A Methodology for Designing Robust Central/Self-Organising Multi-Agent Systems

Organic Computing (OC) has the objective to use principles that are detected in natural systems. Consequently, OC tries to develop systems that are adaptive, flexible and robust at the same time utilising advantage of the organic properties of OC. In this regard, the robustness of OC systems is a key property, because the environments of such systems are dynamic. In this paper, we propose an interdisciplinary methodology, “Robust Multi-Agent System” (RobustMAS) to characterise robustness of multi-agent systems. RobustMAS allows building robust multi-agent systems in presence of disturbances using the OC concept. It uses a hybrid approach (a combination of central and self-organising form) that is robust enough against disturbances. In this way, RobustMAS guarantees an acceptable system performance by limiting the degradation of the performance in the presence of disturbances. In other words, RobustMAS combines the use of a central Observer/Controller (O/C) architecture, autonomous agents, disturbances and deviations from the planned behaviour aiming to solve coordination problems in multi-agent systems. In this context, RobustMAS solves the conflict between a central controller (i.g., a coordination algorithm) and the autonomy of agents so that the system robustness can be achieved. More accurately, RobustMAS introduces a hybrid coordination of a multi-agent system. This hybrid coordination takes place in three steps: path planning, observation, controlling. Simultaneously, we introduce a metric for the quantitative determination of the robustness.