Decentralized control of multi-robot systems

Abstract

Decentralized control of multi-robot systems is a very active research area, that aims at developing methodologies for achieving coordinated behaviors in multi-robot systems, without the need for a centralized control unit. The main idea is that of having each robot able to exchange information with its neighbors: based on such information, each robot can compute its own control input.

Within this scenario, we are developing methodologies for controlling topological properties of the graph that represents the interconnections among the robots. Such topological properties (such as the algebraic connectivity) have a strong influence on the overall behavior of the multi-robot system.

Furthermore, we are developing methodologies for obtaining high-level complex behaviors, in multi-robot systems. In fact, in the literature, several strategies can be found for obtaining simple behaviors, such as creating a formation, or converging to a common position. Our work aims at pushing forward the paradigm of decentralized control, achieving more complex tasks, that are more similar to common, daily-life, activities, such as those performed by groups of human operators.

Video

People