A MODEL OF OBJECT COLOR ON THE BACKGROUND OF AN ARBITRARY TERRAIN

Abstract

In recent years, interest in the use of autonomous mobile robots (AMR) in areas of human activity related to security and military operations has increased significantly. Such robots should be weakly dependent on GPS and independently navigate unfamiliar terrain in the interests of performing the task assigned to them. To do this, each robot must find an object on the territory that can be considered a ground landmark. The purpose of the article is to create a new objects color model, which in the future gives the opportunity to attribute these objects to landmarks and carry out their recognition in time close to real.For the first time, a method on the use of the "snapshots" method, i.e., the analysis of sequential images recorded on the matrix of the video camera in the process of scanning the space. The connection between adjacent color columns of images is provided by a transformation matrix that connects two adjacent color columns of image pixels by Moore-Penrose inversion. Approximation errors indicate the loss of linearity of the transformation, but at the same time can serve as a feature for the detection of a landmark by the robot. The mathematical model of a landmark color for autonomous robot with scanning video camera developed in the article, taking into account the background of the surrounding terrain, is valid for any navigation conditions in the daytime.