Article outline

Problem: The solution to one of the theoretical problems of distributed cloud management in a group of robots is the decomposition of tasks between robots of a group.

Methods: the methods of system analysis and multiagent approach in the development of cloud computing system architecture.

The various variants of the decomposition of tasks in a group of robots using cloud computing technologies proposed in the results of the research make it possible to use the system to control a group of robots when solving complex computing tasks that require a large number of local calculations and approvals. Three ways of decomposition are proposed: on the basis of linear distribution, swarm interaction and synthesis of solutions. Linear distribution involves dividing the problem into smaller ones in the form of a hierarchical tree, in which the selected subtasks of the lowest level of decomposition are solved in separate cloud applications. Roaming interaction assumes that the largest tasks are solved individually by individual robots from the group, and the selected tasks at each level of decomposition are solved in separate cloud applications. In the synthesis of solutions, individual subtasks are integrated into the overall solution, and the number of cloud applications in this case corresponds to a set of methods (approaches) by which a solution of the particular problem can be found. Depending on the general task assigned to the group of robots and the functionality of individual robots, one of the methods of decomposition is selected. At the next level of the management system, effective methods for allocating the resources of the computing system are proposed for their optimal use, taking into account the chosen decomposition model.