Intelligent autonomous navigation system for UAV in randomly changing environmental conditions

Abstract

The safe and reliable navigation of such autonomous systems as unmanned aerial vehicles (UAV) is a complex open problem in robotics, where a robotic system must simultaneously do many tasks of perception, control and localization. This task is especially complicated when working in an uncontrolled, unpredictable environment, for example, on city streets, in wooded areas, etc. In these cases, the autonomous agent must not only be guided to avoid collisions, but also interact safely with other agents in the environment. The developed system allows navigation of unmanned aerial vehicles in difficult environmental conditions. The results of training and the operation of the autonomous navigation system in the forest are presented. The system finds and follows the paths that are fairly difficult to distinguish. The results of field experiments are presented. Presentation of the model is presented on the youtube.com channel.