With the advancement of technology, flying robots are used in many applications such as mapping, inspection and safety, transportation of goods and military operations. In order to navigate these drones, there is an urge for real-time positioning. In outdoor environments, this is often done using satellite positioning systems. But in indoor environments, gnss are not operational. Therefore, in order to navigate flying robots inside built environments, other positioning methods should be implemented, such as the use of video cameras. It is also possible to use the data of the inertia sensor as a complement to image data. In interior area, due to the lack of texture and lighting problems, in addition to the effects of image points, the effects of image lines can also be used. Based on the above, researchers have proposed many algorithms to determine the position of flying robots. However, in order to compare these algorithms, it is sufficient to examine the general error of determining the position and time of implementation of the algorithm and to analyze the reasons for increasing or decreasing the geometric accuracy or the reason for the failure of algorithms in challenging environments. In this research, the positioning of EuRoC drones is performed using selected techniques, namely PL-SVO, ORB-SLAM 2 and vins fusion algorithms, and their results have been analyzed and compared. Also, the reasons for the success or failure of algorithms in robot positioning operations in different conditions are stated by referring to the robot physical information and radiometric properties of images. Among the mentioned algorithms, vins fusion with RMSE of less than one meter for the two tests and about 1.2 meters for the other test has the highest accuracy. Also, the ORB-SLAM 2 algorithm has an RMSE of less than one meter in the parts of the flight that have succeeded in estimating the position of the flying robot.