Pith. sign in

REVIEW 4 major objections 5 minor 39 references

Geofenced Unmanned Aerial Robotic Defender for Deer Detection and Deterrence (GUARD)

T0 review · 4 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

Pith's one-line read An integrated drone can detect deer with 92% accuracy, patrol fields with 15% less energy than the standard back-and-forth route, and recharge itself for dusk-to-dawn operation.

desk verdict The 92% detection claim is contradicted by the paper's own Table I and must be fixed, but the system integration and coverage comparison are worth a referee's time. read the letter →

arxiv 2505.10770 v1 pith:3JDH2VHP submitted 2025-05-16 cs.RO cs.AIcs.MA

classification cs.ROcs.AIcs.MA
keywords deerdeterrenceUAVYOLOv5antcolonyoptimizationcoveragepathplanningautonomouschargingprecisionagricultureagriculturalrobotics
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

This paper argues that a small, low-cost drone can take over the job of keeping deer out of a vegetable farm, replacing fences and scare devices that are too expensive or too easily ignored. The authors built an integrated prototype that combines a YOLOv5 detector for real-time deer spotting, an ant-colony route planner that minimizes energy, and a contact charging dock meant to keep the drone flying from dusk to dawn. On the paper's tests the detector reached 92% average accuracy on its held-out set, and the ant-colony routes used about 15% less energy than the standard back-and-forth coverage pattern. If those results hold in the field, the same modular stack could be pointed at other monitoring-and-response tasks in agriculture and beyond.

What carries the argument

The mechanism that carries the argument is the tight coupling of three subsystems around an energy-aware waypoint loop. The detector is YOLOv5, a real-time object-detection network fine-tuned on annotated deer images; its output feeds a supervisory module that can interrupt the route. The coverage planner builds a grid of waypoints 38 meters apart, removes edges that collide with obstacles, and searches for a tour from the charging station through all valid waypoints and back using ant colony optimization, with an edge heuristic $\eta_{hij}=1/(\lambda d_{ij}+\gamma\theta_{hij})$ where $\lambda=0.1164$ kJ/m and $\gamma=0.0173$ kJ/deg weight straight-line distance and turning angle. The charging station uses passive 60-degree V-rails and spring-loaded pogo pins, with fiducial-marker landing to correct position before touchdown. What these pieces do together is convert a farm map, a battery gauge, and a camera feed into an autonomous patrol that can break off to haze deer and return to charge.

What would settle it

Run the fine-tuned detector on a set of white-tailed deer images or on footage from the target farm's dusk patrol and compare its detections with ground-truth sightings from trail cameras or observers; if accuracy on the local species falls well below the 92% reported on the barasingha test set, the central detection claim does not transfer to the intended deployment.

Watch

Extended reading notes

Core claim

The central claim is that the three hard problems in autonomous deer deterrence—seeing the deer, routing the patrol, and keeping the battery alive—can be solved on one inexpensive quadcopter and connected through a supervisory loop. The vision module is a fine-tuned YOLOv5 network that the paper reports as detecting deer with 92% average accuracy on its held-out test set (mAP@0.5 = 0.693, best F1 = 0.69 at a confidence threshold of 0.337). The coverage module models each flight leg by a weighted energy cost of straight-line distance and turn angle, then searches the waypoint tour with ant colony optimization; both the Ant System and Max-Min Ant System variants beat the conventional back-and-forth path, with a reported energy saving of roughly 15% and with valid tours found in the majority of trials. The charging module is a proof-of-concept dock whose V-rail and pogo-pin contacts, combined with fiducial-marker precision landing, is intended to return the drone to service without human battery swaps and thereby support dusk-to-dawn operation. The paper also describes a reinforcement-learning supervisor trained in a photorealistic simulator to coordinate detection, deterrence, and charging, with field integration still ongoing.

Load-bearing premise

The system's value on the target farm depends on a detector fine-tuned on barasingha deer images recognizing white-tailed deer in real Minnesota fields, and no cross-species or in-field validation is reported.

Editorial extensions

If this is right

  • A single drone could patrol a 13-acre farm through the night without human battery swaps, provided the charging dock's electrical contact and alignment work as designed.
  • Because the ant-colony planner beat the back-and-forth baseline in both single- and dual-drone simulations, the same energy-saving routes should scale to larger fields and multiple drones.
  • The modular split between vision, planning, charging, and supervision means the reinforcement-learning policy can be improved or replaced without rebuilding the lower layers already validated in simulation and small-plot trials.
  • With only the detector retrained, the same stack could address other monitoring-and-response tasks such as wildlife surveys, poaching patrols, or wildfire spotting.

Reading between the lines

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

  • The 92% accuracy figure is a benchmark on the training dataset's own held-out images, so it should be read as detector capacity rather than as measured field performance on the farm.
  • The 15% energy saving comes from a cost model that counts only straight-line distance and turn angle; wind, climb, and obstacle-avoidance maneuvers could shrink or reverse that gap in real sorties.
  • Dusk-to-dawn operation is an architectural promise: the charging circuit and landing-pad prototype are still under construction, so the end-to-end recharge-and-resume loop has not yet been demonstrated.
  • If the simulation-trained supervisor transfers poorly to real conditions, the deployed system may regress to the static coverage planner, though the vision and routing gains would remain intact.
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

4 major / 5 minor

Summary. The manuscript describes GUARD, an integrated unmanned aerial vehicle (UAV) system intended to detect and deter deer on a small Minnesota vegetable farm. The system combines a YOLOv5-based vision module for real-time deer detection, an ant colony optimization (ACO) coverage path planner that minimizes energy expenditure, a proof-of-concept contact-based charging dock, and a reinforcement learning (RL) supervisor trained in an AirSim/Unreal Engine simulation. The authors report that the detector achieves 92% average detection accuracy on a held-out test set, that ACO coverage paths are more energy-efficient than a back-and-forth baseline (with the Conclusions stating a 15% improvement), and that the charging dock enables extended dusk-to-dawn operation. The paper includes results from the vision module, path planning simulations, and qualitative discussions of the charging dock and RL agent, but several hardware and control components are explicitly stated to be under construction or only simulated.

Significance. If the reported claims were fully substantiated, the system would represent a useful low-cost approach to autonomous wildlife deterrence in precision agriculture, combining perception, planning, and battery management for small farms. The paper has some strengths: the energy cost constants used in the planner are taken from prior literature rather than fitted to the results; the detector is evaluated on a held-out test set; and the system is motivated by a real farm partnership. However, the central claims are not supported by the evidence in the manuscript. The 92% detection accuracy is contradicted by the paper's own Table I, the coverage comparison excludes a substantial fraction of ACO trials, and the charging dock and RL supervisor are not yet implemented or field-validated. As a result, the significance of the contribution as presented cannot be assessed; the current manuscript primarily documents an in-progress prototype.

major comments (4)
  1. [Section V.A, Table I, Abstract, Conclusions] The claim of "average detection accuracy of 92%" is undefined and is inconsistent with the metrics reported in Table I. Table I lists mAP@0.5 = 0.693, best F1 = 0.69, maximum recall = 0.86, and true-positive rate = 0.69. Since the claimed 92% exceeds the reported maximum recall, it cannot be a per-instance detection rate, and no definition of "accuracy" is provided anywhere in the paper. This is a load-bearing issue because deer detection is the trigger for the entire deterrence pipeline; if the actual performance is closer to the reported true-positive rate of 0.69, the system's core promise fails. The abstract and conclusions repeat the 92% figure without qualification, so the headline claim is unsupported by the paper's own evaluation.
  2. [Section V.B, Figure 11, Conclusions] The coverage path planning comparison is biased by the exclusion of invalid ACO trials. Section V.B states that out of 30 trials, AS generated valid solutions 19 times for the dual-drone problem and 28 times for the single-drone problem, while MMAS generated valid solutions 18 and 26 times, respectively. The statistics in Figure 11 are computed only over the valid solutions, whereas the back-and-forth baseline always produces a valid solution. This discards a substantial fraction of ACO runs that may represent real failure modes or higher-cost paths, and it inflates the apparent advantage of ACO. Additionally, the Conclusions claim that ACO covers fields "15% more efficiently" than the baseline, but no such figure is reported in Section V, and the figure does not provide error bars or significance tests. The 15% claim is therefore unsupported by the results as presented.
  3. [Sections IV.A, IV.D, Abstract, Conclusions] The system-level claims of "extended dusk-to-dawn operation" via an autonomous charging dock and of an "integrated UAV prototype" are not supported by the state of the components. Section IV.A states that the charging dock is a proof-of-concept: "CAD models have been validated in Gazebo; plywood prototypes are under construction," and the authors "intend to complete the circuit design" and manufacture a prototype in the next phase. Section IV.D states that the RL policy is trained only in simulation and that "integration and field validation of this agent are ongoing." Thus neither the autonomous charging nor the RL-based coordination has been demonstrated in the field, so the abstract and conclusions overstate the maturity of the system. This is load-bearing because the claimed contribution is the integrated system, not merely the individual simulated components.
  4. [Section IV.B, Section I] The detector is fine-tuned on a dataset of barasingha deer (Rucervus duvaucelii) obtained from Roboflow Universe, while the target pest identified in Section I is the white-tailed deer on a Minnesota farm. The paper provides no cross-species evaluation and no field validation on the target species. Given that the authors themselves cite work (Crossling et al. [8]) showing that models trained on stock imagery generalize poorly to on-farm data, the external validity of the detection results for the actual deployment scenario is questionable. This is a significant gap because the deer species differ in appearance and habitat, and the detection results in Section V.A are the only quantitative evidence for the system's perception capability.
minor comments (5)
  1. [Section I, Section II.A] There are several typographical and grammatical errors, e.g., "sytem" in the Introduction and "Some animals are may cause damage" in Section II.A. These should be corrected.
  2. [Section IV.A] The sentence "an autonomous charging routine that handles return-to-dock and." appears to be incomplete; the final clause is missing its object.
  3. [Section IV.C] The waypoint spacing is described as "arbitrarily set" to 38 meters, and the authors acknowledge that "more work has to be done to determine the ideal spacing." This free parameter affects the coverage results, and its arbitrariness should be discussed as a limitation in the evaluation.
  4. [Figure 11, Section V.B] The caption states that one trial was conducted for the back-and-forth baseline "since it always returns the same solution," but no variance or statistical significance is reported for the ACO means. It would be helpful to report standard deviations or confidence intervals over the valid trials.
  5. [Reference [24]] The dataset URL in reference [24] appears to contain a typo ("robomow.com" instead of a Roboflow URL), and the dataset name is inconsistently spelled ("Rucervus dasuceveli" vs. "Rucervus duvaucelii").

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the reported results are evaluated on held-out data and independent literature parameters, not fitted inputs renamed as predictions.

full rationale

None of the paper's load-bearing claims reduces to its inputs by construction. The vision result is a fine-tuned YOLOv5 detector evaluated on a held-out test subset of the same annotated dataset; although the abstract's 92% figure is not defined and contradicts Table I's reported mAP@0.5=0.693, best F1=0.69, and true-positive rate=0.69, that discrepancy is an internal-consistency or correctness problem, not a circularity. No parameter was fit to the test set and then renamed as a prediction. The coverage comparison uses an energy cost model cost(s)=lambda*d_s+gamma*theta_s with lambda=0.1164 kJ/m and gamma=0.0173 kJ/deg taken from prior literature [25] rather than fitted to the reported ACO/back-and-forth outcomes; both planners are evaluated under the same cost function, so the 15% improvement is a legitimate optimization comparison. The charging station is presented as a proof of concept, and no measured dock performance is claimed. There are no load-bearing self-citations and no uniqueness theorem imported from the authors' previous work. Accordingly, no circular step meets the evidentiary standard required by the review instructions.

Assumptions & free parameters 1 free parameters · 3 assumptions · 0 invented entities

The central claims depend on the detector generalizing to the target species, the energy model from prior drones applying to the PX4 quadcopter, and the simulator standing in for real farm behavior. No independent evidence is provided for these transfers. One hand-chosen parameter, the waypoint spacing, directly affects the coverage comparison.

free parameters (1)
  • waypoint spacing = 38 m
    Chosen arbitrarily in Section IV.C; determines coverage resolution and affects the energy comparison between ACO and back-and-forth.
assumptions (3)
  • domain assumption The YOLO model fine-tuned on Rucervus duvaucelii (barasingha) deer images will detect white-tailed deer on the target Minnesota farm.
    Section IV.B uses a barasingha deer dataset; Section I identifies white-tailed deer as the target pest. No cross-species validation or dataset mixing is reported.
  • domain assumption The energy cost model with lambda=0.1164 kJ/m and gamma=0.0173 kJ/deg from Modares et al. (2017) applies to the PX4 quadcopter used in this study.
    Invoked in Section IV.C to define the path objective and to compute the 15% efficiency gain; if the constants do not match the platform, the comparison is quantitatively unreliable.
  • domain assumption AirSim with procedurally generated deer scenes is a valid proxy for real deterrence behavior and crop-protection outcomes.
    The RL policy is trained solely in simulation (Section IV.D) and no sim-to-real transfer result is presented; the paper itself states field validation is ongoing.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Geofenced Unmanned Aerial Robotic Defender for Deer Detection and Deterrence (GUARD)." pith.science (2026). https://pith.science/paper/3JDH2VHP

@misc{pith2026250510770,
  author       = {Pith},
  title        = {Pith review of: Geofenced Unmanned Aerial Robotic Defender for Deer Detection and Deterrence (GUARD)},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3JDH2VHP}},
  note         = {Machine review of arXiv:2505.10770}
}
read the original abstract

Wildlife-induced crop damage, particularly from deer, threatens agricultural productivity. Traditional deterrence methods often fall short in scalability, responsiveness, and adaptability to diverse farmland environments. This paper presents an integrated unmanned aerial vehicle (UAV) system designed for autonomous wildlife deterrence, developed as part of the Farm Robotics Challenge. Our system combines a YOLO-based real-time computer vision module for deer detection, an energy-efficient coverage path planning algorithm for efficient field monitoring, and an autonomous charging station for continuous operation of the UAV. In collaboration with a local Minnesota farmer, the system is tailored to address practical constraints such as terrain, infrastructure limitations, and animal behavior. The solution is evaluated through a combination of simulation and field testing, demonstrating robust detection accuracy, efficient coverage, and extended operational time. The results highlight the feasibility and effectiveness of drone-based wildlife deterrence in precision agriculture, offering a scalable framework for future deployment and extension.

Figures

Figures reproduced from arXiv: 2505.10770 by the authors.

Figure 1
Figure 1. High-level system architecture showing integration of computer vision, path planning, and autonomous charging modules. [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. Ground-based autonomous charging dock. (a) Inter [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Illustration of our methodology for energy-efficient [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (8 more)
Figure 4
Figure 4. Figure 4: Birds-eye view of the farm our solution is designed [PITH_FULL_IMAGE:figures/full_fig_p004_4.png]
Figure 5
Figure 5. Figure 5: Illustration of the parameters that contribute to energy [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 6
Figure 6. Figure 6: Virtual deer herd with realistic behaviours. [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]
Figure 7
Figure 7. Figure 7: Top-down view of a UAV detecting and responding [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]
Figure 8
Figure 8. Figure 8: Key evaluation curves for the deer-detection model. [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]
Figure 9
Figure 9. Figure 9: Qualitative YOLOv5 detections at 80% confidence in [PITH_FULL_IMAGE:figures/full_fig_p007_9.png]
Figure 10
Figure 10. Figure 10: Representative MMAS coverage paths for (a) two [PITH_FULL_IMAGE:figures/full_fig_p007_10.png]
Figure 11
Figure 11. Figure 11: Comparison of results of the back-and-forth, AS, and MMAS coverage planning algorithms for both the single-drone [PITH_FULL_IMAGE:figures/full_fig_p008_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

39 extracted references · 37 canonical work pages

  1. [8]

    Automated pest control: Computer vision for wildlife surveillance,

    C. J. Crossling, H. Vadapalli, and D. van der Haar, “Automated pest control: Computer vision for wildlife surveillance,” inProc. 2024 7th Int. Conf. on Computational Intelligence and Intelligent Systems , pp. 38–43, 2024

  2. [1]

    Minnesota hunters take aim at program allowing farmers to shoot large numbers of deer,

    T. Kennedy, “Minnesota hunters take aim at program allowing farmers to shoot large numbers of deer,” TCA News Service, 2024. Place: Chicago Publisher: Tribune Content Agency LLC

  3. [2]

    Density, population trends, and economic impacts of white-tailed deer in iowa,

    J. Tyndall et al., “Density, population trends, and economic impacts of white-tailed deer in iowa,” Tech. Rep., Iowa DNR, 2024

  4. [3]

    Rutgers cooperative extension white-tailed deer drone survey,

    R. Gene Huntington, E. Huntington, and S. Green, “Rutgers cooperative extension white-tailed deer drone survey,” New Jersey Agricultural Experiment Station, 2023

  5. [4]

    S. A. Valitzski, Evaluation of sound as a deterrent for reducing deer- vehicle collisions, PhD thesis, University of Georgia, 2007

  6. [5]

    Drones outperform dogs for hazing bears: a compar- ison of carnivore aversive conditioning tools,

    W. M. Sarmento, “Drones outperform dogs for hazing bears: a compar- ison of carnivore aversive conditioning tools,” Frontiers in Conservation Science, vol. 5, p. 1478450, 2025

  7. [6]

    You only look once: Unified, real-time object detection,

    J. Redmon, S. Divvala, R. Girshick, and A. Farhadi, “You only look once: Unified, real-time object detection,” 2016

  8. [7]

    Collectively ad- vancing deep learning for animal detection in drone imagery: Successes, challenges, and research gaps,

    D. Axford, F. Sohel, M. Vanderklift, and A. Hodgson, “Collectively ad- vancing deep learning for animal detection in drone imagery: Successes, challenges, and research gaps,” Ecological Informatics, p. 102842, 2024

Show all 39 references
  1. [9]

    Airsim: High-fidelity visual and physical simulation for autonomous vehicles,

    S. Shah, D. Dey, C. Lovett, and A. Kapoor, “Airsim: High-fidelity visual and physical simulation for autonomous vehicles,” in Field and Service Robotics, pp. 621–635, Springer, 2018

  2. [10]

    Autonomous Detection and Deterrence of Pigeons on Buildings by Drones,

    F. Schiano, D. Natter, D. Zambrano, and D. Floreano, “Autonomous Detection and Deterrence of Pigeons on Buildings by Drones,” IEEE Access, vol. 10, pp. 1745–1755, 2022

  3. [11]

    Autonomous Pest Bird Deterring for Agricultural Crops Using Teams of Unmanned Aerial Vehicles,

    Z. Wang and K. C. Wong, “Autonomous Pest Bird Deterring for Agricultural Crops Using Teams of Unmanned Aerial Vehicles,” in 2019 12th Asian Control Conference (ASCC) , pp. 108–113, June 2019. Accepted to the Novel Approaches for Precision Agriculture and Forestry with Autonomo...

  4. [12]

    Online localization of radio-tagged wildlife with an autonomous aerial robot system,

    O. M. Cliff, R. Fitch, S. Sukkarieh, D. L. Saunders, and R. Heinsohn, “Online localization of radio-tagged wildlife with an autonomous aerial robot system,” in Robotics: Science and Systems , 2015

  5. [13]

    Automated Individual Cattle Identification Using Video Data: A Unified Deep Learning Architecture Approach,

    Y . Qiao, C. Clark, S. Lomax, H. Kong, D. Su, and S. Sukkarieh, “Automated Individual Cattle Identification Using Video Data: A Unified Deep Learning Architecture Approach,” Frontiers in Animal Science, vol. 2, p. 759147, Nov. 2021

  6. [14]

    Automated aerial animal detection when spatial resolution conditions are varied,

    J. Brown, Y . Qiao, C. Clark, S. Lomax, K. Rafique, and S. Sukkarieh, “Automated aerial animal detection when spatial resolution conditions are varied,” Computers & Electronics in Agriculture, vol. 193, p. 106689, Feb. 2022

  7. [15]

    Improving daytime detection of deer for surveillance and management,

    T. W. Logan, A. Ashton-Butt, and A. I. Ward, “Improving daytime detection of deer for surveillance and management,” European Journal of Wildlife Research, vol. 65, p. 83, Oct. 2019

  8. [16]

    Deer surveillance system in public parks using deep learning,

    M. A. C. Rodriguez, “Deer surveillance system in public parks using deep learning,” Master’s thesis, National College of Ireland, 2022

  9. [17]

    What is YOLOv5: A deep look into the internal features of the popular object detector,

    R. Khanam and M. Hussain, “What is YOLOv5: A deep look into the internal features of the popular object detector,” arXiv:2407.20892 [cs.CV], 2024

  10. [18]

    An Automated Battery Management System to Enable Persistent Missions With Multiple Aerial Vehicles,

    N. K. Ure, G. Chowdhary, T. Toksoz, J. P. How, M. A. Vavrina, and J. Vian, “An Automated Battery Management System to Enable Persistent Missions With Multiple Aerial Vehicles,” IEEE/ASME Trans. Mechatronics, vol. 20, pp. 275–286, Feb. 2015

  11. [19]

    Auto- mated charging and docking station for security UA Vs,

    H. Nieuwoudt, J. Welgemoed, T. v. Niekerk, and R. Phillips, “Auto- mated charging and docking station for security UA Vs,” in 2023 14th International Conference on Mechanical and Intelligent Manufacturing Technologies (ICMIMT), pp. 32–38, May 2023

  12. [20]

    Precision landing for low-maintenance remote operations with UA Vs,

    M. Moreira et al. , “Precision landing for low-maintenance remote operations with UA Vs,” Drones, vol. 5, no. 4, p. 103, 2021

  13. [21]

    UA V positioning mecha- nisms in landing stations: Classification and engineering design review,

    M. Galimov, R. Fedorenko, and A. Klimchik, “UA V positioning mecha- nisms in landing stations: Classification and engineering design review,” Sensors, vol. 20, no. 13, p. 3648, 2020

  14. [22]

    Autocharge: Autonomous charging for perpetual quadrotor missions,

    A. Saviolo, J. Mao, R. Balu, V . Radhakrishnan, and G. Loianno, “Autocharge: Autonomous charging for perpetual quadrotor missions,” in 2023 IEEE Int. Conf. on Robotics and Automation (ICRA) , pp. 5400–5406, 2023

  15. [23]

    Design and implementation of drones charging station,

    S. Kokkinos, C. Mourgelas, E. Micha, E. Chatzistavrakis, and I. V oyi- atzis, “Design and implementation of drones charging station,” in Proc. 27th Pan-Hellenic Conference on Progress in Computing and Informatics, pp. 116–122, 2023

  16. [24]

    Rucervus dasuceveli Ravindra Tripathi ZDCD0 Dataset (v1),

    R. Tripathi, “Rucervus dasuceveli Ravindra Tripathi ZDCD0 Dataset (v1),” universe.robomow.com/deerproject/rucervusdauceveli, 2025. Ac- cessed May 2025

  17. [25]

    UB-ANC planner: Energy efficient coverage path planning with multiple drones,

    J. Modares, F. Ghanei, N. Mastronarde, and K. Dantu, “UB-ANC planner: Energy efficient coverage path planning with multiple drones,” in 2017 IEEE Int. Conf. on Robotics and Automation (ICRA) , pp. 6182–6189, June 2017

  18. [26]

    The traveling salesman problem: a guided tour of combinatorial optimization,

    E. L. Lawler, “The traveling salesman problem: a guided tour of combinatorial optimization,” Wiley-Interscience Series in Discrete Math- ematics, 1985

  19. [27]

    Distributed optimization by ant colonies,

    A. Colorni, M. Dorigo, and V . Maniezzo, “Distributed optimization by ant colonies,” in Proc. First European Conf. on Artificial Life , vol. 142, pp. 134–142, Paris, France, 1991

  20. [28]

    Dorigo, Optimization, learning and natural algorithms , PhD thesis, Politecnico di Milano, 1992

    M. Dorigo, Optimization, learning and natural algorithms , PhD thesis, Politecnico di Milano, 1992

  21. [29]

    Ant colony optimization,

    M. Dorigo, M. Birattari, and T. St ¨utzle, “Ant colony optimization,” IEEE Computational Intelligence Magazine, vol. 1, no. 4, pp. 28–39, 2006

  22. [30]

    MAX-MIN ant system and local search for the traveling salesman problem,

    T. St ¨utzle and H. Hoos, “MAX-MIN ant system and local search for the traveling salesman problem,” in ICEC, pp. 309–314, IEEE, 1997

  23. [31]

    MAX–MIN ant system,

    T. St ¨utzle and H. Hoos, “MAX–MIN ant system,” Future Generation Computer Systems, vol. 16, no. 8, pp. 889–914, 2000

  24. [32]

    Traveling Salesman Problem with Ant Colony Optimization,

    V . Murugananthan et al., “Traveling Salesman Problem with Ant Colony Optimization,” in 2023 2nd International Conf. on Edge Computing and Applications (ICECAA), pp. 481–485, July 2023

  25. [33]

    Concentrated coverage path planning algorithm of UA V formation for aerial photography,

    Y . Cao, X. Cheng, and J. Mu, “Concentrated coverage path planning algorithm of UA V formation for aerial photography,” IEEE Sensors Journal, vol. 22, no. 11, pp. 11098–11111, Jun. 2022

  26. [34]

    A UA V-based roe deer fawn detection system,

    M. Israel, “A UA V-based roe deer fawn detection system,” Int. Archives Photogrammetry, Remote Sensing and Spatial Information Sciences, vol. XXXVIII-1/C22, pp. 51–55, 2012

  27. [35]

    A Coverage Path Planning Algorithm for Self-Organizing Drone Swarms,

    K. Bezas, G. Tsoumani, and K. Oikonomou, “A Coverage Path Planning Algorithm for Self-Organizing Drone Swarms,” in 2021 Int. Balkan Conf. on Communications and Networking (BalkanCom) , pp. 122–126, Sept. 2021

  28. [36]

    Coverage path planning for spraying drones,

    E. Vazquez-Carmona, J. Vasquez-Gomez, J. Herrera-Lozada, and M. Antonio-Cruz, “Coverage path planning for spraying drones,” Com- puters & Industrial Engineering, vol. 168, p. 108125, June 2022

  29. [37]

    Coverage Path Planning for UA V Based on Improved Back-and-Forth Mode,

    X. Mu, W. Gao, X. Li, and G. Li, “Coverage Path Planning for UA V Based on Improved Back-and-Forth Mode,” IEEE Access, vol. 11, pp. 114840–114854, 2023

  30. [38]

    Coverage Path Planning Optimization of Heterogeneous UA Vs Group for Precision Agriculture,

    R. I. Mukhamediev et al. , “Coverage Path Planning Optimization of Heterogeneous UA Vs Group for Precision Agriculture,” IEEE Access, vol. 11, pp. 5789–5803, 2023

  31. [39]

    Coverage path planning with unmanned aerial vehicles for 3D terrain reconstruction,

    M. Torres, D. A. Pelta, J. L. Verdegay, and J. C. Torres, “Coverage path planning with unmanned aerial vehicles for 3D terrain reconstruction,” Expert Systems with Applications, vol. 55, pp. 441–451, Aug. 2016. Accepted to the Novel Approaches for Precision Agriculture and For...

Pith tools

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