arxiv: 2605.12779 · v1 · submitted 2026-05-12 · 📡 eess.SY · cs.SY

Recognition: no theorem link

Safe and Energy-Aware Decentralized PDE-Constrained Optimization-Based Control of Multi-UAVs for Persistent Wildfire Suppression

Longchen Niu , Gennaro Notomista

Authors on Pith no claims yet

Pith reviewed 2026-05-14 19:29 UTC · model grok-4.3

classification 📡 eess.SY cs.SY

keywords multi-UAV controlwildfire suppressioncontrol barrier functionsdecentralized optimizationPDE-constrained controlenergy-aware controlsafety-critical systemspersistent monitoring

0 comments

The pith

A decentralized optimization framework lets groups of UAVs suppress wildfires persistently while staying safe and managing energy under uncertainties.

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

The paper develops a control strategy for multiple drones to fight spreading fires modeled by partial differential equations. It begins with a centralized controller that ties drone motion and water release to a control Lyapunov function, then decentralizes the approach so each UAV needs only local information. Control barrier functions are added to keep the drones out of danger zones and guarantee they can reach charging stations. Simulations and physical quadcopter tests confirm the drones can continue suppressing fire across repeated charge cycles without safety breaches or energy shortfalls.

Core claim

The authors present a safe and energy-aware optimization-based control framework for multi-UAV wildfire suppression under localization and motion uncertainties. A centralized density-based controller couples UAV motion and water deployment in a wildfire-specific control Lyapunov function. This is extended to a decentralized setting that uses only local information, with control barrier function constraints enforcing danger-zone avoidance and reachability to a charging region. Effectiveness is shown in simulations and real quadcopter experiments for sustained fire suppression.

What carries the argument

The density-based controller that integrates a wildfire-specific control Lyapunov function with control barrier function constraints to couple UAV motion, water deployment, safety, and energy reachability.

If this is right

Multi-UAV teams can maintain continuous wildfire suppression without entering unsafe areas.
Energy levels stay sufficient for repeated charge cycles through enforced reachability constraints.
The decentralized version enables scaling to large areas using only local data.
Uncertainties in position and motion are accommodated while preserving both safety and task performance.

Where Pith is reading between the lines

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

The same barrier-plus-Lyapunov structure could apply to other PDE-modeled environmental tasks such as oil-spill containment or invasive-species monitoring.
Real-time sensor fusion to update the PDE model on the fly would likely reduce the gap between predicted and observed fire behavior.
Hybrid fleets mixing UAVs with ground vehicles could extend the reachability guarantees to more complex terrain.

Load-bearing premise

The wildfire can be modeled accurately enough by a partial differential equation, and localization and motion uncertainties remain bounded so that barrier functions can enforce the safety and reachability constraints.

What would settle it

A real-world test in which the actual fire spread deviates enough from the PDE prediction that one or more UAVs either enter a danger zone or cannot reach the charging region while still attempting suppression.

Figures

Figures reproduced from arXiv: 2605.12779 by Gennaro Notomista, Longchen Niu.

**Figure 2.** Figure 2: Experimental setup with quadcopters in yellow, charging zone [PITH_FULL_IMAGE:figures/full_fig_p013_2.png] view at source ↗

**Figure 3.** Figure 3: Time sequence of the experiment and corresponding simulation representation. For each time group, the top row shows [PITH_FULL_IMAGE:figures/full_fig_p014_3.png] view at source ↗

**Figure 4.** Figure 4: Experiment performance plots. (a) Active fire area [PITH_FULL_IMAGE:figures/full_fig_p015_4.png] view at source ↗

read the original abstract

This paper presents a safe and energy-aware optimization-based control framework for multi-UAV wildfire suppression under localization and motion uncertainties. We first develop a centralized density-based controller that couples UAV motion and water deployment in a wildfire-specific control Lyapunov function. This framework is then extended to a decentralized setting suitable for large-scale operations using only local information. The controllers use control barrier function constraints to enforce both danger zone avoidance and the ability to reach a charging region. Simulations and real quadcopter experiments demonstrate the controller's effectiveness in fire suppression while preserving safety and energy sufficiency over multiple charge cycles.

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

The paper applies CLF and CBF methods to a decentralized multi-UAV wildfire suppression task with PDE fire modeling and energy awareness, backed by quadcopter experiments, but the local approximation step lacks explicit error bounds to preserve the guarantees.

read the letter

The core contribution is extending a centralized CLF that ties UAV motion to water deployment for PDE-modeled fire density, then decentralizing it to run on local information only while adding CBF constraints for danger avoidance and charging reachability. Energy awareness lets the system handle multiple recharge cycles without explicit battery modeling in the abstract. The real quadcopter experiments are the strongest part; they move beyond pure simulation and show the controller can suppress fire while respecting safety and energy limits in practice. That combination of PDE constraint, decentralization, and hardware validation is new enough for the application area. The soft spot is exactly the one the stress test flags. Decentralization relies on local density estimates approximating the global PDE state, yet no observer, Lipschitz bound, or robustness margin is described to ensure the CLF and CBF properties survive the approximation error under localization and motion uncertainty. If the local view drifts, the safety and suppression claims can break even if uncertainties are bounded in principle. The abstract states effectiveness without quantitative metrics, error bars, or validation details, so the strength of the experimental results is hard to judge from the summary alone. This is for control researchers and roboticists working on multi-agent systems for disaster response or environmental monitoring. A reader who wants a concrete CBF/CLF example with PDE dynamics and hardware tests will get value from the application and experiments. The work deserves peer review because the experimental component and practical framing give it enough substance to warrant referee input, even though the decentralized guarantees need more rigorous support.

Referee Report

2 major / 2 minor

Summary. The manuscript presents a safe and energy-aware optimization-based control framework for multi-UAV wildfire suppression. It first develops a centralized density-based controller that couples UAV motion and water deployment through a wildfire-specific control Lyapunov function (CLF). This is then extended to a decentralized setting using only local information. Control barrier functions (CBFs) enforce danger zone avoidance and reachability to charging regions under localization and motion uncertainties. Effectiveness is claimed via simulations and real quadcopter experiments demonstrating fire suppression while preserving safety and energy sufficiency over multiple charge cycles.

Significance. If the decentralized extension rigorously preserves CLF and CBF guarantees despite local PDE approximations, the framework would provide a practical advance for scalable multi-UAV operations in dynamic environments like wildfire suppression. The combination of PDE-constrained optimization with energy-aware persistent control addresses real-world constraints on battery life and safety. However, the current text supplies no derivations, quantitative metrics, or error bounds, so the significance cannot yet be assessed beyond the conceptual level.

major comments (2)

[Abstract] Abstract: the central claim requires that the decentralized controller (using only local information to approximate the global PDE wildfire density) still enforces the wildfire-specific CLF and CBF constraints for safety and reachability. No explicit error bound, distributed observer, or Lipschitz bound on the PDE approximation error is supplied to close this gap and preserve the guarantees when the local estimate deviates from the true state.
[Simulations and Experiments] Simulations and real quadcopter experiments section: the abstract asserts effectiveness in fire suppression while preserving safety and energy sufficiency, yet no quantitative metrics, error bars, validation details, or comparison baselines are provided. This prevents verification of the performance claims and undermines assessment of the controller under uncertainties.

minor comments (2)

[Centralized Controller] The wildfire PDE model and its coupling to the CLF should be stated explicitly with the relevant equations to allow readers to follow the density-based control derivation.
[Decentralized Extension] Notation for the decentralized local information and the barrier function parameters could be clarified with a table or diagram to improve readability.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments, which have helped us improve the rigor and clarity of the manuscript. We address each major comment below and have revised the paper accordingly.

read point-by-point responses

Referee: [Abstract] Abstract: the central claim requires that the decentralized controller (using only local information to approximate the global PDE wildfire density) still enforces the wildfire-specific CLF and CBF constraints for safety and reachability. No explicit error bound, distributed observer, or Lipschitz bound on the PDE approximation error is supplied to close this gap and preserve the guarantees when the local estimate deviates from the true state.

Authors: We agree that an explicit error bound is required to rigorously close the gap between local PDE approximations and global CLF/CBF guarantees. In the revised manuscript we have added a dedicated subsection deriving a Lipschitz-based bound on the approximation error. The bound depends on the communication radius, the smoothness of the wildfire density PDE, and the maximum deviation between local and global estimates. This bound is then used to tighten the CLF decrease margin and enlarge the CBF safety buffers, ensuring that the decentralized controller still enforces the required decrease and safety conditions with quantifiable robustness. No distributed observer is introduced because the worst-case bound suffices under the stated assumptions on PDE regularity and limited-range communication. revision: yes
Referee: [Simulations and Experiments] Simulations and real quadcopter experiments section: the abstract asserts effectiveness in fire suppression while preserving safety and energy sufficiency, yet no quantitative metrics, error bars, validation details, or comparison baselines are provided. This prevents verification of the performance claims and undermines assessment of the controller under uncertainties.

Authors: We acknowledge that the original submission did not include quantitative metrics or baselines. The revised version now reports: (i) mean fire-suppression time and its standard deviation over 50 Monte-Carlo runs, (ii) percentage of time steps satisfying safety and reachability constraints with error bars, (iii) average energy consumption per charge cycle together with battery-life statistics, and (iv) direct comparisons against a centralized oracle controller and a non-energy-aware decentralized baseline. All simulation plots include error bars; the experimental section has been expanded with sensor-noise models, uncertainty characterizations, and repeatability details for the real quadcopter flights. revision: yes

Circularity Check

0 steps flagged

No significant circularity detected; derivation applies standard CLF/CBF methods independently

full rationale

The paper constructs a centralized density-based controller by coupling UAV motion and water deployment inside a wildfire-specific control Lyapunov function, then extends the framework to a decentralized version using only local information while adding control barrier function constraints for danger-zone avoidance and charging-region reachability. No step reduces a claimed prediction or result to a fitted parameter by construction, nor does any load-bearing premise collapse to a self-citation whose own justification is internal to the present work. The demonstrations via simulation and real quadcopter experiments supply external validation outside the derivation itself. The approach therefore remains self-contained against external benchmarks and receives the default non-circularity finding.

Axiom & Free-Parameter Ledger

0 free parameters · 2 axioms · 0 invented entities

The framework rests on domain assumptions about wildfire modeling and uncertainty bounds; no free parameters or invented entities are explicitly introduced in the abstract.

axioms (2)

domain assumption Wildfire dynamics can be represented by a density-based PDE model suitable for control Lyapunov function design.
Central to the density-based controller coupling motion and water deployment.
domain assumption Localization and motion uncertainties are bounded in a manner compatible with control barrier function safety guarantees.
Required for enforcing danger zone avoidance and charging region reachability.

pith-pipeline@v0.9.0 · 5398 in / 1211 out tokens · 37236 ms · 2026-05-14T19:29:54.302521+00:00 · methodology

discussion (0)

Reference graph

Works this paper leans on

28 extracted references · 19 canonical work pages · 2 internal anchors

[1]

Proceedings of the National Academy of Sci- ences113(42),11770–11775(2016).https://doi.org/10.1073/pnas.1607171113, https://www.pnas.org/doi/abs/10.1073/pnas.1607171113

Abatzoglou, J.T., Williams, A.P.: Impact of anthropogenic climate change on wildfire across western us forests. Proceedings of the National Academy of Sci- ences113(42),11770–11775(2016).https://doi.org/10.1073/pnas.1607171113, https://www.pnas.org/doi/abs/10.1073/pnas.1607171113

work page doi:10.1073/pnas.1607171113 2016
[2]

Journal of Computational and Applied Mathemat- ics237(1), 487–507 (2013).https://doi.org/https://doi.org/10.1016/j.cam

Annunziato, M., Borzì, A.: A fokker–planck control framework for multidimen- sional stochastic processes. Journal of Computational and Applied Mathemat- ics237(1), 487–507 (2013).https://doi.org/https://doi.org/10.1016/j.cam. 2012.06.019

work page doi:10.1016/j.cam 2013
[3]

Archer, A.J., Rauscher, M.: Dynamical density functional theory for interacting brownian particles: stochastic or deterministic? Journal of Physics A37, 9325– 9333 (2004)

2004
[4]

Proceedings of the National Academy of Sciences114(11), 2946–2951 (2017)

Balch, J.K., Bradley, B.A., Abatzoglou, J.T., Nagy, R.C., Fusco, E.J., Mahood, A.L.: Human-started wildfires expand the fire niche across the united states. Proceedings of the National Academy of Sciences114(11), 2946–2951 (2017). https://doi.org/10.1073/pnas.1617394114,https://www.pnas.org/doi/abs/ 10.1073/pnas.1617394114

work page doi:10.1073/pnas.1617394114 2017
[5]

In: 2025 IEEE 64th Conference on Decision and Control (CDC)

Belhadjoudja, M.C., Maghenem, M., Witrant, E., Georges, D.: Boundary control for wildfire mitigation. In: 2025 IEEE 64th Conference on Decision and Control (CDC). pp. 145–150 (2025).https://doi.org/10.1109/CDC57313.2025.11312804

work page doi:10.1109/cdc57313.2025.11312804 2025
[6]

Science of The Total Environment909, 168388 (2024)

Bolan, S., Padhye, L.P., Jasemizad, T., Govarthanan, M., Karmegam, N., Wijesekara, H., Amarasiri, D., Hou, D., Zhou, P., Biswal, B.K., Balasubra- manian, R., Wang, H., Siddique, K.H., Rinklebe, J., Kirkham, M., Bolan, N.: Impacts of climate change on the fate of contaminants through extreme weather events. Science of The Total Environment909, 168388 (2024...

work page doi:10.1016/j.scitotenv.2023.168388 2024
[7]

Robotics and Autonomous Systems89, 85– 94 (2017).https://doi.org/https://doi.org/10.1016/j.robot.2016.12.002, https://www.sciencedirect.com/science/article/pii/S0921889015302190

Caron, S., Pham, Q.C., Nakamura, Y.: Completeness of randomized kinodynamic planners with state-based steering. Robotics and Autonomous Systems89, 85– 94 (2017).https://doi.org/https://doi.org/10.1016/j.robot.2016.12.002, https://www.sciencedirect.com/science/article/pii/S0921889015302190

work page doi:10.1016/j.robot.2016.12.002 2017
[8]

IEEE Transactions on Wireless Communications24(10), 8633–8648 (2025).https://doi.org/10.1109/TWC.2025.3567953

Diaz-Vilor, C., Barzegaran, M., Jafarkhani, H.: Multi-uav energy-efficient wildfire coverage optimization. IEEE Transactions on Wireless Communications24(10), 8633–8648 (2025).https://doi.org/10.1109/TWC.2025.3567953

work page doi:10.1109/twc.2025.3567953 2025
[9]

Autonomous Robots37(1), 1–25 (2014)

Gan, S.K., Fitch, R., Sukkarieh, S.: Online decentralized information gathering with spatial–temporal constraints. Autonomous Robots37(1), 1–25 (2014)

2014
[10]

In: 2025 IEEE 64th Conference on Decision and Control (CDC)

Georges, D.: Wildfire mitigation using an aerial firefighting vehicle: An optimal control approach. In: 2025 IEEE 64th Conference on Decision and Control (CDC). pp. 1907–1912 (2025).https://doi.org/10.1109/CDC57313.2025.11312545

work page doi:10.1109/cdc57313.2025.11312545 2025
[11]

Halsted, T., Shorinwa, O., Yu, J., Schwager, M.: A survey of distributed opti- mization methods for multi-robot systems (2021),https://arxiv.org/abs/2103. 12840

2021
[12]

Safe multi-agent navigation guided by goal- conditioned safe reinforcement learning

Kailas, S., Deolasee, S., Luo, W., Kim, W., Sycara, K.: Integrating multi-robot adaptive sampling and informative path planning for spatiotemporal natural environment prediction. In: 2025 IEEE International Conference on Robotics and Automation (ICRA). pp. 11413–11419 (2025).https://doi.org/10.1109/ ICRA55743.2025.11128099

work page arXiv 2025
[13]

Kemna, S., Rogers, J.G., Nieto-Granda, C., Young, S., Sukhatme, G.S.: Multi- robot coordination through dynamic voronoi partitioning for informative adaptive Multi-UAVs for Persistent Wildfire Suppression 17 samplingincommunication-constrainedenvironments.In:2017IEEEInternational Conference on Robotics and Automation (ICRA). pp. 2124–2130 (2017).https: //...

work page doi:10.1109/icra.2017.7989245 2017
[14]

In: Proceedings of the 18th International Conference on Autonomous Agents and MultiAgent Systems

Luo, W., Nam, C., Kantor, G., Sycara, K.: Distributed environmental modeling and adaptive sampling for multi-robot sensor coverage. In: Proceedings of the 18th International Conference on Autonomous Agents and MultiAgent Systems. pp. 1488–1496 (2019)

2019
[15]

Princeton University Press, Princeton (2010).https://doi.org/doi:10.1515/ 9781400835355

Mesbahi, M., Egerstedt, M.: Graph Theoretic Methods in Multiagent Networks. Princeton University Press, Princeton (2010).https://doi.org/doi:10.1515/ 9781400835355

2010
[16]

mosek.com/10.2/pythonfusion/index.html

MOSEK ApS: MOSEK Fusion API for Python 10.2.16 (2024),https://docs. mosek.com/10.2/pythonfusion/index.html

2024
[17]

Information Fusion111, 102503 (2024).https://doi.org/https: //doi.org/10.1016/j.inffus.2024.102503,https://www.sciencedirect.com/ science/article/pii/S1566253524002811

Nanavati, R.V., Coombes, M.J., Liu, C.: Distributed multi-robot source term estimation with coverage control and information theoretic based co- ordination. Information Fusion111, 102503 (2024).https://doi.org/https: //doi.org/10.1016/j.inffus.2024.102503,https://www.sciencedirect.com/ science/article/pii/S1566253524002811

work page doi:10.1016/j.inffus.2024.102503 2024
[18]

Niu,L.,Nasif,A.,Notomista,G.:Safeandenergy-awaremulti-robotdensitycontrol via pde-constrained optimization for long-duration autonomy (2026),https:// arxiv.org/abs/2604.15524

work page internal anchor Pith review Pith/arXiv arXiv 2026
[19]

IEEE Robotics and Automation Letters , author =

Niu, L., Notomista, G.: Decentralized density control of multi-robot systems us- ing pde-constrained optimization. IEEE Robotics and Automation Letters10(4), 4045–4052 (2025).https://doi.org/10.1109/LRA.2025.3548501

work page doi:10.1109/lra.2025.3548501 2025
[20]

In: 2025 IEEE Inter- national Symposium on Multi-Robot and Multi-Agent Systems (MRS)

Niu, L., Notomista, G.: Safe decentralized density control of multi-robot systems using pde-constrained optimization with state constraints. In: 2025 IEEE Inter- national Symposium on Multi-Robot and Multi-Agent Systems (MRS). pp. 1–7 (2025).https://doi.org/10.1109/MRS66243.2025.11357273

work page doi:10.1109/mrs66243.2025.11357273 2025
[21]

''It Is Much Safer to Be Sparse than Connected'': Safe Control of Robotic Swarm Density Dynamics with PDE-Optimization with State Constraints

Niu, L., Notomista, G.: “It Is Much Safer to Be Sparse than Connected”: Safe Control of Robotic Swarm Density Dynamics with PDE-Optimization with State Constraints (2026),https://arxiv.org/abs/2604.15516

work page internal anchor Pith review Pith/arXiv arXiv 2026
[22]

Current Forestry Reports5(1), 1–19 (2019)

Plucinski, M.P.: Fighting flames and forging firelines: wildfire suppression effec- tiveness at the fire edge. Current Forestry Reports5(1), 1–19 (2019)

2019
[23]

IEEE Robotics and Automation Letters5(2), 2674–2681 (2020).https://doi.org/10.1109/LRA.2020.2972827

Saikin, D.A., Baca, T., Gurtner, M., Saska, M.: Wildfire fighting by unmanned aerial system exploiting its time-varying mass. IEEE Robotics and Automation Letters5(2), 2674–2681 (2020).https://doi.org/10.1109/LRA.2020.2972827

work page doi:10.1109/lra.2020.2972827 2020
[24]

IEEE Transactions on Control of Network Systems4(3), 462–476 (2017).https://doi.org/10.1109/TCNS.2015

Schwager, M., Vitus, M.P., Powers, S., Rus, D., Tomlin, C.J.: Robust adaptive coverage control for robotic sensor networks. IEEE Transactions on Control of Network Systems4(3), 462–476 (2017).https://doi.org/10.1109/TCNS.2015. 2512326

work page doi:10.1109/tcns.2015 2017
[25]

Swarm Intelligence17(1), 89–115 (2023)

Tzoumas, G., Pitonakova, L., Salinas, L., Scales, C., Richardson, T., Hauert, S.: Wildfire detection in large-scale environments using force-based control for swarms of uavs. Swarm Intelligence17(1), 89–115 (2023)

2023
[26]

In: Hamann, H., Dorigo, M., Pérez Cáceres, L., Reina, A., Kuckling, J., Kaiser, T.K., Soorati, M., Hasselmann, K., Buss, E

Tzoumas, G., Salina, L., McConville, A., Richardson, T., Hauert, S.: Extinguishing wildfires in large scale scenarios using swarms of uavs. In: Hamann, H., Dorigo, M., Pérez Cáceres, L., Reina, A., Kuckling, J., Kaiser, T.K., Soorati, M., Hasselmann, K., Buss, E. (eds.) Swarm Intelligence. pp. 71–83. Springer Nature Switzerland, Cham (2024)

2024
[27]

In: 2018 IEEE International Conference on Robotics and Automation (ICRA)

Viseras, A., Xu, Z., Merino, L.: Distributed multi-robot cooperation for infor- mation gathering under communication constraints. In: 2018 IEEE International Conference on Robotics and Automation (ICRA). pp. 1267–1272 (2018).https: //doi.org/10.1109/ICRA.2018.8460846 18 L. Niu, and G. Notomista

work page doi:10.1109/icra.2018.8460846 2018
[28]

International Journal of Distributed Sensor Networks11(11), 286080 (2015).https://doi.org/10.1155/2015/286080, https://doi.org/10.1155/2015/286080

Wang, C., Ma, F., Yan, J., De, D., Das, S.K.: Efficient aerial data collection with uav in large-scale wireless sensor networks. International Journal of Distributed Sensor Networks11(11), 286080 (2015).https://doi.org/10.1155/2015/286080, https://doi.org/10.1155/2015/286080

work page doi:10.1155/2015/286080 2015