Development of the method of stochastic dynamic programming with complete information about the state vector in the problem of information and telecommunication robot optimal control

Abstract

The paper is focused on an information and telecommunication robot (ITR) that is a cluster of precision mobile mechanical objects (sensors or multisensors) that perform a common mission concerning either monitoring or protection of the territory or the coordinates of the search ob-jects location.ITRs are equipped with wireless telecommunication systems and high-performance small-sized on-board computers that allow the use of cloud technologies and fog computing. A group of unmanned aerial vehicles (UAVs) that perform a common mission (such as searching for victims of natural disasters) and are equipped with interoperable wireless tele-communications systems can be considered as a flying ITR. Mobile sensor networks (robotic sensor networks and systems) are also considered as a possible technical implementation of ITR. From a theoretical point of view, these control objects can be classified as composite dy-namic systems (CDS). The article also analyzes the notion of trajectories of composite dynam-ical systems that are called branched trajectories in the scientific literature. These trajectories consist of several sections of joint movement of CDS subsystems and some sections of their individual movement to the targets along separate branches of the trajectory. Analysis of the physical content of the phenomena that accompany the process of ITR (and especially of mobile ITR) functioning has shown that such clusters are under accidental influence. Therefore, sto-chastic CDS should be considered as a model of ITR functioning. To solve the problem of opti-mizing the process of stochastic CDS control, there arises a task of developing a method of sto-chastic dynamic programming with complete information about the vector of the ITR state. The article describes a solution to the problem of stochastic composite dynamic system control op-timization, with an arbitrary branching scheme provided that complete information about the state vector is available.