System wyznaczania ścieżek przemieszczeń dla grupy autonomicznych mobilnych robotów-agentów rozpoznawczo-inspekcyjnych

W artykule opisano oryginalny system wieloagentowy, którego podstawowym celem jest rozpoznanie, a następnie przeprowadzanie inspekcji nieznanego obszaru przy wykorzystaniu grupy autonomicznych mobilnych robotów-agentów. Wnioski z przeprowadzonych symulacji systemu pozwalają na rozpoznanie obszaru wiedzy, dotyczącego szczególnych aspektów systemów wieloagentowych takich jak min: metody ustalania ścieżek przemieszczeń agentów mobilnych czy wpływ liczby agentów na globalną efektywność systemu.

The paper describes an original multi-agent system [1, 2, 10, 11] whose primary purpose is to recognize and, subsequently, to inspect continuously a limited area. The agents are autonomous mobile robots, equipped with a set of sensors that are capable of scanning the examined area and collecting information on significant features. The main decision problem necessary to be solved in the system is to determine the paths for the robotic agents in such a way that the area is recognized in the shortest time, and the inspection process is the most effective from the point of view of the state of knowledge about the significant features of the examined area. The agents should act collectively in such a way as to balance their load and, hence, to optimize the system performance. In the paper there are described in detail the consecutive steps of the algorithm (Fig. 1), and, in particular, the methodology for determining the direction of motion of robotic agents (Subsections 5.3 and 5.4). The presented multi-agent system was implemented in the Webots environment [4]. Educational robots of the e-puck class were used as robots-agents (Fig. 2), [3]. Implementation of the system made it possible to perform a series of experiments which allowed drawing interesting conclusions (Section 6) on the effectiveness of the system in achieving its primary objectives. At the end of the paper the experiment is illustrated (Fig. 7), and also the growth chart of recognition of the area for each robotic agent is presented (Fig. 8).