Title: Consensus-Driven Swarm Morphogenesis: Decentralized Heterogeneous Formations via Localized Bubble Meshing Abstract: This research introduces a decentralized multi-agent control framework enabling heterogeneous robotic swarms to achieve and reconfigure complex formations without centralized coordination or global positioning infrastructure. We propose a novel hybrid approach integrating a wireless mesh network for local positioning—where agent states are fused through trilateration and Kalman filtering—with a distributed shape partitioning algorithm based on bubble meshing dynamics. Unlike conventional potential field or randomized assignment methods, our method employs simulated inter-bubble forces derived from molecular packing models to generate agent-specific goal states, ensuring uniform topological distribution within non-analytical formation contours. Agents autonomously negotiate goal assignments through collaborative optimization of displacement costs, utilizing visibility graphs and the Hungarian algorithm to reach swarm-wide consensus. Simulations in Gazebo with 50 heterogeneous agents validate superior performance in mesh quality (ϵt<2.3 ϵt <2.3, ϵg<0.28 ϵg <0.28) and total displacement reduction compared to artificial force and fractal-based approaches. A hardware implementation using vision-localized omni-wheel robots demonstrates real-time applicability. This work advances decentralized multi-agent systems by unifying local sensing, mesh communication, and geometric swarm morphogenesis into a scalable, robust formation control paradigm suitable for unstructured environments.