Abstract: Although Multi-Agent Systems (MAS) can handle various problems in heterogeneous environments, unknown problems may arise at runtime due to their inherent heterogeneity when dealing with cyber-physical systems. Therefore, designers of intelligent cyber-physical systems must adopt approaches that enable adaptability and fault tolerance at runtime. This work proposes techniques for dynamically adding, removing, and swapping resources (sensors or actuators) in Embedded MAS at runtime. These approaches utilize the BDI model and a customized agent architecture capable of perceiving the availability of microcontrollers integrated into the system through serial communication. A case study was conducted to analyze five scenarios that an embedded MAS can deal with, starting from where there are no existing physical resources and new resources must be added at runtime and a scenario where failures, replacement, and the need for upgrade scenarios occur. Our approach was evaluated and tested in three different BDI frameworks, demonstrating that swapping resources at runtime is a promissory feature to guarantee the adaptability of intelligent cyber-physical systems.
Loading