Data-driven adaptive formation control based on preview mechanism for networked multi-robot systems with communication delays

This study addresses the formation control task of networked multi-robot systems operating in an edge-end network architecture with communication delays. We propose a data-driven distributed adaptive formation controller DDAPM for stable and precise formation performance. Implemented at each edge-node, the DDAPM controller includes a single closed-loop angular velocity control module and a double closed-loop linear velocity control module connected in parallel. It integrates a preview mechanism to capture the intentions of robots and delay compensators to mitigate the adverse effects of communication delays. We analyze the stability and consensus of the NMRS utilizing DDAPM controllers. Numerical simulations are conducted to validate the feasibility, effectiveness, and adaptability of the DDAPM. Comparative simulations between the DDAPM and other controllers further substantiate its advantages. Additionally, an ablation study is performed to highlight the effectiveness and necessity of delay compensators and adaptive parameter pseudo-partial derivatives. Finally, we discuss the study's limitations and propose potential directions for future research.