From the lab · November 16, 2026
New preprint on consensus with merging agents
A hybrid consensus framework for agents that physically merge when they come close
We are excited to share a new preprint, “Consensus over Networks with Merging Agents.”
The paper studies consensus among physical agents that dock and merge when they come sufficiently close. It introduces a hybrid protocol in which agents follow consensus dynamics between docking events, while each merge combines both their states and their network neighborhoods. This structure keeps merged agents together and makes the evolving network a blow-up of a contraction of the original graph.
Using this graph-theoretic characterization, the paper shows that all agents merge at their initial centroid in finite time, with an explicit guarantee that holds regardless of the order of the merges and without requiring sensing links beyond those in the original network. Extensions address formation assembly, collision avoidance, and leader–follower networks, with examples in modular self-assembly and cooperative transport.