Distributed Swarm Morphogenesis with Neural Cellular Automata
Distributed Swarm Morphogenesis with Neural Cellular Automata
This research presents a modular framework that integrates neural cellular automata (NCA) with site assignment and collision-aware control to enable robots to assemble desired shapes through local interactions. The authors introduce a decentralized attachment-site discovery mechanism where robots in the formation share NCA states to identify new sites, allowing progressive growth without predefined lists. To address the gap between simulated growth and physical assembly, the method employs offline sparse-retention fine-tuning that simulates partial attachment and a false-positive loss to penalize invalid activations. Simulations show the system can form diverse lattice morphologies scaling to several hundred robots with improved accuracy. Experiments using Crazyflie quadrotors validate the pipeline using robot-relative observations reconstructed from external tracking.
The paper proposes a modular framework combining neural cellular automata with collision-aware control to enable decentralized swarm morphogenesis, validated via simulations of hundreds of robots and experiments with Crazyflie quadrotors using external tracking for robot-relative observations.
Source: arXiv Robotics — research abstracts · Read original article ↗
Loading article text…
What the source reports
Publisher-reported claims, with original evidence. These results have not been independently verified by RoboSignal.
- Environment
- SimulationOpen source S3
- paper: not_reported
Source excerpts and review record
Automatically extracted; no manual editorial approval recorded.
NCA states to identify new attachment sites, enabling progressive growth without a predefined list of sites during deployment. To bridge the mismatch between conventional NCA growth and gradual physical assembly, we additionally propose offline sparse-retention fine-tuning that simulates partial attachment, together with a false-positive loss that penalizes invalid site activations. Simulations demonstrate formation
Open source S3
of diverse lattice morphologies, scaling to several hundred robots, and substantial improvements in morphology accuracy from the proposed fine-tuning. Experiments with Crazyflie quadrotors validate the integrated execution pipeline using robot-relative observations reconstructed from external tracking.
Open source S4
Source：arXiv Robotics — research abstracts · arxiv.org