pith. sign in

arxiv: 2606.20239 · v1 · pith:OSD7VAIAnew · submitted 2026-06-18 · 🧮 math.OC

Optimizing Agricultural Drone Operations: From Launch and Recovery Siting to Tiered Routing Strategies

Ethan Kolby , Josh Noble , Max Z. Li This is my paper

Pith reviewed 2026-06-26 16:06 UTC · model grok-4.3

classification 🧮 math.OC
keywords agricultural dronesfacility locationp-median heuristicroute planningproblem decompositionoptimization heuristicsdrone spraying
0
0 comments X

The pith

A p-median heuristic for drone launch siting cuts planning time from over 97 seconds to under 1.2 seconds with only a 4% drop in covered field area.

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

The paper develops an operational framework for planning drone spraying in agriculture and benchmarks two main components against exact solvers. For choosing where to place launch and recovery sites, it replaces a mixed-integer program with a p-median heuristic that finishes in seconds rather than minutes while still servicing nearly the same total area. For the actual flight routes, it partitions the fields into six to eight spatial clusters and solves each subproblem separately, cutting computation time by roughly a factor of ten with little loss in performance. Together these changes bring full daily plans within reach on ordinary computers. The work addresses the practical barrier that current optimization tools scale too slowly for routine farm decisions.

Core claim

The paper establishes that the p-median heuristic for facility siting reduces runtime by three orders of magnitude from over 97 seconds to under 1.2 seconds with only a 4% reduction in serviced field area compared to the mixed-integer program baseline. For route planning, partitioning the target area into 6 to 8 spatial clusters reduces computation time by an order of magnitude with minimal degradation in serviced area. The resulting framework supports minute-scale planning on commodity hardware.

What carries the argument

p-median heuristic for selecting launch and recovery sites to minimize total distance to fields, paired with tiered decomposition of the routing problem into 6-8 spatial clusters.

If this is right

  • Daily spraying plans become computable in minutes on standard hardware instead of requiring longer runtimes.
  • The approach scales to larger numbers of fields without specialized computing resources.
  • Small reductions in covered area are acceptable when offset by large gains in planning speed.
  • Optimization tools move from research prototypes to practical daily decision support for drone operations.

Where Pith is reading between the lines

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

  • Adding weather constraints to the siting and routing steps could reduce the risk that plans become unusable on the day of operation.
  • Testing the 6-8 cluster choice on a broader set of irregular field layouts would show how stable the speed-accuracy trade-off remains.
  • Solving siting and routing together in one model might recover some of the 4% coverage that is currently traded for speed.

Load-bearing premise

That the 4% coverage loss and the performance of 6-8 clusters will remain acceptable when the methods are applied to field geometries and conditions outside the tested instances.

What would settle it

Running the p-median heuristic and exact MIP on a fresh collection of real farm fields with different shapes and sizes, then checking whether runtime savings stay near three orders of magnitude and coverage loss stays near 4%.

Figures

Figures reproduced from arXiv: 2606.20239 by Ethan Kolby, Josh Noble, Max Z. Li.

Figure 1
Figure 1. Figure 1: They are places where aircraft would take off, land, and swap batteries. We distinguish [PITH_FULL_IMAGE:figures/full_fig_p009_1.png] view at source ↗
Figure 1
Figure 1. Figure 1: An example of a launch and recovery “facility,” which we approach from a facility siting [PITH_FULL_IMAGE:figures/full_fig_p010_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: A geometric illustration of facility siting and active node assignments across an example [PITH_FULL_IMAGE:figures/full_fig_p016_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Example field illustrating how passive nodes are assigned coverage rewards. Nodes in the [PITH_FULL_IMAGE:figures/full_fig_p017_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: An example of the tiered routing generation. (a) The macro-level tour established by [PITH_FULL_IMAGE:figures/full_fig_p019_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Left: Facility selection runtime vs. facility selection methods. Right: Routing opti￾mization runtime vs. facility selection method [PITH_FULL_IMAGE:figures/full_fig_p023_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Passive node coverage percentage vs. facility selection method. [PITH_FULL_IMAGE:figures/full_fig_p023_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Passive coverage vs. active node spacing and number of subgroups. [PITH_FULL_IMAGE:figures/full_fig_p025_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Coverage efficiency vs. active node spacing and number of subgroups. [PITH_FULL_IMAGE:figures/full_fig_p025_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Routing runtime vs. active node spacing and number of subgroups. [PITH_FULL_IMAGE:figures/full_fig_p026_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Routing runtime vs. active node spacing for a fixed number of 4 subgroups. [PITH_FULL_IMAGE:figures/full_fig_p026_10.png] view at source ↗
read the original abstract

Drones are increasingly used in agriculture, where tight margins demand efficient planning. Current optimization tools suffer from exponential runtimes as problem sizes grow, necessitating practical heuristics for daily operations. This paper presents an operational framework and benchmarking analysis for drone spraying operations. We evaluate the trade-offs between facility siting methods and tiered routing parameters. For facility siting, comparing a Mixed-Integer Program (MIP) baseline against a $p$-Median heuristic shows that the heuristic reduces runtime by three orders of magnitude, from over 97 seconds to under 1.2 seconds, with only a 4\% reduction in serviced field area. For route planning, a tiered problem decomposition approach partitioning the target area into 6 to 8 spatial clusters reduces computation time by an order of magnitude with minimal degradation in serviced area. This framework achieves minute-scale planning on commodity hardware, demonstrating operational relevance. Future research will incorporate weather modeling, integrated optimization of facility location and routing, and validation across diverse field geometries.

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.

Referee Report

2 major / 0 minor

Summary. The manuscript presents an operational framework for agricultural drone spraying operations. It benchmarks a p-median heuristic against a mixed-integer program (MIP) baseline for launch and recovery facility siting, and evaluates a tiered routing strategy that decomposes the target area into 6-8 spatial clusters. The central empirical claims are that the p-median heuristic reduces runtime from over 97 seconds to under 1.2 seconds with only a 4% reduction in serviced field area, while the tiered decomposition reduces computation time by an order of magnitude with minimal degradation in serviced area, enabling minute-scale planning on commodity hardware.

Significance. If the reported runtime-coverage trade-offs prove reproducible, the work supplies practical, scalable heuristics that bridge exact optimization and daily operational use in agriculture. The explicit numerical comparisons (three-order-of-magnitude speed-up for siting, order-of-magnitude reduction for routing) constitute concrete evidence of operational relevance on the tested instances.

major comments (2)
  1. [Abstract] Abstract: the performance claims (runtime reduction from >97 s to <1.2 s with 4 % area loss; order-of-magnitude improvement for 6-8 clusters) are presented without any description of the underlying test instances, including number of fields, field geometries, instance sizes, or data sources. These details are load-bearing for assessing whether the reported deltas are representative or stable.
  2. [Abstract] Abstract: the statement that the tiered decomposition yields 'minimal degradation in serviced area' is not accompanied by quantitative metrics (e.g., exact percentage loss per cluster count) or by the method used to form the spatial clusters, preventing evaluation of the robustness of the 6-8 cluster recommendation.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed feedback on the abstract. The comments correctly identify opportunities to improve clarity by adding context on test instances and quantitative metrics. We will revise the abstract to address both points while preserving the manuscript's core claims and length constraints.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the performance claims (runtime reduction from >97 s to <1.2 s with 4 % area loss; order-of-magnitude improvement for 6-8 clusters) are presented without any description of the underlying test instances, including number of fields, field geometries, instance sizes, or data sources. These details are load-bearing for assessing whether the reported deltas are representative or stable.

    Authors: We agree that the abstract would benefit from a concise description of the test instances to contextualize the performance claims. In the revised version we will insert a short clause summarizing the instance characteristics (number of fields, representative geometries, scale, and data provenance) drawn from the experimental section, enabling readers to evaluate representativeness without expanding the abstract beyond its current scope. revision: yes

  2. Referee: [Abstract] Abstract: the statement that the tiered decomposition yields 'minimal degradation in serviced area' is not accompanied by quantitative metrics (e.g., exact percentage loss per cluster count) or by the method used to form the spatial clusters, preventing evaluation of the robustness of the 6-8 cluster recommendation.

    Authors: We concur that replacing the qualitative phrase with explicit metrics and a brief mention of the clustering approach would strengthen the abstract. The revision will report the observed area-loss percentages for the 6- and 8-cluster cases (as quantified in the results) and indicate the spatial partitioning method, thereby making the 6-8 cluster recommendation more transparent while remaining within abstract length limits. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical runtime and coverage deltas from direct computation

full rationale

The paper reports explicit computational comparisons (MIP baseline vs. p-median heuristic; tiered 6-8 cluster decomposition) on its test instances, giving concrete runtime reductions (97s to 1.2s; order-of-magnitude) and area losses (4%). No equations, fitted parameters, self-citations, or ansatzes are invoked to derive these deltas; they are measured outputs. The manuscript itself notes the results are instance-specific and flags future validation on other geometries, confirming the claims do not reduce to self-definition or prediction-by-construction.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

Abstract-only review; no free parameters, axioms, or invented entities are stated.

pith-pipeline@v0.9.1-grok · 5705 in / 1015 out tokens · 40309 ms · 2026-06-26T16:06:35.990217+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

36 extracted references · 33 canonical work pages

  1. [1]

    Eye in the sky: 360° inspection of bridge infrastructure using uncrewed aerial vehicles,

    S. S. C. Congress, J. Escamilla III, H. Chimauriya, and A. J. Puppala, “Eye in the sky: 360° inspection of bridge infrastructure using uncrewed aerial vehicles,”Transportation Research Record, 2024.doi:10.1177/03611981231185138

    work page doi:10.1177/03611981231185138 2024

  2. [2]

    Uav-based wildland fire air toxics data collection and analysis,

    P. Ragbir et al., “Uav-based wildland fire air toxics data collection and analysis,”Sensors, vol. 23, no. 7, p. 3561, 2023.doi:10.3390/s23073561

    work page doi:10.3390/s23073561 2023

  3. [3]

    Unmanned aerial vehicles for search and rescue: A survey,

    Z. Lyu et al., “Unmanned aerial vehicles for search and rescue: A survey,”Remote Sensing, vol. 15, no. 13, p. 3266, 2023.doi:10.3390/rs15133266

    work page doi:10.3390/rs15133266 2023

  4. [4]

    Spatial coverage in routing and path planning problems,

    K. Glock and A. Meyer, “Spatial coverage in routing and path planning problems,”European Journal of Operational Research, vol. 305, no. 1, pp. 1–20, 2023.doi:10.1016/j.ejor.2022. 02.031

    work page doi:10.1016/j.ejor.2022 2023

  5. [5]

    W. K. Mennell,Heuristics for solving three routing problems: Close-enough traveling sales- man problem, close-enough vehicle routing problem, and sequence-dependent team orienteer- ing problem. University of Maryland, College Park, 2009. [Online]. Available:http://hdl. handle.net/1903/9822

    2009

  6. [6]

    Decomposition strategies for vehicle routing heuristics,

    A. Santini, M. Schneider, T. Vidal, and D. Vigo, “Decomposition strategies for vehicle routing heuristics,”INFORMS Journal on Computing, vol. 35, no. 3, pp. 543–559, 2023.doi:10. 1287/ijoc.2023.1288

    arXiv 2023

  7. [7]

    The covering salesman problem,

    J. R. Current and D. A. Schilling, “The covering salesman problem,”Transportation Science, vol. 23, no. 3, pp. 208–213, 1989.doi:10.1287/trsc.23.3.208

    work page doi:10.1287/trsc.23.3.208 1989

  8. [8]

    The covering tour problem,

    M. Gendreau, G. Laporte, and F. Semet, “The covering tour problem,”Operations Research, vol. 45, no. 4, pp. 568–576, 1997.doi:10.1287/opre.45.4.568 29

    work page doi:10.1287/opre.45.4.568 1997

  9. [9]

    An integer-programming-based approach to the close-enough traveling salesman problem,

    B. Behdani and J. C. Smith, “An integer-programming-based approach to the close-enough traveling salesman problem,”INFORMS Journal on Computing, vol. 26, no. 3, pp. 415–432, 2014.doi:10.1287/ijoc.2013.0574

    work page doi:10.1287/ijoc.2013.0574 2014

  10. [10]

    A branch-and-bound algorithm for the close-enough traveling salesman problem,

    W. P. Coutinho, R. Martinelli, and R. S. de Camargo, “A branch-and-bound algorithm for the close-enough traveling salesman problem,”INFORMS Journal on Computing, vol. 28, no. 4, pp. 772–787, 2016.doi:10.1287/ijoc.2016.0711

    work page doi:10.1287/ijoc.2016.0711 2016

  11. [11]

    Hierarchical coverage path planning in com- plex 3d environments,

    C. Cao, J. Zhang, M. Travers, and H. Choset, “Hierarchical coverage path planning in com- plex 3d environments,” in2020 IEEE International Conference on Robotics and Automation (ICRA), IEEE, 2020, pp. 8437–8443.doi:10.1109/ICRA40945.2020.9196575

    work page doi:10.1109/icra40945.2020.9196575 2020

  12. [12]

    Maximum coverage capacitated facility location problem with range constrained drones,

    D. Chauhan, A. Unnikrishnan, and M. Figliozzi, “Maximum coverage capacitated facility location problem with range constrained drones,”Transportation Research Part C: Emerging Technologies, vol. 99, pp. 1–18, 2019.doi:10.1016/j.trc.2018.12.001

    work page doi:10.1016/j.trc.2018.12.001 2019

  13. [13]

    The flying sidekick traveling salesman problem: Optimization of drone-assisted parcel delivery,

    C. C. Murray and A. G. Chu, “The flying sidekick traveling salesman problem: Optimization of drone-assisted parcel delivery,”Transportation Research Part C: Emerging Technologies, vol. 54, pp. 86–109, 2015.doi:10.1016/j.trc.2015.03.005

    work page doi:10.1016/j.trc.2015.03.005 2015

  14. [14]

    The drone-assisted vehicle routing problem with robot stations,

    A. Morim, G. Campuzano, P. Amorim, M. Mes, and E. Lalla-Ruiz, “The drone-assisted vehicle routing problem with robot stations,”Expert Systems with Applications, vol. 238, p. 121 741, 2024.doi:10.1016/j.eswa.2023.121741

    work page doi:10.1016/j.eswa.2023.121741 2024

  15. [15]

    Drones for automated parcel delivery: Use case exploration and analysis,

    S. Zieher et al., “Drones for automated parcel delivery: Use case exploration and analysis,” Case Studies on Transport Policy, vol. 12, p. 101 253, 2024.doi:10.1016/j.trip.2024. 101253

    work page doi:10.1016/j.trip.2024 2024

  16. [16]

    Research on collaborative delivery path planning for trucks and drones in parcel delivery,

    T. Fu, S. Li, and Z. Li, “Research on collaborative delivery path planning for trucks and drones in parcel delivery,”Sensors, vol. 25, no. 10, p. 3087, 2025.doi:10.3390/s25103087

    work page doi:10.3390/s25103087 2025

  17. [17]

    Optimizing an on-demand delivery mode based on trucks and drones,

    L. Zhen, J. Gao, S. Wang, L. Gilbert, and X. Yue, “Optimizing an on-demand delivery mode based on trucks and drones,”Transportation Science, 2025.doi:10.1287/trsc.2024.0693

    work page doi:10.1287/trsc.2024.0693 2025

  18. [18]

    A method for solving traveling-salesman problems,

    G. A. Croes, “A method for solving traveling-salesman problems,”Operations Research, vol. 6, no. 6, pp. 791–812, 1958.doi:10.1287/opre.6.6.791 30

    work page doi:10.1287/opre.6.6.791 1958

  19. [19]

    An effective heuristic algorithm for the traveling-salesman problem,

    S. Lin and B. W. Kernighan, “An effective heuristic algorithm for the traveling-salesman problem,”Operations Research, vol. 21, no. 2, pp. 498–516, 1973.doi:10.1287/opre.21.2. 498

    work page doi:10.1287/opre.21.2 1973

  20. [20]

    A review on the state of the art in copter drones and flight control systems,

    J. Peksa and D. Mamchur, “A review on the state of the art in copter drones and flight control systems,”Sensors, vol. 24, no. 11, p. 3349, 2024.doi:10.3390/s24113349

    work page doi:10.3390/s24113349 2024

  21. [21]

    Deployment of a fixed-wing small uncrewed aircraft system for urban traffic–related air pollution assessment,

    J. Silberstein, R. Sasse, M. K. Harjam¨ aki, et al., “Deployment of a fixed-wing small uncrewed aircraft system for urban traffic–related air pollution assessment,”Journal of Atmospheric and Oceanic Technology, 2025.doi:10.1175/JTECH-D-24-0042.1

    work page doi:10.1175/jtech-d-24-0042.1 2025

  22. [22]

    An evaluation of fixed-wing unmanned aerial vehicle trends and corre- lations with respect to nato classification, region, eis date and operational specifications,

    E. Mitridis et al., “An evaluation of fixed-wing unmanned aerial vehicle trends and corre- lations with respect to nato classification, region, eis date and operational specifications,” Aerospace, vol. 10, no. 1, p. 53, 2023.doi:10.3390/aerospace10040382

    work page doi:10.3390/aerospace10040382 2023

  23. [23]

    A review on uav-based applications for precision agriculture,

    D. C. Tsouros, S. Bibi, and P. G. Sarigiannidis, “A review on uav-based applications for precision agriculture,”Information, vol. 10, no. 11, p. 349, 2019.doi:10.3390/info10110349

    work page doi:10.3390/info10110349 2019

  24. [24]

    Optimization of coverage path planning for agricultural drones in weed-infested fields using semantic segmentation,

    F. A. Lara-Molina et al., “Optimization of coverage path planning for agricultural drones in weed-infested fields using semantic segmentation,”Agriculture, vol. 15, no. 12, p. 1262, 2025. doi:10.3390/agriculture15121262

    work page doi:10.3390/agriculture15121262 2025

  25. [25]

    Deep learning-based weed detection using uav images: A comparative study,

    T. B. Shahi, S. Dahal, C. Sitaula, A. Neupane, and W. Guo, “Deep learning-based weed detection using uav images: A comparative study,”Drones, vol. 7, no. 10, p. 624, 2023.doi: 10.3390/drones7100624

    work page doi:10.3390/drones7100624 2023

  26. [26]

    A novel path planning approach for plant protection uav based on ddpg and ila optimization algorithm,

    P. Wang et al., “A novel path planning approach for plant protection uav based on ddpg and ila optimization algorithm,”Computers and Electronics in Agriculture, vol. 239, p. 111 006, 2025.doi:https://doi.org/10.1016/j.compag.2025.111006

    work page doi:10.1016/j.compag.2025.111006 2025

  27. [27]

    Integration of uav-LiDAR and uav-photogrammetry for infrastructure monitoring and bridge assessment,

    F. Gaspari, F. Ioli, F. Barbieri, E. Belcore, and L. Pinto, “Integration of uav-LiDAR and uav-photogrammetry for infrastructure monitoring and bridge assessment,” inThe Interna- tional Archives of the Photogrammetry, Remote Sensing and Spatial Information Sciences, vol. XLIII-B2-2022, 2022, pp. 995–1002.doi:10.5194/isprs-archives-XLIII-B2-2022- 995-2022 31

    work page doi:10.5194/isprs-archives-xliii-b2-2022- 2022

  28. [28]

    Uav photogramme- try for concrete bridge inspection using object-based image analysis (obia),

    S. Zollini, M. Alicandro, D. Dominici, R. Quaresima, and M. Giallonardo, “Uav photogramme- try for concrete bridge inspection using object-based image analysis (obia),”Remote Sensing, vol. 12, no. 19, p. 3180, 2020.doi:10.3390/rs12193180

    work page doi:10.3390/rs12193180 2020

  29. [29]

    Supporting search and rescue operations with uavs,

    S. Waharte and N. Trigoni, “Supporting search and rescue operations with uavs,” in2010 IEEE International Conference on Emerging Security Technologies (EST), 2010, pp. 142–147. doi:10.1109/EST.2010.31

    work page doi:10.1109/est.2010.31 2010

  30. [30]

    Aerial person detection for search and rescue: A survey,

    X. Zhang et al., “Aerial person detection for search and rescue: A survey,”Journal of Remote Sensing, vol. 9, p. 0474, 2025.doi:10.34133/remotesensing.0474

    work page doi:10.34133/remotesensing.0474 2025

  31. [31]

    Robust air-to-air channel model for swarms of drones in sar missions,

    S. Shaheen, W. Saad, et al., “Robust air-to-air channel model for swarms of drones in sar missions,”IEEE Access, vol. 11, pp. 68 890–68 896, 2023.doi:10 . 1109 / ACCESS . 2023 . 3292517

    2023

  32. [32]

    A review on early wildfire de- tection from unmanned aerial vehicles using deep learning-based computer vision algorithms,

    A. Bouguettaya, H. Zarzour, A. M. Taberkit, and A. Kechida, “A review on early wildfire de- tection from unmanned aerial vehicles using deep learning-based computer vision algorithms,” Signal Processing, vol. 190, p. 108 309, 2022.doi:10.1016/j.sigpro.2021.108309

    work page doi:10.1016/j.sigpro.2021.108309 2022

  33. [33]

    Remote sens- ing applications for prescribed burn research: A review,

    A. LoPresti, M. T. Hayden, K. Siegel, B. Poulter, E. N. Stavros, and L. E. Dee, “Remote sens- ing applications for prescribed burn research: A review,”International Journal of Wildland Fire, vol. 33, WF23130, 2024.doi:10.1071/WF23130

    work page doi:10.1071/wf23130 2024

  34. [34]

    Integer programming formulation of traveling salesman problems,

    C. E. Miller, A. W. Tucker, and R. A. Zemlin, “Integer programming formulation of traveling salesman problems,”Journal of the ACM (JACM), vol. 7, no. 4, pp. 326–329, 1960.doi: 10.1145/321043.321046

    work page doi:10.1145/321043.321046 1960

  35. [35]

    Computer solutions of the traveling salesman problem,

    S. Lin, “Computer solutions of the traveling salesman problem,”Bell System Technical Jour- nal, vol. 44, no. 10, pp. 2245–2269, 1965.doi:10.1002/j.1538-7305.1965.tb04146.x

    work page doi:10.1002/j.1538-7305.1965.tb04146.x 1965

  36. [36]

    Machine learning and optimization pipeline for agricultural drone aerial imaging operations,

    X. Ni et al., “Machine learning and optimization pipeline for agricultural drone aerial imaging operations,” Jul. 2024.doi:10.2514/6.2024-3657 32 Table 4: Performance Metrics (Mean±Std. Dev.) vs. Grid Spacing and Number of Subgroups. Subgroups Runtime (s) Coverage (%) Efficiency (%) Active Node Spacing: 4.68 m 1 706.95±93.79 89.03±7.66 0.68±0.04 2 351.02±...

    work page doi:10.2514/6.2024-3657 2024