A comprehensible artificial intelligence based application that securely stores critical flight data of unmanned aerial vehicles and enables autonomous landing based on terrain conditions

The aim of the master's thesis is to enable unmanned aerial vehicles to perform a safe landing by accurately detecting the obstacles and variable terrain conditions that they may encounter during landing. In this process, it is also aimed that the decisions taken by the unmanned aerial vehicles can be explained in a stable manner. Within the scope of the study, the design of the unmanned aerial vehicle was realized with sensors, Raspberry Pi 3B+ and software prepared in Python programming language. The unmanned aerial vehicle was flown in different terrain conditions and data were collected for different environmental conditions and the data obtained were used for training the deep learning model designed specifically for the study. The deep learning model developed for the study developed safe landing strategies by analyzing environmental data during landing. Blockchain integration ensured secure and seamless transmission of telemetry data. The results show that the developed system significantly improves the landing safety and operational efficiency of unmanned aerial vehicles.