Robust Scalability of Multi-Agent Systems

Multi-agent systems (MAS) are implemented in many different areas and forms in many technical applications today and will become even more important in the future. An important body of work has appeared over the years but many key questions have not been addressed so far.

Consider for instance a simple network of agents, where one is affected by a disturbance. Due to the coupling with other agents, the disturbance will travel through the network. This project will consider the question when and how the local errors grow while the disturbance propagates through the system. We will also develop suitable control algorithms, that can be implemented at the individual agents, to ensure that disturbances are attenuated and that the error signals are bounded independently of the size and the structure of the network. Whether and how this can be achieved will depend on the dynamics of the agents as well as the type of coupling and imperfections in the communication between the agents such as noise, delays or dropouts.

This problem is well known in some networks, i.e. in vehicle platoons, where it is called string stability. Hence, we will combine methods proven to be suitable to study string stability, methods suitable for handling of communication imperfections and results derived for general multi-agent systems. Research in this area will contribute to advances in MAS such as consensus and pinning networks, and will enable safe operation of these networks in realistic settings.