Intelligent binocular compound eye vision system for detecting azimuth and distance to object on plane

The article is devoted to the prototype of an artificial binocular compound eye vision system for detecting the azimuth and distance to an object on a plane using an artificial neural network. A analytical review of modern distance and azimuth detection systems based on active and passive sensors is given. An intelligent binocular vision system is proposed, which is a passive optical sensor that allows you to determine the azimuth and distance to a round object of arbitrary size, emitting in the visible or infrared ranges of the electromagnetic spectrum. The general architecture of the compound eye vision system is considered. The main structural elements of the system are: an optical module, a hardware and software controller and a neural network module. The optical module uses a pair of lenses to convert the light signal from the object into two pixel Fourier images, which are fed to the input of the hardware and software controller. The controller performs primary processing of pixel Fourier images and converts them into two bit masks, the elements of which correspond to separate facets (each facet integrates four adjacent columns of the pixel image). The resulting bit masks are fed into a neural network module, which, based on their analysis, determines the coordinates of the object in the form of distance and azimuth.