Uncertainty Propagation and Cooperative Control Theory of Coupled Disturbances in Multi-Robot Systems

Abstract

Multi-robot systems (MRS) are increasingly used in complex and uncertain environments, such as search and rescue missions or autonomous manufacturing. An important but less studied problem is how the impact that affects multiple robots at the same time propagates in the system and affects the cooperative control performance because of sharing dynamics, communication links or environmental fields. This study uses stochastic dynamics, graph theory and robust control theory to understand how uncertainty propagates in MRS and how it affects the stability of strata and the efficiency of tasks. The results show that the pairing disturbance propagates in an anisotropic way, and the network topology and control settings are the main modulators. When the characteristic mode of the pairing system is aligned with the main uncertainty direction, the disturbance is amplified, which produces “wave effect” and reduces the accuracy of consensus. In contrast, distributed disturbance observer and adaptive gain adjustment can reduce propagation by as much as 42% in simulation training. In this study, “disturbance sensitivity spectrum” is introduced to understand how different robots deal with problems. It also provides a “collaborative damping” framework that can change pairing: it is no longer a weakness, but a help to prevent interruption. This helps us to do some things, such as creating scattered observers, allocating control according to topology, and monitoring uncertainty in real time so that MRS can work normally. This work is helpful to understand how pairing can change uncertainty and provide insights for building powerful and adaptable robots in disorderly places.