Structured Neural Network Based Quadcopter Control Under Overland Monitoring

Abstract

In some areas where economic activity is carried out, the presence of mountains and forests is observed. In order to provide information support for the development of infrastructure and agriculture in these territories, in some cases the overland monitoring is required using unmanned technologies, in particular, quadcopters. To ensure autonomous maneuvering of the quadcopter under overland monitoring, it is proposed to use a structured hierarchical neural network control model, which includes two subnets: “reasonable” and “instinctive”. The training of these networks is carried out on various scenarios of the behavior of the quadcopter relative to overcoming possible obstacles in the five fields of vision. As a basic model for the formation of such scenarios, it is proposed to use a fuzzy inference system with input characteristics in the form of linguistic variables that reflect fuzzy areas of space within which the presence of obstacles and the distance to them are interpreted verbally, i.e. as the terms of the corresponding input linguistic variables. Overcoming obstacles is supposed to be carried out on the basis of fuzzy conclusions of the proposed system, formulated as output linguistic variables, reflecting changes in the angle of rotation in the horizontal plane, flight altitude and path velocity of the quadcopter.