Pith. sign in

REVIEW 5 major objections 5 minor 43 references

LLM-Flock: Decentralized Multi-Robot Flocking via Large Language Models and Influence-Based Consensus

T0 review · 5 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read LLM-Flock claims that coupling independent LLM-generated plans with an influence-based consensus protocol prevents centroid collapse and divergence, producing stable multi-robot formations.

desk verdict A simple degree-based plan consensus rule does stabilize LLM-generated formations, but the paper overclaims: it synchronizes plans, not correctness, and never compares to prior LLM methods. read the letter →

arxiv 2505.06513 v1 pith:MN7TT7VC submitted 2025-05-10 cs.RO

Peihan Li , Lifeng Zhou This is my paper · ORCID
classification cs.RO
keywords multi-robotsystemsflockingformationcontrollargelanguagemodelsdecentralizedconsensusinfluence-basedplanningplanadoptionCrazyfliedrones
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

The paper claims that the instability of LLM-only multi-robot flocking can be fixed without central control by adding a lightweight consensus layer. Each robot asks its own LLM for a full formation plan, then repeatedly replaces that plan with the plan of whichever neighbor can talk to the largest fraction of the team; the team thereby converges on one plan and moves to the corresponding goal positions. The authors argue that this influence-based plan consensus prevents the centroid collapse and chaotic divergence that unstructured LLM planners exhibit, and they support it with simulations across five language models and formations of three to ten robots, plus physical Crazyflie drone tests. If right, it means language-model reasoning and classical decentralized coordination are complementary rather than competing: the LLM proposes, the communication graph disposes.

What carries the argument

The load-bearing mechanism is influence-based plan consensus. Influence is defined as the fraction of the team within communication range, $I_i(t)=|\mathcal{N}_i(t)|/N$, so a robot with more neighbors is deemed more influential; each robot adopts the plan of the neighbor with strictly higher influence, then assigns itself the nearest available goal in the adopted plan. This makes communication topology, not geometric correctness, the arbiter of which plan wins, and it is what turns inconsistent LLM outputs into a single shared formation plan.

What would settle it

In a three-robot triangle run with a strong LLM, deliberately inject the distorted circle plan shown in Appendix D.1 as the plan of the most-connected robot; if the team adopts that plan wholesale and the final Procrustes error remains above the convergence threshold used in Section 4.2, then influence-based consensus alone does not guarantee stable convergence.

Watch

Extended reading notes

Core claim

The paper's central discovery is that independent LLM-generated formation plans, which by themselves collapse to the team centroid or scatter chaotically, can be reconciled into a single coherent plan by a simple decentralized adoption rule: at each negotiation round a robot replaces its plan with the plan of whichever neighbor has the largest fraction of the team within communication range, provided that fraction exceeds its own. Because plans propagate outward from the most-connected robots, the team converges to a shared formation and then moves to the corresponding goal positions, with each robot choosing the nearest unassigned waypoint. The authors report stable convergence across five different language models and team sizes of three to ten robots, in simulation and on physical Crazyflie drones.

Load-bearing premise

The whole approach assumes that the plan offered by the most-connected neighbor is a good enough formation plan and that each robot's LLM will actually move it toward its assigned goal, since influence scores measure communication centrality rather than geometric validity.

Editorial extensions

If this is right

  • Teams of LLM-driven robots can reach a shared formation plan using only local communication, with no central controller and no retraining of the language models.
  • Adding the influence-based adoption rule turns failure modes documented for unstructured LLM planners, such as centroid collapse and divergence, into convergent formations across triangle, square, circle, and cross shapes.
  • Convergence speed and residual error depend on the reasoning strength of the underlying LLM, with stronger models converging faster and to lower Procrustes error, while general-purpose models still improve under consensus.
  • The same protocol transfers from simulation to physical hardware, as Crazyflie drones form the target shapes in a motion-capture arena, showing the framework is not simulation-only.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Because influence is pure communication degree, the protocol is effectively a max-degree leader-election rule on the communication graph; an immediate testable consequence is that convergence time should scale with graph diameter and that a poorly connected robot holding the best plan can never win.
  • The Appendix D.1 failure with Claude 3.5 suggests a cheap fix the authors do not test: validating candidate plans against the desired inter-robot distance before adoption would filter faulty high-influence plans without changing the decentralized structure.
  • The consensus protocol is agnostic to what generates the plans, so replacing the LLM with any stochastic or rule-based planner should reproduce the same convergence dynamics; if true, the paper's contribution is the negotiation protocol rather than the language model.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The paper proposes LLM-Flock, a decentralized multi-robot formation control framework in which each robot uses a large language model to generate a candidate global formation plan and then iteratively negotiates plan adoption through an influence-based consensus protocol. The influence score of a robot is its degree centrality (Eq. 1), and a robot adopts a neighbor's plan when that neighbor has a higher influence score and a different plan (Algorithm 1). Motion execution is also LLM-driven, with each robot querying its model for the next waypoint toward its assigned goal. The framework is evaluated in simulation across five LLM backends and several formation geometries, with Procrustes shape error as the quantitative metric, and in physical experiments with Crazyflie drones. The central claim is that the influence-based consensus protocol prevents the collapse-to-centroid and divergence observed in unstructured LLM-only planning, yielding stable convergence to the desired formations.

Significance. If the central claim is established, the paper makes a useful empirical contribution: it shows that a simple, communication-based plan-selection rule can reduce the failure modes of independent LLM planners in multi-robot formation tasks. The strengths of the paper are its reproducible setting (code and video are provided), the breadth of LLM backends considered, the inclusion of real-robot Crazyflie experiments, and the honest discussion of failure cases in Appendix D. However, the significance is conditional because the consensus protocol selects plans by communication centrality rather than by geometric validity, so the protocol guarantees plan agreement but not formation correctness. In addition, the quantitative evidence shows that weaker models still produce high residual errors with wide confidence intervals, and the abstract's claim of improvement over 'previous LLM-based methods' is not tested against any prior method. The work is a reasonable step, but the claims need to be reframed or the protocol needs to be strengthened.

major comments (5)
  1. [Section 3.2, Eq. (1), Algorithm 1, Appendix D.1] The adoption rule selects the plan of the neighbor with the largest degree centrality and contains no geometric validity check or plan-quality term. Consequently, the protocol guarantees convergence to a common plan, not to a correct formation. This distinction is not merely theoretical: Appendix D.1 shows a faulty Claude 3.5 plan propagating to the entire team (Fig. 13(b)), and the failure is attributed to the influence score being orthogonal to plan correctness. The abstract's claim that consensus 'drives the system toward a coherent and stable flocking formation' and Section 4.2's claim of 'stable convergence across all tested language models' therefore overstate what the protocol establishes. The claims should be reformulated as convergence in plan agreement, or the protocol should incorporate a mechanism that rejects plans violating the formation constraints.
  2. [Section 3.2, Eq. (2), Algorithm 1, lines 8-10] The goal-assignment procedure is described as a decentralized greedy rule, but as written each robot constructs its own available set from its local copy of the adopted plan and removes only its own choice. Nothing in Algorithm 1 synchronizes the 'available' sets or prevents two robots from selecting the same waypoint. The text states that 'once a goal position is assigned to a robot, it becomes unavailable for others,' but no communication or tie-breaking rule in the protocol implements this. The manuscript should specify the claim/coordination mechanism for goal assignment, or prove that the nearest-available rule is conflict-free under the assumed communication model; otherwise the reported formations may not be reproducible from the protocol alone.
  3. [Section 4.1, Eq. (2)] The baseline comparison confounds the influence-based consensus protocol with the goal-assignment rule. In the baseline, each robot follows its initially assigned LLM goal, while in LLM-Flock robots reassign goals by nearest-available-goal selection after adopting a plan (Eq. 2). The improved performance reported in Fig. 2 and Fig. 11 could therefore be partly due to the goal-reassignment mechanism rather than to the consensus protocol itself. To support the attribution, the baseline should use the same nearest-available-goal assignment on the initial plan, or the experiments should include an ablation that changes only the adoption rule while holding the goal-assignment mechanism fixed.
  4. [Abstract, Section 4] The abstract claims 'notable improvements ... over previous LLM-based methods,' but no comparison to prior LLM-based multi-robot coordination methods (e.g., RoCo, SMART-LLM, ZeroCAP) is reported anywhere in the paper. The only comparative experiment is against the no-consensus baseline using the same LLM backend. Either the authors should add direct comparisons to at least one prior LLM-based formation-control method, or the abstract and conclusion should be revised to limit the claim to improvement over the unstructured-LLM baseline.
  5. [Section 4.2, Fig. 11, Appendix D.2] The sentence 'LLM-Flock enables stable convergence across all tested language models and team sizes' is not supported by the quantitative data. In Fig. 11(e), after 19 timesteps, Claude 3.5 and Qwen-Max show mean Procrustes errors well above 50 with wide 95% confidence intervals for the 10-robot circle, and Appendix D.2 documents a Qwen-Max execution failure in a triangle task where the robot oscillates without stabilizing despite full plan consensus. The claim should be qualified to the strongest reasoning models, or the quantitative threshold for 'stable convergence' should be defined and shown to be met by all models.
minor comments (5)
  1. [Section 4.2, Eq. (3), Appendix C] The notation in Eq. (3) is confusing: it refers to 'xi(t) and yi denote the aligned positions of robot i and actual target formation, respectively,' while Appendix C defines the Procrustes error through R*(xi - xbar) - (yi - ybar). Please make the notation consistent and define the correspondence between robot indices and target points.
  2. [Section 4.2 and Appendix B] The text says 'More results are included in Appendix C,' but the extended quantitative results appear in Appendix B.2, while Appendix C defines the Procrustes metric. Please correct the cross-reference.
  3. [Algorithm 1] The goal-assignment step appears only inside the branch where a robot adopts a new plan. Please specify whether a robot that retains its current plan also recomputes its goal, and how the 'available' set is initialized in that case.
  4. [Appendix A.2] The extracted plan in the text lists '[47.11, 56]' as Vertex 1, but the LLM output above it says '[47.11, 55]'. Please correct the transcription.
  5. [General] The paper uses 'flocking' to describe the task of converging to a static geometric formation. Classical flocking typically involves ongoing motion with separation, alignment, and cohesion terms. Please clarify that the task is static formation acquisition rather than dynamic flocking, or justify the terminology.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation found; the consensus rule is evaluated by an external geometric metric, and the paper explicitly acknowledges dependence on LLM plan quality.

full rationale

The paper reports an algorithmic framework rather than a fitted derivation, so there is no fitted parameter that is later renamed as a prediction. The influence score (Eq. 1) is degree centrality, and Algorithm 1's adoption rule propagates the plan of the most-connected neighbor; this guarantees plan agreement by construction, but plan agreement is not the quantity used to support the central claim. The reported success metric is Procrustes shape error (Eq. 3) against the ideal geometric formation, which is external to the selection rule and can remain large even after consensus. The paper's own Appendix D.1 demonstrates exactly this: a faulty Claude 3.5 plan 'ends up propagating to the entire team, and the final formation does not represent an ideal uniformly spaced circular shape.' Section 6 similarly states that 'the overall performance of LLM-Flock inherently depends on the zero-shot planning accuracy of the underlying language model.' These admissions show that the consensus mechanism is not a self-fulfilling definition of formation quality. The evaluation is also external to the method's construction: it compares against a no-consensus baseline (Figs. 2 and 11), across multiple LLMs, and on physical Crazyflie drones. The only self-citations ([13], [24]) support the motivation that unstructured LLM planners fail, but the same failure is reproduced within this paper's own baseline experiments, so those citations are not load-bearing. No uniqueness theorem, fitted parameter, or ansatz is imported from prior work, and no known result is merely renamed. Accordingly, the derivation chain is self-contained and no circular step is present.

Assumptions & free parameters 0 free parameters · 4 assumptions · 0 invented entities

The framework introduces no new physical entities and has no fitted constants. The load-bearing assumptions are that LLM outputs are parseable and geometrically plausible, that degree-based influence is a useful selector of plans, and that the communication graph stays sufficiently connected. These are stated in the method sections but not proven.

assumptions (4)
  • domain assumption LLM outputs are parseable and geometrically plausible formation plans.
    The consensus step selects among LLM plans, so if no plan is a valid formation the protocol cannot fix it. Appendix D.1 documents a faulty plan.
  • ad hoc to paper Degree-based influence (|N_i|/N) is a useful proxy for plan quality.
    Eq. (1) defines influence purely by neighbor count; the paper offers intuition but no justification that high-degree robots generate better plans.
  • domain assumption The communication graph remains connected enough for adoption to propagate to all robots.
    Consensus requires a path between any robot and the eventual leader; rcomm is fixed and connectivity is not analyzed.
  • domain assumption Nearest-available-goal greedy assignment produces a coherent team motion.
    Eq. (2) assigns goals greedily; the paper states this 'ensures a smooth transition' but provides no analysis of matching conflicts.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LLM-Flock: Decentralized Multi-Robot Flocking via Large Language Models and Influence-Based Consensus." pith.science (2026). https://pith.science/paper/MN7TT7VC

@misc{pith2026250506513,
  author       = {Pith},
  title        = {Pith review of: LLM-Flock: Decentralized Multi-Robot Flocking via Large Language Models and Influence-Based Consensus},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MN7TT7VC}},
  note         = {Machine review of arXiv:2505.06513}
}
read the original abstract

Large Language Models (LLMs) have advanced rapidly in recent years, demonstrating strong capabilities in problem comprehension and reasoning. Inspired by these developments, researchers have begun exploring the use of LLMs as decentralized decision-makers for multi-robot formation control. However, prior studies reveal that directly applying LLMs to such tasks often leads to unstable and inconsistent behaviors, where robots may collapse to the centroid of their positions or diverge entirely due to hallucinated reasoning, logical inconsistencies, and limited coordination awareness. To overcome these limitations, we propose a novel framework that integrates LLMs with an influence-based plan consensus protocol. In this framework, each robot independently generates a local plan toward the desired formation using its own LLM. The robots then iteratively refine their plans through a decentralized consensus protocol that accounts for their influence on neighboring robots. This process drives the system toward a coherent and stable flocking formation in a fully decentralized manner. We evaluate our approach through comprehensive simulations involving both state-of-the-art closed-source LLMs (e.g., o3-mini, Claude 3.5) and open-source models (e.g., Llama3.1-405b, Qwen-Max, DeepSeek-R1). The results show notable improvements in stability, convergence, and adaptability over previous LLM-based methods. We further validate our framework on a physical team of Crazyflie drones, demonstrating its practical viability and effectiveness in real-world multi-robot systems.

Figures

Figures reproduced from arXiv: 2505.06513 by the authors.

Figure 1
Figure 1. LLM-Flock framework for decentralized multi-robot flocking. Each robot first uses an [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Comparison of formation results with and without influence-based consensus. (a)-(d) [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Qualitative results demonstrating LLM-Flock on diverse formation tasks. (a)-(d) show the [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (11 more)
Figure 4
Figure 4. Figure 4: Procrustes shape error over time across different formation tasks and LLM backends. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Snapshots of five Crazyflie drones forming a circle using proposed LLM-Flock framework. [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: Snapshots of simulation with three robots forming an equilateral triangle. The number on [PITH_FULL_IMAGE:figures/full_fig_p013_6.png]
Figure 7
Figure 7. Figure 7: Three robots forming a triangle using OpenAI o3-mini. (a)–(d): With influence-based con [PITH_FULL_IMAGE:figures/full_fig_p016_7.png]
Figure 8
Figure 8. Figure 8: Six robots forming a triangle using OpenAI o3-mini. (a)–(d): LLM-Flock enables conver [PITH_FULL_IMAGE:figures/full_fig_p016_8.png]
Figure 9
Figure 9. Figure 9: Eight robots forming a square. (a)–(d): Influence-based plan consensus leads to uniform [PITH_FULL_IMAGE:figures/full_fig_p017_9.png]
Figure 10
Figure 10. Figure 10: Ten robots forming a circle. (a)–(d): Robots converge to evenly spaced positions around [PITH_FULL_IMAGE:figures/full_fig_p017_10.png]
Figure 11
Figure 11. Figure 11: Procrustes shape error over time across different formation tasks and LLM backends. [PITH_FULL_IMAGE:figures/full_fig_p018_11.png]
Figure 12
Figure 12. Figure 12: Real drone experiments using proposed LLM-Flock framework with o3-mini as LLM [PITH_FULL_IMAGE:figures/full_fig_p020_12.png]
Figure 13
Figure 13. Figure 13: Comparison of formation plans generated by Claude 3.5 Sonnet and ideal formation for [PITH_FULL_IMAGE:figures/full_fig_p022_13.png]
Figure 14
Figure 14. Figure 14: Snapshots from a simulation in which the LLM fails to execute correct motion behavior. [PITH_FULL_IMAGE:figures/full_fig_p023_14.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

43 extracted references · 27 canonical work pages

  1. [1]

    Shijie, S

    L. Shijie, S. Yefeng, G. Liang, and Z. Kai. Trajectory tracking method of agricultural machin- ery multi-robot formation operation based on mpc delay compensator. Smart Agriculture, 6 (3):69, 2024

    work page 2024

  2. [2]

    L. Peng, F. Guan, L. Perneel, H. Fayyad-Kazan, and M. Timmerman. Decentralized multi- robot formation control with communication delay and asynchronous clock. Journal of Intel- ligent & Robotic Systems , 89:465–484, 2018

    work page 2018

  3. [3]

    Konda, H

    R. Konda, H. M. La, and J. Zhang. Decentralized function approximated q-learning in multi- robot systems for predator avoidance.IEEE Robotics and Automation Letters, 5(4):6342–6349, 2020

    work page 2020

  4. [4]

    Bouraqadi, A

    N. Bouraqadi, A. Doniec, and E. de Douai. Flocking-based multi-robot exploration. In Na- tional conference on control architectures of robots, pages 23–24, 2009

    work page 2009

  5. [5]

    Notomista, C

    G. Notomista, C. Pacchierotti, and P. R. Giordano. Multi-robot persistent environmental mon- itoring based on constraint-driven execution of learned robot tasks. In 2022 International Conference on Robotics and Automation (ICRA) , pages 6853–6859. IEEE, 2022

    work page 2022

  6. [6]

    K.-C. Ma, Z. Ma, L. Liu, and G. S. Sukhatme. Multi-robot informative and adaptive planning for persistent environmental monitoring. In Distributed Autonomous Robotic Systems: The 13th International Symposium, pages 285–298. Springer, 2018

    work page 2018

  7. [7]

    Burgard, M

    W. Burgard, M. Moors, C. Stachniss, and F. E. Schneider. Coordinated multi-robot exploration. IEEE Transactions on robotics, 21(3):376–386, 2005

    work page 2005

  8. [8]

    Olfati-Saber

    R. Olfati-Saber. Flocking for multi-agent dynamic systems: Algorithms and theory. IEEE Transactions on automatic control, 51(3):401–420, 2006

    work page 2006

Show all 43 references
  1. [9]

    S. Shao, J. Zhang, T. Wang, A. Shankar, and C. Maple. Dynamic obstacle-avoidance algorithm for multi-robot flocking based on improved artificial potential field. IEEE Transactions on Consumer Electronics, 70(1):4388–4399, 2023

    2023

  2. [10]

    De Denus, J

    M. De Denus, J. Anderson, and J. Baltes. Heuristic formation control in multi-robot systems using local communication and limited identification. In RoboCup 2009: Robot Soccer World Cup XIII 13, pages 437–448. Springer, 2010

    2009

  3. [11]

    Antonelli, F

    G. Antonelli, F. Arrichiello, F. Caccavale, and A. Marino. Decentralized time-varying forma- tion control for multi-robot systems. The International Journal of Robotics Research , 33(7): 1029–1043, 2014

    2014

  4. [12]

    Tao, T.-E

    Z. Tao, T.-E. Lin, X. Chen, H. Li, Y . Wu, Y . Li, Z. Jin, F. Huang, D. Tao, and J. Zhou. A survey on self-evolution of large language models. arXiv preprint arXiv:2404.14387, 2024

    2024 arXiv

  5. [13]

    P. Li, Z. An, S. Abrar, and L. Zhou. Large language models for multi-robot systems: A survey. arXiv preprint arXiv:2502.03814, 2025

    2025 arXiv

  6. [14]

    Mandi, S

    Z. Mandi, S. Jain, and S. Song. Roco: Dialectic multi-robot collaboration with large language models. In 2024 IEEE International Conference on Robotics and Automation (ICRA) , pages 286–299. IEEE, 2024

    2024

  7. [15]

    Y . Wang, R. Xiao, J. Y . L. Kasahara, R. Yajima, K. Nagatani, A. Yamashita, and H. Asama. Dart-llm: Dependency-aware multi-robot task decomposition and execution using large lan- guage models. arXiv preprint arXiv:2411.09022, 2024

    2024 arXiv

  8. [16]

    Y . Wu, Y . Tao, P. Li, G. Shi, G. S. Sukhatmem, V . Kumar, and L. Zhou. Hierarchical LLMs In-the-loop Optimization for Real-time Multi-Robot Target Tracking under Unknown Hazards, Sept. 2024. URL http://arxiv.org/abs/2409.12274. arXiv:2409.12274. 10

    2024

  9. [17]

    Achiam, S

    J. Achiam, S. Adler, S. Agarwal, L. Ahmad, I. Akkaya, F. L. Aleman, D. Almeida, J. Altenschmidt, S. Altman, S. Anadkat, et al. Gpt-4 technical report. arXiv preprint arXiv:2303.08774, 2023

    2023 arXiv

  10. [18]

    Grattafiori, A

    A. Grattafiori, A. Dubey, A. Jauhri, A. Pandey, A. Kadian, A. Al-Dahle, A. Letman, A. Mathur, A. Schelten, A. Vaughan, et al. The llama 3 herd of models. arXiv preprint arXiv:2407.21783, 2024

    2024 arXiv

  11. [19]

    W. Hunt, S. D. Ramchurn, and M. D. Soorati. A Survey of Language-Based Communication in Robotics. arXiv preprint arXiv:2406.04086 , Sept. 2024. doi:10.48550/arXiv.2406.04086. URL http://arxiv.org/abs/2406.04086

    2024 doi arXiv

  12. [20]

    Y . Kim, D. Kim, J. Choi, J. Park, N. Oh, and D. Park. A Survey on Integration of Large Language Models with Intelligent Robots. arXiv preprint arXiv:2404.09228, Aug. 2024. doi: 10.48550/arXiv.2404.09228. URL http://arxiv.org/abs/2404.09228

    2024 doi arXiv

  13. [21]

    T. Guo, X. Chen, Y . Wang, R. Chang, S. Pei, N. V . Chawla, O. Wiest, and X. Zhang. Large Language Model based Multi-Agents: A Survey of Progress and Challenges. arXiv preprint arXiv:2402.01680, Apr. 2024. doi:10.48550/arXiv.2402.01680. URL http://arxiv.org/ abs/2402.01680

    2024 doi arXiv

  14. [22]

    Huang, W

    L. Huang, W. Yu, W. Ma, W. Zhong, Z. Feng, H. Wang, Q. Chen, W. Peng, X. Feng, B. Qin, et al. A survey on hallucination in large language models: Principles, taxonomy, challenges, and open questions. ACM Transactions on Information Systems, 2023

    2023

  15. [23]

    McDonald, R

    D. McDonald, R. Papadopoulos, and L. Benningfield. Reducing llm hallucination using knowl- edge distillation: A case study with mistral large and mmlu benchmark. Authorea Preprints, 2024

    2024

  16. [24]

    P. Li, V . Menon, B. Gudiguntla, D. Ting, and L. Zhou. Challenges faced by large language models in solving multi-agent flocking. arXiv preprint arXiv:2404.04752, 2024

    2024 arXiv

  17. [25]

    W. Chen, S. Koenig, and B. Dilkina. Why solving multi-agent path finding with large language model has not succeeded yet. arXiv preprint arXiv:2401.03630, 2024

    2024 arXiv

  18. [26]

    C. W. Reynolds. Flocks, herds and schools: A distributed behavioral model. In Proceedings of the 14th annual conference on Computer graphics and interactive techniques , pages 25–34, 1987

    1987

  19. [27]

    Blumenkamp, S

    J. Blumenkamp, S. Morad, J. Gielis, Q. Li, and A. Prorok. A framework for real-world multi- robot systems running decentralized gnn-based policies. In 2022 International Conference on Robotics and Automation (ICRA) , pages 8772–8778. IEEE, 2022

    2022

  20. [28]

    Goarin and G

    M. Goarin and G. Loianno. Graph neural network for decentralized multi-robot goal assign- ment. IEEE Robotics and Automation Letters , 2024

    2024

  21. [29]

    R. Lowe, Y . I. Wu, A. Tamar, J. Harb, O. Pieter Abbeel, and I. Mordatch. Multi-agent actor- critic for mixed cooperative-competitive environments. Advances in neural information pro- cessing systems, 30, 2017

    2017

  22. [30]

    Rashid, M

    T. Rashid, M. Samvelyan, C. S. De Witt, G. Farquhar, J. Foerster, and S. Whiteson. Monotonic value function factorisation for deep multi-agent reinforcement learning. Journal of Machine Learning Research, 21(178):1–51, 2020

    2020

  23. [31]

    S. Chen, Y . Sun, P. Li, L. Zhou, and C.-T. Lu. Spatial temporal graph neural networks for de- centralized control of robot swarms. In Proceedings of the 31st ACM International Conference on Advances in Geographic Information Systems , pages 1–4, 2023. 11

    2023

  24. [32]

    S. Chen, Y . Sun, P. Li, L. Zhou, and C.-T. Lu. Learning decentralized flocking controllers with spatio-temporal graph neural network. In 2024 IEEE International Conference on Robotics and Automation (ICRA), pages 2596–2602. IEEE, 2024

    2024

  25. [33]

    Tolstaya, F

    E. Tolstaya, F. Gama, J. Paulos, G. Pappas, V . Kumar, and A. Ribeiro. Learning decentralized controllers for robot swarms with graph neural networks. In Conference on robot learning , pages 671–682. PMLR, 2020

    2020

  26. [34]

    A. Khan, E. Tolstaya, A. Ribeiro, and V . Kumar. Graph policy gradients for large scale robot control. In Conference on robot learning, pages 823–834. PMLR, 2020

    2020

  27. [35]

    S. S. Kannan, V . L. Venkatesh, and B.-C. Min. Smart-llm: Smart multi-agent robot task plan- ning using large language models. In 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 12140–12147. IEEE, 2024

    2024

  28. [36]

    V . L. N. Venkatesh and B.-C. Min. ZeroCAP: Zero-Shot Multi-Robot Context Aware Pattern Formation via Large Language Models, Sept. 2024. URL http://arxiv.org/abs/2404. 02318. arXiv:2404.02318

    2024 arXiv

  29. [37]

    https://www.bitcraze.io/

    Bitcraze crazyflie. https://www.bitcraze.io/. Accessed: 2025-04-15

    2025

  30. [38]

    J. C. Gower and G. B. Dijksterhuis. Procrustes problems, volume 30. Oxford University Press, USA, 2004

    2004

  31. [39]

    D. G. Kendall. A survey of the statistical theory of shape. Statistical Science , 4(2):87–99, 1989

    1989

  32. [40]

    Openai o3-mini, January 2025

    OpenAI. Openai o3-mini, January 2025. URL https://openai.com/index/ openai-o3-mini/. Accessed April 30, 2025

    2025

  33. [41]

    Claude 3.5 sonnet, October 2024

    Anthropic. Claude 3.5 sonnet, October 2024. URL https://www.anthropic.com/news/ claude-3-5-sonnet . Accessed April 30, 2025

    2024

  34. [42]

    D. Guo, D. Yang, H. Zhang, J. Song, R. Zhang, R. Xu, Q. Zhu, S. Ma, P. Wang, X. Bi, et al. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948, 2025

    2025 arXiv

  35. [43]

    Position: [x, y]

    A. Yang, B. Yang, B. Zhang, B. Hui, B. Zheng, B. Yu, C. Li, D. Liu, F. Huang, H. Wei, et al. Qwen2. 5 technical report. arXiv preprint arXiv:2412.15115, 2024. 12 A LLM Prompting Details To illustrate how robots in LLM-Flock interact with language models during decentralized fo...

    2024 arXiv

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.