Photogrammetric based image acquisition and flight optimization by unmanned aerial vehicle

Unmanned Aerial Vehicles (UAV) and Unmanned Aerial Vehicles Systems (UAS) are one of the most important research topics of the last century. Due to the widespread use of civilians and easy access by researchers, UAS are preferred in critical areas of use in nature and human life. Due to the rapidly developing UAS technology, new issues arise, and researches that appeal to everyone are carried out. The increase in the use of UAVs in the field of geomatics engineering in recent years has led to some problems. One of the subjects that researchers especially work on is UAV flight optimization. Today, energy is vital for UAVs, most of which (except those carrying solar cells) are fueled by finite sources. In daily life, the spontaneous optimization processes of the human brain have recently been the subject of Artificial Intelligence (AI) applications that can think like a human. Computer learning, deep learning, Artificial Neural Networks (ANN) and many similar optimization algorithms are also effective in the improvement of AI subjects. It is obvious that AI algorithms will provide important advantages for daily life by solving the need for optimization problems in electronic industry. The purpose of this research is to determine the optimum conditions underneath that the UAV will keep within the air and perform flight optimisation by analyzing the leads to photogrammetric use. Following this purpose, all parameters regarding the flight time were diversified, and different check results were observed. As a result of the field studies performed in different types of areas such as rural areas, urban areas and different slope conditions such as sloping area and multi-slope area, fifty-three trial flights were managed, and observations were recorded. The Battery Status and Flight Time determined as output parameters were tested with the trained network using input parameters (UAV Type, Ground Sampling Range, Overlap Rate and Atmospheric Conditions) determined in fifty-three different UAV flights performed at the field in the optimization study using ANN. The optimization of flight planning parameters was aimed by evaluating the images taken to produce photogrammetric products. Flight optimization is taken into account as a whole, the input parameters are varied, and therefore the regression results of the output parameters are examined. Within the framework of ANN, the effects of different training algorithms (Gradient Descent -GD- Algorithm and Levenberg-Marquet-LM-Algorithm) on optimization were investigated, and the states of inputs before and after normalization were compared. A graphical interface (GUI) was prepared, and optimized knowledge was connected to the GUI as a centre for the users to achieve the results. it's aimed to provide a concept for optimisation by process samples of flight designing that are most often employed in geomatics engineering. The results disclosed that the optimization model, within which all input parameters were used, calculable with 82% accuracies with post-normalization data within the GD algorithmic rule for the most effective results. within the estimation results wherever completely different parameters were excluded from the optimization, it absolutely was seen that the best results were 69% in the GD algorithm with pre-normalization data. As a result, it absolutely was determined that the optimisation of the flight with ANN within the use of photogrammetric UAVs.