Navigation based on computer vision is a cheap method for drone’s position calculation. This paper presents an implementation and simulation of a navigation system based on extreme correlation. Simulation was done using ROS (Robot Operating System), Gazebo (3D dynamic simulator), a drone model and a 3D urban flight environment. Correlation between the captured image and the georeferenced image was done to calculate the drone’s position. Testes showed a position rms error less than 3m and 40ms execution time.
Инноватика-2021 : сборник материалов XVII Международной школы-конференции студентов, аспирантов и молодых ученых, 22-23 апреля 2021 г., г. Томск, Россия. Томск, 2021. С. 129-132