arxiv: 2604.11353 · v1 · submitted 2026-04-13 · 📡 eess.SY · cs.SY

Leader-Follower Density Control of Multi-Agent Systems with Interacting Followers: Feasibility and Convergence Analysis

Beniamino Di Lorenzo , Gian Carlo Maffettone , Mario di Bernardo This is my paper

Pith reviewed 2026-05-10 16:01 UTC · model grok-4.3

classification 📡 eess.SY cs.SY

keywords leader-follower systemsdensity controlmulti-agent systemsPDE modelsfeasibility conditionsinteraction kernelsfeedback controlstability analysis

0 comments

The pith

Necessary and sufficient conditions determine when leaders can steer interacting followers to a target density, with a feedback law ensuring local stability.

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

The paper shows that follower-follower interactions impose strict limits on which spatial density distributions can be achieved by controllable leaders in large multi-agent groups. It derives exact feasibility conditions that relate any chosen target distribution to the strength of those interactions, diffusion, and the total mass of leaders. A feedback control law is constructed that drives the follower density to the target from nearby initial states, with an explicit bound on the basin where this works. This matters because it exposes fundamental phase-transition thresholds beyond which no amount of leader effort succeeds, unlike simpler models that ignore follower interactions.

Core claim

We derive necessary and sufficient feasibility conditions linking the target distribution to interaction strength, diffusion, and leader mass. We also construct a feedback control law that guarantees local stability of the closed-loop density dynamics together with an explicit estimate of the basin of attraction. The analysis identifies sharp feasibility thresholds and phase transitions beyond which the desired configuration cannot be reached regardless of control effort.

What carries the argument

The macroscopic PDE model for follower density evolution that includes an interaction kernel for follower-follower dynamics, together with the feedback control law that acts through the leaders.

If this is right

Target distributions become unreachable once interaction strength exceeds thresholds set by diffusion and leader mass.
The feedback law drives the density to the target from all initial conditions inside the explicit basin.
Phase transitions appear in controllability as interaction parameters vary.
Macroscopic predictions match behavior observed in finite-population agent-based simulations.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

Swarm designers may need to tune leader numbers or modify interaction rules to remain inside the feasible region.
The same threshold analysis could be applied to time-varying targets or to swarms with heterogeneous interaction kernels.
The basin estimate offers a practical test for whether a planned leader trajectory will succeed before deployment.

Load-bearing premise

The continuous PDE description with the chosen interaction kernel accurately represents the collective behavior of finite numbers of agents.

What would settle it

A numerical experiment in which a target density violating the derived feasibility threshold is not reached by any leader input, or the closed-loop density diverges when started outside the stated basin of attraction.

read the original abstract

We address density control problems for large-scale multi-agent systems in leader-follower settings, where a group of controllable leaders must steer a population of followers toward a desired spatial distribution. Unlike prior work, we explicitly account for follower-follower interactions, capturing realistic behaviors such as flocking and collision avoidance. Within a macroscopic framework based on partial differential equations governing the density dynamics, we derive (i) necessary and sufficient feasibility conditions linking the target distribution to interaction strength, diffusion, and leader mass, and (ii) a feedback control law guaranteeing local stability with an explicit estimate of the basin of attraction. Our analysis reveals sharp feasibility thresholds, phase transitions beyond which no control effort can achieve the desired configuration. Numerical simulations in one- and two-dimensional domains validate the theoretical results at the macroscopic level, and agent-based simulations on finite populations confirm the practical deployability of the proposed framework.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Desk Editor's Note private letter to a colleague

This adds follower interactions to density control with feasibility conditions and a local stabilizer, but the finite-N validation is thin.

read the letter

This paper adds follower-follower interactions to the leader-follower density control problem and works out necessary and sufficient feasibility conditions plus a stabilizing feedback law with basin estimate. That's the main new element compared to earlier work that omitted those terms. They build on standard PDE density models but now include the interaction kernel, showing how it affects reachability and creating phase transitions where targets become unreachable. The simulations in one and two dimensions, along with finite-agent runs, support the claims at least at a basic level. It does a decent job of highlighting practical limits on what leaders can achieve when followers interact, like in flocking or avoidance scenarios. The soft spot is the mean-field PDE approximation for finite systems. Fluctuations in actual agent counts could cross the feasibility lines even if the continuous model says otherwise. The paper mentions agent-based simulations but lacks clear N-scaling or approximation error analysis, so the thresholds' reliability for moderate population sizes remains a question. If the derivations include rigorous mean-field convergence, that would address it. Readers in multi-agent control and PDE modeling of collectives would get value from this, especially those wanting to handle interactions explicitly. It deserves serious peer review as it fills a noted gap with analysis and checks, though reviewers will likely want more on the discrete-to-continuous link. I'd recommend putting it through peer review.

Referee Report

3 major / 2 minor

Summary. The paper develops a macroscopic PDE model for density control of a large population of interacting followers steered by a smaller set of controllable leaders. It derives necessary and sufficient feasibility conditions that relate the target density to the strength of follower-follower interactions, diffusion coefficient, and total leader mass; these conditions exhibit sharp thresholds and phase transitions. A feedback control law is constructed that renders the target locally asymptotically stable, with an explicit estimate of the basin of attraction. The theoretical results are illustrated by numerical solutions of the PDE in one and two dimensions and by agent-based simulations of finite populations.

Significance. If the derivations are correct, the work supplies the first set of sharp, parameter-linked feasibility thresholds for interacting leader-follower density control and supplies a stabilizing feedback law together with a basin estimate. These results are directly relevant to applications such as robotic swarms and crowd management where follower interactions cannot be neglected. The combination of analytic conditions, local stability proof, and both continuum and discrete simulations constitutes a solid contribution to the mean-field control literature.

major comments (3)

[§3] §3 (Feasibility analysis): The necessary-and-sufficient conditions are derived under the continuum PDE limit. Because the target application is a finite-N multi-agent system, the manuscript must supply either an N-scaling argument or explicit error bounds between the empirical measure and the PDE solution near the feasibility thresholds; without such bounds the claim that the thresholds remain predictive for practical population sizes is not yet supported.
[§4] §4 (Control design and basin estimate): The local stability proof and basin estimate rely on the linearized PDE around the target. The manuscript should verify that the same feedback law, when applied to the finite-N system, inherits a comparable basin for sufficiently large N; currently only qualitative agent-based trajectories are shown.
[§5] §5 (Numerical validation): The agent-based simulations confirm qualitative agreement but do not report quantitative metrics (e.g., Wasserstein distance or L1 error to the PDE solution) as a function of N or proximity to the derived thresholds. Such metrics are needed to assess how well the PDE-derived feasibility boundaries translate to discrete systems.

minor comments (2)

Notation for the interaction kernel and the leader control term should be introduced once and used consistently; several symbols appear to be redefined in different sections.
The abstract states that the conditions are 'necessary and sufficient,' yet the text should explicitly state the precise function space in which sufficiency holds (e.g., L^1 or H^{-1}).

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the thorough review and valuable suggestions. We have carefully considered each comment and provide detailed responses below. We believe the revisions will strengthen the connection between the theoretical results and practical finite-agent implementations.

read point-by-point responses

Referee: [§3] §3 (Feasibility analysis): The necessary-and-sufficient conditions are derived under the continuum PDE limit. Because the target application is a finite-N multi-agent system, the manuscript must supply either an N-scaling argument or explicit error bounds between the empirical measure and the PDE solution near the feasibility thresholds; without such bounds the claim that the thresholds remain predictive for practical population sizes is not yet supported.

Authors: We agree that providing a link between the PDE-derived thresholds and finite-N systems is essential. In the revised manuscript, we will add a discussion in Section 3 on the mean-field approximation error. Specifically, we will invoke quantitative propagation-of-chaos results for the underlying interacting particle system (citing relevant works on McKean-Vlasov equations), which establish that the Wasserstein distance between the empirical measure and the PDE solution scales as O(N^{-1/2}) for fixed time horizons, provided the interaction kernel is sufficiently regular. Near the feasibility thresholds, we will note that the constants in the error bounds may deteriorate, but the thresholds themselves remain indicative for large but finite N, as supported by our agent-based simulations. This addresses the predictive power for practical population sizes. revision: yes
Referee: [§4] §4 (Control design and basin estimate): The local stability proof and basin estimate rely on the linearized PDE around the target. The manuscript should verify that the same feedback law, when applied to the finite-N system, inherits a comparable basin for sufficiently large N; currently only qualitative agent-based trajectories are shown.

Authors: The stability result is derived for the infinite-population PDE limit, which is the appropriate macroscopic model for large-scale systems. For finite N, the agent-based simulations in the current manuscript already illustrate that the proposed feedback law successfully steers the population toward the target for the tested values of N. To further support the inheritance of the basin of attraction, we will include in the revision a theoretical remark based on the fact that the finite-N dynamics converge to the PDE dynamics in the mean-field limit. Consequently, for sufficiently large N, the local asymptotic stability and the estimated basin carry over to the discrete system up to a small perturbation. We will also enhance the numerical section with additional simulations starting from initial conditions near the boundary of the estimated basin to demonstrate robustness. revision: partial
Referee: [§5] §5 (Numerical validation): The agent-based simulations confirm qualitative agreement but do not report quantitative metrics (e.g., Wasserstein distance or L1 error to the PDE solution) as a function of N or proximity to the derived thresholds. Such metrics are needed to assess how well the PDE-derived feasibility boundaries translate to discrete systems.

Authors: We concur that quantitative metrics would provide stronger evidence of the approximation quality. In the revised manuscript, we will augment Section 5 with quantitative comparisons. Specifically, we will compute the L^1 norm and the 1-Wasserstein distance between the binned empirical density from the agent-based simulations and the PDE solution, for increasing values of N (e.g., 50, 200, 1000) and for parameter regimes both well inside and close to the feasibility boundaries. These results will be presented in new figures and tables, confirming the convergence to the continuum predictions as N grows and highlighting any sensitivity near the thresholds. revision: yes

Circularity Check

0 steps flagged

No significant circularity; derivations start from standard PDE models

full rationale

The paper's central claims derive necessary-and-sufficient feasibility conditions and a stabilizing feedback law directly from the macroscopic PDE density model with interaction kernels. These steps use standard mean-field analysis, Lyapunov methods, and explicit basin estimates without reducing to self-definitional loops, fitted inputs renamed as predictions, or load-bearing self-citations. The interaction kernel and PDE are taken as given modeling assumptions (not derived within the paper), and numerical validation is presented separately from the analytic thresholds. No step equates a claimed prediction to its own input by construction.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The framework rests on the standard assumption that follower density obeys a PDE with diffusion and interaction terms; no free parameters or new entities are introduced in the abstract.

axioms (1)

domain assumption Follower density evolves according to a PDE that includes diffusion and follower-follower interaction terms.
This is the macroscopic modeling choice that enables the feasibility and stability analysis.

pith-pipeline@v0.9.0 · 5459 in / 1171 out tokens · 77986 ms · 2026-05-10T16:01:33.796756+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

38 extracted references · 38 canonical work pages · 1 internal anchor

[1]

Freeway traffic control: A survey,

S. Siri, C. Pasquale, S. Sacone, and A. Ferrara, “Freeway traffic control: A survey,”Automatica, vol. 130, p. 109655, 2021

work page 2021
[2]

Review of road traffic control strategies,

M. Papageorgiou, C. Diakaki, V . Dinopoulou, A. Kotsialos, and Y . Wang, “Review of road traffic control strategies,”Proceedings of the IEEE, vol. 91, no. 12, pp. 2043–2067, 2003

work page 2043
[3]

Robust optimal density control of robotic swarms,

C. Sinigaglia, A. Manzoni, F. Braghin, and S. Berman, “Robust optimal density control of robotic swarms,”Automatica, vol. 176, p. 112218, 2025

work page 2025
[4]

Swarm robotics: Past, present, and future [point of view],

M. Dorigo, G. Theraulaz, and V . Trianni, “Swarm robotics: Past, present, and future [point of view],”Proceedings of the IEEE, vol. 109, no. 7, pp. 1152–1165, 2021

work page 2021
[5]

Swarm robotics: a review from the swarm engineering perspective,

M. Brambilla, E. Ferrante, M. Birattari, and M. Dorigo, “Swarm robotics: a review from the swarm engineering perspective,”Swarm Intelligence, vol. 7, no. 1, pp. 1–41, 2013

work page 2013
[6]

Multi-robot-assisted human crowd control for emergency evacuation: A stabilization approach,

Z. Yuan, T. Zheng, M. Nayyar, A. R. Wagner, H. Lin, and M. Zhu, “Multi-robot-assisted human crowd control for emergency evacuation: A stabilization approach,” in2023 American Control Conference (ACC), 2023

work page 2023
[7]

Simulating dynamical features of escape panic,

D. Helbing, I. Farkas, and T. Vicsek, “Simulating dynamical features of escape panic,”Nature, vol. 407, no. 6803, pp. 487–490, 2000

work page 2000
[8]

Rectification and confinement of photokinetic bacteria in an optical feedback loop,

H. Massana-Cid, C. Maggi, G. Frangipane, and R. Di Leonardo, “Rectification and confinement of photokinetic bacteria in an optical feedback loop,”Nature Communications, vol. 13, no. 1, p. 2740, 2022

work page 2022
[9]

Data- driven inference of digital twins for high-throughput phenotyping of motile and light-responsive microorganisms,

A. Giusti, D. Salzano, M. di Bernardo, and T. E. Gorochowski, “Data- driven inference of digital twins for high-throughput phenotyping of motile and light-responsive microorganisms,”Journal of The Royal Society Interface, vol. 23, no. 234, 2026

work page 2026
[10]

Dis- sipation of stop-and-go waves via control of autonomous vehicles: Field experiments,

R. E. Stern, S. Cui, M. L. Delle Monache, R. Bhadani, M. Bunting, M. Churchill, N. Hamilton, R. Haulcy, H. Pohlmann, F. Wuet al., “Dis- sipation of stop-and-go waves via control of autonomous vehicles: Field experiments,”Transportation research part C: emerging technologies, vol. 89, pp. 205–221, 2018

work page 2018
[11]

Controlling noncooperative herds with robotic herders,

A. Pierson and M. Schwager, “Controlling noncooperative herds with robotic herders,”IEEE Transactions on Robotics, vol. 34, no. 2, pp. 517–525, 2017

work page 2017
[12]

Optimal control problems in transport dynamics with additive noise,

S. Almi, M. Morandotti, and F. Solombrino, “Optimal control problems in transport dynamics with additive noise,”Journal of Differential Equations, vol. 373, pp. 1–47, 2023

work page 2023
[13]

Leader–follower density control of spatial dynamics in large-scale multiagent systems,

G. C. Maffettone, A. Boldini, M. Porfiri, and M. d. Bernardo, “Leader–follower density control of spatial dynamics in large-scale multiagent systems,”IEEE Transactions on Automatic Control, vol. 70, no. 10, pp. 6783–6798, 2025

work page 2025
[14]

Shepherding and herdability in com- plex multiagent systems,

A. Lama and M. di Bernardo, “Shepherding and herdability in com- plex multiagent systems,”Physical Review Research, vol. 6, no. 3, p. L032012, 2024

work page 2024
[15]

A continuification-based control solution for large-scale shepherding,

B. Di Lorenzo, G. C. Maffettone, and M. di Bernardo, “A continuification-based control solution for large-scale shepherding,”Eu- ropean Journal of Control, p. 101324, 2025

work page 2025
[16]

Solving the shepherding problem: heuristics for herding autonomous, interacting agents,

D. Str ¨ombom, R. P. Mann, A. M. Wilson, S. Hailes, A. J. Morton, D. J. Sumpter, and A. J. King, “Solving the shepherding problem: heuristics for herding autonomous, interacting agents,”Journal of the royal society interface, vol. 11, no. 100, p. 20140719, 2014

work page 2014
[17]

Optimal transport for time-varying multi-agent coverage control,

I. Napolitano and M. di Bernardo, “Optimal transport for time-varying multi-agent coverage control,”arXiv preprint arXiv:2601.21753, 2026

work page arXiv 2026
[18]

Mixed reality environment and high-dimensional continuification control for swarm robotics,

G. C. Maffettone, L. Liguori, E. Palermo, M. Di Bernardo, and M. Por- firi, “Mixed reality environment and high-dimensional continuification control for swarm robotics,”IEEE Transactions on Control Systems Technology, 2024

work page 2024
[19]

Sparse shepherding control of large-scale multi-agent systems via Reinforcement Learning

L. Catello, I. Napolitano, D. Salzano, and M. di Bernardo, “Sparse shepherding control of large-scale multi-agent systems via reinforcement learning,”arXiv preprint arXiv:2511.21304, 2025

work page internal anchor Pith review Pith/arXiv arXiv 2025
[20]

Mean-field optimal control,

M. Fornasier and F. Solombrino, “Mean-field optimal control,”ESAIM: Control, Optimisation and Calculus of Variations, vol. 20, no. 4, pp. 1123–1152, 2014

work page 2014
[21]

Mean-field optimal control by leaders,

M. Fornasier, B. Piccoli, N. P. Duteil, and F. Rossi, “Mean-field optimal control by leaders,”Proceedings of the 53rd IEEE Conference on Decision and Control, pp. 6957–6962, 2014

work page 2014
[22]

Optimal control problems in transport dynamics,

M. Bongini and G. Buttazzo, “Optimal control problems in transport dynamics,”Mathematical Models and Methods in Applied Sciences, vol. 27, no. 03, pp. 427–451, 2017

work page 2017
[23]

Mean-field selective optimal control via transient leadership,

G. Albi, S. Almi, M. Morandotti, and F. Solombrino, “Mean-field selective optimal control via transient leadership,”Applied Mathematics & Optimization, vol. 85, no. 2, p. 22, 2022

work page 2022
[24]

Kinetic description of swarming dynamics with topological interaction and transient leaders,

G. Albi and F. Ferrarese, “Kinetic description of swarming dynamics with topological interaction and transient leaders,”Multiscale Modeling & Simulation, vol. 22, no. 3, pp. 1169–1195, 2024

work page 2024
[25]

A. Lama, M. di Bernardo, and S. H. Klapp, “Nonreciprocal field DI LORENZOet al.: LEADER-FOLLOWER DENSITY CONTROL OF MUL TI-AGENT SYSTEMS WITH INTERACTING FOLLOWERS 15 theory for decision-making in multi-agent control systems,”Nature Communications, vol. 16, no. 1, p. 8450, 2025

work page 2025
[26]

Leadership Through Influ- ence: What Mechanisms Allow Leaders to Steer a Swarm?

S. Bernardi, R. Eftimie, and K. J. Painter, “Leadership Through Influ- ence: What Mechanisms Allow Leaders to Steer a Swarm?”Bulletin of Mathematical Biology, vol. 83, no. 6, p. 69, Jun. 2021

work page 2021
[27]

Macroscopic descriptions of follower-leader systems,

S. Bernardi, G. Estrada-Rodriguez, H. Gimperlein, and K. J. Painter, “Macroscopic descriptions of follower-leader systems,” Kinetic and Related Models, vol. 14, no. 6, pp. 981– 1002, 2021. [Online]. Available: https://www.aimsciences.org/article/id/ d8004c4d-a77c-48ce-b44c-66d8d44c2c53

work page 2021
[28]

A primer of swarm equilibria,

A. J. Bernoff and C. M. Topaz, “A primer of swarm equilibria,”SIAM Journal on Applied Dynamical Systems, vol. 10, no. 1, pp. 212–250, 2011

work page 2011
[29]

Derivation of macroscopic equations for individual cell-based models: a formal approach,

M. Bodnar and J. J. L. Velazquez, “Derivation of macroscopic equations for individual cell-based models: a formal approach,”Mathematical methods in the applied sciences, vol. 28, no. 15, pp. 1757–1779, 2005

work page 2005
[30]

Stigmergy: from mathematical modelling to control,

A. Boldini, M. Civitella, and M. Porfiri, “Stigmergy: from mathematical modelling to control,”Royal Society Open Science, vol. 11, no. 9, p. 240845, 2024

work page 2024
[31]

H. L. Royden and P. Fitzpatrick,Real analysis. Macmillan New York, 1988, vol. 32

work page 1988
[32]

On information and sufficiency,

S. Kullback and R. A. Leibler, “On information and sufficiency,”The annals of mathematical statistics, vol. 22, no. 1, pp. 79–86, 1951

work page 1951
[33]

H. K. Khalil,Nonlinear systems. Prentice Hall, 2002

work page 2002
[34]

Decentralized continuification control of multi-agent systems via distributed density estimation,

B. Di Lorenzo, G. C. Maffettone, and M. di Bernardo, “Decentralized continuification control of multi-agent systems via distributed density estimation,”IEEE Control Systems Letters, vol. 9, pp. 1580–1585, 2025

work page 2025
[35]

Con- tinuification control of large-scale multiagent systems in a ring,

G. C. Maffettone, A. Boldini, M. Di Bernardo, and M. Porfiri, “Con- tinuification control of large-scale multiagent systems in a ring,”IEEE Control Systems Letters, vol. 7, pp. 841–846, 2022

work page 2022
[36]

A continuation method for large-scale modeling and control: from odes to pde, a round trip,

D. Nikitin, C. Canudas-de Wit, and P. Frasca, “A continuation method for large-scale modeling and control: from odes to pde, a round trip,” IEEE Transactions on Automatic Control, vol. 67, no. 10, pp. 5118– 5133, 2021

work page 2021
[37]

Mean-field control barrier functions: A framework for real-time swarm control,

S. W. Fung and L. Nurbekyan, “Mean-field control barrier functions: A framework for real-time swarm control,”Proceedings of the 2025 American Control Conference (ACC), 2025

work page 2025
[38]

Banach con- trol barrier functions for large-scale swarm control,

X. Gao, G. Pascual, S. Brown, and S. Mart ´ınez, “Banach con- trol barrier functions for large-scale swarm control,”arXiv preprint arXiv:2602.05011, 2026

work page arXiv 2026