A conceptual model for increasing the speed of decision-making based on images obtained from UAVs

To reduce the load on the operator of an unmanned aerial vehicle (UAV) during long search and rescue, and monitoring missions, the concept of an automatic system is proposed, which directly on board performs a prelimi-nary analysis of images received from a high-resolution navigation video camera, determines areas of interest, and sets the position of an additional camera with a reduced viewing angle to scale the image of the selected area. This allows the operator to speed up the final decision and reduces the re-sponse time to the detection, identification of an object, as well as to the preparation of a mission report. To develop a technical system that will be able to solve the tasks, a complex hierarchical model is considered, consist-ing of three components: a software system for image pre-processing and analysis, an electromechanical camera positioning system, and a higher-level human-machine complex. It is determined that the model of the first com-ponent should be based on the use of a deep learning artificial neural net-work using inference trees. The Simulink model of the positioning system contains a controller that improves the dynamics of two interconnected elec-tric drives by using three control loops in each of them. The features of in-formation perception by the operator of the UAV control complex are ana-lysed and the need to consider the effect of global precedence is noted. The results of the simulation of the electromechanical link are presented and the complexes of further research are outlined