried out to confirm that it is incredibly safe. Besides, we performed a flight test in the indoor construction site using the UAV and demonstrated that it could be used for inspection work. Furthermore, the walkthrough view was generated from the photographed image for making daily progress management easily. As the results, it was confirmed that some works of walking around the construction site for checking would be reduced, and productivity will be improved. In the near future, we aim to collect data by autonomous flight at night when work is stopped and to create a database of daily information.