A Rotation-Invariant Embedded Platform for (Neural) Cellular Automata
Dominik Woiwode, Jakob Marten, Bodo Rosenhahn
TL;DR
The paper tackles the problem of realizing neural cellular automata (NCA) in physical, distributed robotic systems by introducing a rotation-invariant, battery-powered hardware platform. It couples a symmetric 5×5 cell array with edge-to-edge communication and a lightweight firmware–simulator stack, enabling independent operation of cells and local communication under arbitrary orientation. The authors develop hardware-aware training to achieve rotation-invariant shape classification and validate the approach with two experiments: firefly-like synchronization and isotropic shape recognition across multiple datasets. The work provides open-source hardware, firmware, and a browser-based simulator, offering a practical platform for education and research in distributed self-organization and emergent computation.
Abstract
This paper presents a rotation-invariant embedded platform for simulating (neural) cellular automata (NCA) in modular robotic systems. Inspired by previous work on physical NCA, we introduce key innovations that overcome limitations in prior hardware designs. Our platform features a symmetric, modular structure, enabling seamless connections between cells regardless of orientation. Additionally, each cell is battery-powered, allowing it to operate independently and retain its state even when disconnected from the collective. To demonstrate the platform's applicability, we present a novel rotation-invariant NCA model for isotropic shape classification. The proposed system provides a robust foundation for exploring the physical realization of NCA, with potential applications in distributed robotic systems and self-organizing structures. Our implementation, including hardware, software code, a simulator, and a video, is openly shared at: https://github.com/dwoiwode/embedded_nca