Decentralised self-organisation of pivoting cube ensembles using geometric deep learning

2025-09-03

AI Summary

This work addresses autonomous reconfiguration of 2D homogeneous rotating cubic modular robots using only local neighborhood information. We propose a decentralized control framework integrating geometric deep learning and reinforcement learning: a group-equivariant graph neural network embeds grid symmetry priors, while multi-round message passing enhances global awareness from local observations. Each module makes distributed decisions solely based on its own state and communication with immediate neighbors—no global information or central coordination is required. Experiments demonstrate near-optimal reconfiguration performance under purely local interaction, validating scalability, robustness, and transferability to related systems (e.g., sliding-module robots and CubeSat swarms). Our core contribution is the first systematic integration of symmetry-driven geometric deep learning into distributed modular robot reconfiguration, overcoming the performance limitations of conventional local policies.

Technology Category

• Intelligent Robots: Learning & Optimization for ROB
• Search and Optimization: Learning to Search
• Machine Learning: Adversarial Learning & Robustness

Application Category

• Graph Algorithms and Modeling for the Web: Graph neural networks and deep learning approaches for Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Responsible Web: Machine-in-the-loop, human agency and autonomy

Abstract

We present a decentralized model for autonomous reconfiguration of homogeneous pivoting cube modular robots in two dimensions. Each cube in the ensemble is controlled by a neural network that only gains information from other cubes in its local neighborhood, trained using reinforcement learning. Furthermore, using geometric deep learning, we include the grid symmetries of the cube ensemble in the neural network architecture. We find that even the most localized versions succeed in reconfiguring to the target shape, although reconfiguration happens faster the more information about the whole ensemble is available to individual cubes. Near-optimal reconfiguration is achieved with only nearest neighbor interactions by using multiple information passing between cubes, allowing them to accumulate more global information about the ensemble. Compared to standard neural network architectures, using geometric deep learning approaches provided only minor benefits. Overall, we successfully demonstrate mostly local control of a modular self-assembling system, which is transferable to other space-relevant systems with different action spaces, such as sliding cube modular robots and CubeSat swarms.

Problem

• Decentralized reconfiguration of pivoting cube modular robots
• Neural network control with local neighborhood information
• Geometric deep learning incorporating grid symmetries

Innovation

• Decentralized neural network control with local information
• Geometric deep learning incorporating grid symmetries
• Near-optimal reconfiguration through neighbor interactions