Problems of Mathematical Pattern Recognition in Robotic Systems

Abstract

Unmanned systems have proven to be effective, but there are still challenges that need to be addressed to improve and modify them, such as increasing flight time, control range, and transmitting information through communication channels to ground control points in the required quantity and at the required speed. UAVs that need to determine their position are equipped with a GPS sensor device for spatial orientation, which allows them to determine that they have arrived at a given point, the azimuth to the next point, the height relative to the level of the world's oceans, and other information. Also, during the flight, the sensor may lose the signal from the satellite or the module may reboot and the UAV will lose its orientation in space. The solution is to increase the bandwidth and noise immunity of the information transmission channels, which can be achieved by concentrating on board the maximum number of devices that operate in software mode without constantly exchanging information with the control centre. 
The orientation of an unmanned aerial vehicle on the ground is usually achieved using GPS satellite communication systems, which can be unstable under certain conditions. A promising solution to the problem of terrain orientation in the absence of a satellite navigation system signal is the use of a terrain positioning algorithm without the use of GPS satellite communication systems. A new positioning algorithm has been developed and implemented using the image matching method to search for an image element on a terrain map, which has increased the speed of computation, reduced the likelihood of search errors and significantly improved the accuracy of the algorithm. The template search method allows to find the required area on the map with almost 100% correspondence, which makes it possible to use it as a component of the navigation system of an unmanned aerial vehicle in conditions of active electronic countermeasures.