mlsanityData to 14 Oct 2026

Synthesizing Decentralized Controllers with Graph Neural Networks and Imitation Learning

arXiv 2012.14906 ↗First posted 29 Dec 2020cs.LG cs.SY eess.SPcited by 12

On the map: Offline RL and imitation learning from data → Imitation learning methods

Abstract (arXiv)

Dynamical systems consisting of a set of autonomous agents face the challenge of having to accomplish a global task, relying only on local information. While centralized controllers are readily available, they face limitations in terms of scalability and implementation, as they do not respect the distributed information structure imposed by the network system of agents. Given the difficulties in finding optimal decentralized controllers, we propose a novel framework using graph neural networks (GNNs) to learn these controllers. GNNs are well-suited for the task since they are naturally distributed architectures and exhibit good scalability and transferability properties. We show that GNNs learn appropriate decentralized controllers by means of imitation learning, leverage their permutation invariance properties to successfully scale to larger teams and transfer to unseen scenarios at deployment time. The problems of flocking and multi-agent path planning are explored to illustrate the potential of GNNs in learning decentralized controllers.

Key facts. Cited by 12 papers in the map (data through 14 Oct 2026). Placed in Imitation learning methods, one of 13 directions in Offline RL and imitation learning from data. Builds on 31 of its 62 references that are in the map. First posted on arXiv 29 Dec 2020.

Builds on

31 of 62 references.

Closest ancestors · 2

Same direction, newest first

  1. 2020Imitation Learning with Stability and Safety Guarantees16 Dec
  2. 2018An Algorithmic Perspective on Imitation Learning16 Nov

Other references · 29

  1. 2014Adam: A Method for Stochastic Optimizationcited by 37,240
  2. 2016Convolutional Neural Networks on Graphs with Fast Localized Spectral Filteringcited by 2,089
  3. 2013Spectral Networks and Locally Connected Networks on Graphscited by 1,102
  4. 2017Graph Signal Processing: Overview, Challenges and Applicationscited by 413
  5. 2016Structured Sequence Modeling with Graph Convolutional Recurrent Networkscited by 275
  6. 2019Stability Properties of Graph Neural Networkscited by 127
  7. 2020Graphon Neural Networks and the Transferability of Graph Neural Networkscited by 118
  8. 2018Connecting the Dots: Identifying Network Structure via Graph Signal Processingcited by 111
  9. 2018Convolutional Neural Network Architectures for Signals Supported on Graphscited by 95
  10. 2017Deeply AggreVaTeD: Differentiable Imitation Learning for Sequential Predictioncited by 93
  11. 2019Graph Neural Networks for Decentralized Multi-Robot Path Planningcited by 86
  12. 2020Graph Neural Networks: Architectures, Stability and Transferabilitycited by 80
17 more
  1. 2020Graphs, Convolutions, and Neural Networks: From Graph Filters to Graph Neural Networkscited by 75
  2. 2019Optimal Power Flow Using Graph Neural Networkscited by 63
  3. 2019Learning Decentralized Controllers for Robot Swarms with Graph Neural Networkscited by 62
  4. 2020Gated Graph Recurrent Neural Networkscited by 58
  5. 2018Learning an Approximate Model Predictive Controller with Guaranteescited by 57
  6. 2020Message-Aware Graph Attention Networks for Large-Scale Multi-Robot Path Planningcited by 43
  7. 2016Stationary Graph Processes and Spectral Estimationcited by 36
  8. 2020EdgeNets:Edge Varying Graph Neural Networkscited by 36
  9. 2016Stationary signal processing on graphscited by 29
  10. 2019Graph Policy Gradients for Large Scale Robot Controlcited by 28
  11. 2018Forecasting Time Series with VARMA Recursions on Graphscited by 27
  12. 2017A Time-Vertex Signal Processing Frameworkcited by 20
  13. 2017Filtering Random Graph Processes Over Random Time-Varying Graphscited by 8
  14. 2020Wide and Deep Graph Neural Networks with Distributed Online Learningcited by 7
  15. 2020Graph Neural Networks for Distributed Linear-Quadratic Controlcited by 5
  16. 2018Ergodicity in Stationary Graph Processes: A Weak Law of Large Numberscited by 3
  17. 2019Controllability of Bandlimited Graph Processes Over Random Time Varying Graphscited by 3

Built on by

Cited by year

Citing papers per year: 2021: 3; 2022: 2; 2023: 4; 2024: 2; 2025: 1.

12 papers.

  1. 2022Graph Filters for Signal Processing and Machine Learning on Graphscited by 53
  2. 2021Wide and Deep Graph Neural Network with Distributed Online Learningcited by 14
  3. 2021Space-Time Graph Neural Networkscited by 10
  4. 2023Unsupervised Learning of Sampling Distributions for Particle Filterscited by 8
  5. 2021Scalable Perception-Action-Communication Loops with Convolutional and Graph Neural Networkscited by 7
  6. 2022Learning-based Motion Planning in Dynamic Environments Using GNNs and Temporal Encodingcited by 5
  7. 2023Graph learning in robotics: a surveycited by 5
  8. 2024LPAC: Learnable Perception-Action-Communication Loops with Applications to Coverage Controlcited by 3
  9. 2023Graph Neural Networks over the Air for Decentralized Tasks in Wireless Networkscited by 2
  10. 2024Graph Neural Networks Gone Hogwildcited by 2
  11. 2023A Scalable Multi-Robot Framework for Decentralized and Asynchronous Perception-Action-Communication Loopscited by 1
  12. 2025Real-Time Risky Fault-Chain Search using Time-Varying Graph RNNscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2024Generalizability of Graph Neural Networks for Decentralized Unlabeled Motion Planning
  2. 2023Learning Decentralized Flocking Controllers with Spatio-Temporal Graph Neural Network
  3. 2026Complex-Valued GNNs for Distributed Basis-Invariant Control of Planar Systems
  4. 2019Decentralized Multi-Agents by Imitation of a Centralized Controller
  5. 2023Simulation of robot swarms for learning communication-aware coordination
  6. 2023Graph Neural Networks for Decentralized Multi-Agent Perimeter Defense
  7. 2025Distributed Control of Network Systems in the Space of Stabilizing Graph Neural Network Policies
  8. 2026Zero-Shot Scalable Resilience in UAV Swarms: A Decentralized Imitation Learning Framework with Physics-Informed Graph Interactions