Communication-free detection of resource conflicts in multi-agent-based cyber-physical systems

摘要

Multi-agent approaches can be applied to model behaviour and relations of entities in cyber-physical systems. Here entities frequently compete on insufficient resources (e.g., hardware) at the same time. Hence, resource conflicts between several agents are one of the most important conflict types in such multi-agent systems. These conflicts can significantly slow the operation of a system down, or in the worst case, might lead to a system halt. In this paper, we investigate the challenge of efficiently detecting resource conflicts. For this purpose, we introduce a conflict detection model based on beliefs of BDI agents. One benefit of our approach is that conflicts are detected using local belief state information of agents without communication. For evaluation purposes we apply our conflict detection model to a multi-agent system representing a transportation service with moving robots on a fictitious airport to measure the rate of collisions and completed transportation tasks. The evaluation study showed that the system deploying the conflict detection model can avoid collisions between moving agents and agents execute tasks successfully.