Planning the behavior of an autonomous flying robot in a space of subtasks. Inference procedures

The paper gives a brief description of the knowledge representation model for autonomous flying robots in the form of a set of typical subtasks, without reference to a specific area, which makes it possible to construct complex programs of goal-directed activity in a priori undefined conditions. The solution of a test problem is presented, which has shown the efficiency and effectiveness of the proposed model of knowledge representation and processing for building an intelligent problem solver for various autonomous mobile intelligent agents. We developed knowledge-processing procedures that allow flying robots to form a plan of goal-directed behavior based on the automatic growth of a reduction network model for solving complex problems in the space of subtasks. We gave boundary estimates of the complexity of inference procedures, confirming that the proposed model of knowledge representation and processing allows flying robots to automatically build plans for goaldirected behavior with polynomial complexity in underdetermined conditions.