Federated Single-Agent Robotics: Multi-Robot Coordination Without Intra-Robot Multi-Agent Fragmentation

Simulation code for the paper:

Xue Qin, Simin Luan, John See, Cong Yang, Zhijun Li. "Federated Single-Agent Robotics: Multi-Robot Coordination Without Intra-Robot Multi-Agent Fragmentation." 2026. arXiv: 2604.11028

Overview

Protocol-level simulator evaluating three fleet coordination architectures:

• FSAR (Federated Single-Agent Robotics) — proposed approach
• CFC (Centralized Fleet Controller) — baseline
• DHMA (Decomposition-Heavy Multi-Agent) — baseline

Experiments

Script	Description
fsar_simulator.py	Main evaluation: 5 scenarios x 20 runs x 3 architectures + statistical tests
ablation_full.py	Component ablation study
scaling_experiment.py	Fleet scaling evaluation (4, 8, 16 robots)
statistical_analysis.py	Additional statistical analysis

Metrics

• Coordination success rate
• Governance locality
• Recovery containment
• Authority conflicts
• Policy violations
• Reassignment latency
• Human intervention rate
• Audit attributability

Requirements

Python >= 3.11

No external dependencies required.

Usage

Run main evaluation:

python fsar_simulator.py

Run ablation study:

python ablation_full.py

Run scaling experiment:

python scaling_experiment.py

License

Licensed under the Apache License, Version 2.0. See LICENSE for details.

Citation

@article{qin2026fsar,
  title={Federated Single-Agent Robotics: Multi-Robot Coordination Without Intra-Robot Multi-Agent Fragmentation},
  author={Qin, Xue and Luan, Simin and See, John and Yang, Cong and Li, Zhijun},
  journal={arXiv preprint arXiv:2604.11028},
  year={2026}
}