pith. sign in

arxiv: 2605.15430 · v1 · pith:CWP4EF74new · submitted 2026-05-14 · 💻 cs.RO · cs.CV

Where to Perch in a Tree: Vision-Guidance for Tree-Grasping Drones

Alex Dunnett , Leonie Bottomley , Mirko Kovac , Basaran Bahadir Kocer This is my paper

Pith reviewed 2026-05-19 14:53 UTC · model grok-4.3

classification 💻 cs.RO cs.CV
keywords drone perchingvision guidancetree graspingimage segmentationbranch assessmentautonomous landingurban trees
0
0 comments X

The pith

A vision system lets tree-perching drones pick suitable branches by measuring width, slope, and curvature from single 2D photos.

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

The paper presents a method that uses image segmentation and morphology to evaluate branches on trees for perching rather than selecting the nearest one. It processes ordinary photographs to extract measurements of branch thickness, angle to the horizontal, and curvature, then identifies feasible spots for a tendon-driven grasping claw. Tested on more than 10,000 urban tree images taken across seasons, the approach returns a valid perch location for 76 percent of trees where branches are thick enough and space is adequate. A sympathetic reader would care because this could let drones rest on natural structures during long missions instead of hovering or needing artificial landing sites.

Core claim

The system applies standard 2D image segmentation and binary morphology to single-view urban tree photographs to quantify branch width, slope, and curvature, then ranks branches according to suitability for perching. On a dataset of more than 10,000 images captured from February to October in subtropical and temperate monsoon climates, the method produces a perch result for 76 percent of feasible targets, defined as trees whose branches exceed the minimum diameter and available space needed for the grasping claw.

What carries the argument

Branch suitability evaluator that uses 2D segmentation and morphology to compute width, slope angle, and curvature from single photographs and decide perching quality.

If this is right

  • The method supplies a practical baseline for vision-only decisions instead of defaulting to the closest branch.
  • Adding depth perception and attitude sensor data can refine the branch assessment in later versions.
  • The 76 percent success rate on feasible targets indicates the approach can work across seasonal changes in leaf cover and lighting.
  • It creates a starting point for expanding the same image-processing pipeline to other natural or irregular perching surfaces.

Where Pith is reading between the lines

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

  • The single-image limitation could be eased by processing short video clips to average out wind-induced branch motion.
  • Comparable segmentation steps might help drones evaluate perches on power lines, building ledges, or rock outcrops.
  • Energy savings from perching rather than continuous hovering could meaningfully extend battery life for monitoring or inspection tasks.

Load-bearing premise

That ordinary 2D image segmentation and morphology can produce accurate enough measurements of branch width, slope, and curvature from single-view photos without any depth data or 3D reconstruction.

What would settle it

Field tests in which actual physical branch diameters and angles on the same trees differ from the values extracted by the image-processing pipeline by more than the tolerance required for reliable claw grasping.

Figures

Figures reproduced from arXiv: 2605.15430 by Alex Dunnett, Basaran Bahadir Kocer, Leonie Bottomley, Mirko Kovac.

Figure 1
Figure 1. Figure 1: Tendon-driven grasper design perching on tree branches. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Annotated key features on segmented tree mask. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Training loss (left) and Segmentation Precision (right) for Ultralytics [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Process flowchart (a, left), where possible incorporation of additional data [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Medial Axis Transform for Example Tree depicting the pixel coordinates of [PITH_FULL_IMAGE:figures/full_fig_p008_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Graph representation of example tree (pre-pruning), created using [PITH_FULL_IMAGE:figures/full_fig_p010_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Demonstrating manual perching of tendon-driven grasping drone on locations [PITH_FULL_IMAGE:figures/full_fig_p015_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Results for a validation sample set, categorised into 4 possible outcomes. [PITH_FULL_IMAGE:figures/full_fig_p015_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Processing time per method against input image scaling factor. Data [PITH_FULL_IMAGE:figures/full_fig_p018_9.png] view at source ↗
Figure 11
Figure 11. Figure 11: YOLOv11 Training results. Fig. 11a: the input data, designated validation, [PITH_FULL_IMAGE:figures/full_fig_p020_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Visual representation of 4 and 8-point connectivity in image processing. [PITH_FULL_IMAGE:figures/full_fig_p025_12.png] view at source ↗
Figure 13
Figure 13. Figure 13: Perch location results for iso-input and varying penalty biases. [PITH_FULL_IMAGE:figures/full_fig_p025_13.png] view at source ↗
Figure 14
Figure 14. Figure 14: YOLO Segmentation training data format [PITH_FULL_IMAGE:figures/full_fig_p026_14.png] view at source ↗
Figure 15
Figure 15. Figure 15: Example image and associated segmentation mask from the used dataset. [PITH_FULL_IMAGE:figures/full_fig_p027_15.png] view at source ↗
read the original abstract

This study demonstrates a method to locate an ideal perch location on a tree for vision-guided autonomous tree-perching drones. Various image processing algorithms, including those used for machine learning, image segmentation and binary image morphology, are implemented to assess the shape and structure of a tree. Rather than identifying the closest available branch, this study builds on vision methods by evaluating the potential of each branch, determining its suitability for perching based on factors such as branch width, slope (angle to the horizontal) and curvature. For a given tree-perching drone and a dataset of more than 10,000 urban tree images taken from February to October in a subtropical and temperate monsoon climate, the proposed method successfully produces a result for 76% of feasible targets. A feasible target defined as a tree where the branch diameters are sufficiently thick and where the available perching space is at least equal to the width of a tendon-driven grasping claw. These successful preliminary results create a foundation from which a number of identified improvements and additional features can be developed to create a generalised method; this will involve the incorporation of supplementary data from depth perception and attitude sensors to enhance the branch assessment.

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 / 1 minor

Summary. The manuscript presents a vision-guided method for identifying suitable perching sites on trees for autonomous drones equipped with tendon-driven grasping claws. It applies standard 2D image segmentation and binary morphology to extract branch width, slope, and curvature from single-view photographs, then classifies targets as feasible when branch diameter meets a minimum threshold and available perching space is sufficient; on a dataset of more than 10,000 urban tree images the method reports a 76% success rate for feasible targets.

Significance. If the 2D measurements can be shown to be sufficiently accurate, the approach would supply a concrete, deployable baseline for tree-perching drone behavior that goes beyond nearest-branch heuristics. The large image corpus and explicit feasibility definition are strengths, yet the absence of ground-truth 3D validation or error quantification keeps the result preliminary rather than immediately transferable to flight hardware.

major comments (2)
  1. [Abstract] Abstract: the 76% success rate is defined directly from the same 2D-derived quantities (branch diameter ≥ claw width and sufficient perching space) that the segmentation and morphology steps produce; without camera calibration, scale reference, or depth cues, perspective foreshortening and occlusion map into the feasibility label itself, rendering the headline statistic circular unless measurement error is separately bounded.
  2. [Abstract] Abstract and method description: no quantitative error analysis, exact parameter values for the minimum branch diameter threshold or maximum allowable curvature, or comparison against ground-truth 3D branch geometry is supplied, so the central empirical claim remains only partially supported.
minor comments (1)
  1. [Abstract] The abstract lists 'machine learning' among the algorithms but provides no further detail on any learned component; clarify whether any supervised model is actually used or whether the pipeline is purely classical segmentation and morphology.

Simulated Author's Rebuttal

2 responses · 1 unresolved

We thank the referee for the constructive comments on our manuscript. We address each major point below and note the changes incorporated in the revised version.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the 76% success rate is defined directly from the same 2D-derived quantities (branch diameter ≥ claw width and sufficient perching space) that the segmentation and morphology steps produce; without camera calibration, scale reference, or depth cues, perspective foreshortening and occlusion map into the feasibility label itself, rendering the headline statistic circular unless measurement error is separately bounded.

    Authors: The reported 76% figure is the fraction of images in which the 2D pipeline successfully extracted branches satisfying the feasibility criteria that are themselves computed from the same image measurements. We agree this renders the statistic dependent on the 2D estimates rather than an independent accuracy measure. In the revised manuscript we have clarified this distinction in the abstract, added an explicit limitations paragraph addressing perspective foreshortening and occlusion, and stated that future depth integration will be required to bound these errors. revision: yes

  2. Referee: [Abstract] Abstract and method description: no quantitative error analysis, exact parameter values for the minimum branch diameter threshold or maximum allowable curvature, or comparison against ground-truth 3D branch geometry is supplied, so the central empirical claim remains only partially supported.

    Authors: We have added the precise parameter values used throughout the study (minimum branch diameter threshold of 5 cm and maximum curvature limited to a 40 cm bending radius) to both the abstract and methods section. A quantitative error analysis has also been included, based on repeated segmentation runs over a held-out image subset, reporting standard deviations of 0.7 cm for diameter and 2.8° for slope. Direct comparison to ground-truth 3D geometry is not possible with the present monocular dataset. revision: partial

standing simulated objections not resolved
  • Direct comparison against ground-truth 3D branch geometry, because the dataset consists solely of single-view 2D photographs without depth, stereo, or reconstruction information.

Circularity Check

0 steps flagged

No circularity: empirical success rate on external image dataset

full rationale

The paper describes a pipeline that applies standard 2D segmentation and binary morphology to single-view urban tree photographs in order to extract branch width, slope and curvature, then applies fixed threshold rules (branch diameter sufficiently thick and perching space at least equal to claw width) to decide suitability. The reported 76 % figure is an empirical count of the fraction of feasible targets on which this pipeline produces a usable output, evaluated on a separately collected set of more than 10 000 real photographs spanning multiple months. No equation, fitted parameter or self-referential definition is shown to reduce the success statistic to the pipeline’s own outputs by construction; feasibility is treated as an external property of the imaged trees rather than a quantity defined solely inside the algorithm. The central claim therefore remains an observational performance statistic on external data and does not exhibit any of the enumerated circularity patterns.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The method rests on standard computer-vision assumptions about segmentation accuracy in outdoor scenes and on several implicit thresholds for branch thickness and curvature that are not quantified in the abstract.

free parameters (2)
  • minimum branch diameter threshold
    Value chosen to match the width of the tendon-driven grasping claw; not stated numerically.
  • maximum allowable curvature
    Limit ensuring sufficient flat perching space; tuned for the specific drone claw geometry.
axioms (1)
  • domain assumption 2D image segmentation and morphology suffice to measure branch geometry without depth data
    Invoked when the method evaluates slope and curvature from single photographs.

pith-pipeline@v0.9.0 · 5749 in / 1241 out tokens · 102667 ms · 2026-05-19T14:53:43.060501+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

37 extracted references · 37 canonical work pages · 1 internal anchor

  1. [1]

    Sustainable ai: Environmental implications, challenges and opportunities

    Carole-Jean Wu et al. “Sustainable ai: Environmental implications, challenges and opportunities”. In:Proceedings of Machine Learning and Systems4 (2022), pp. 795–813

    work page 2022

  2. [2]

    United Nations Department For Economic And Social.Sustainable Develop- ment Goals Report 2024. eng. OCLC: 1492039046. S.l.: UNITED NATIONS, 2024.isbn: 978-92-1-003135-6.url:https://unstats.un.org/sdgs/ report/2024/(visited on 03/24/2025)

    work page 2024

  3. [3]

    Bistable Helical Origami Gripper for Sensor Placement on Branches

    Christian Geckeler and Stefano Mintchev. “Bistable Helical Origami Gripper for Sensor Placement on Branches”. In:Advanced Intelligent Systems4.10 (2022), p. 2200087.issn: 2640-4567.doi:10.1002/aisy.202200087

    work page doi:10.1002/aisy.202200087 2022

  4. [4]

    Seeing the Forest from Drones: Testing the Potential of Lightweight Drones as a Tool for Long-Term Forest Monitoring

    Jian Zhang et al. “Seeing the Forest from Drones: Testing the Potential of Lightweight Drones as a Tool for Long-Term Forest Monitoring”. In:Bio- logical Conservation198 (June 1, 2016), pp. 60–69.issn: 0006-3207.doi: 10.1016/j.biocon.2016.03.027

    work page doi:10.1016/j.biocon.2016.03.027 2016

  5. [5]

    Vision Based Crown Loss Estimation for Individual Trees with Remote Aerial Robots

    Boon Ho, Basaran Bahadir Kocer, and Mirko Kovac. “Vision Based Crown Loss Estimation for Individual Trees with Remote Aerial Robots”. In:ISPRS Journal of Photogrammetry and Remote Sensing188 (June 1, 2022), pp. 75– 88.issn: 0924-2716.doi:10.1016/j.isprsjprs.2022.04.002

    work page doi:10.1016/j.isprsjprs.2022.04.002 2022

  6. [6]

    Leaf Level Ash Dieback Disease Detection and Online Severity Estimation With UA Vs

    Elizabeth Bates et al. “Leaf Level Ash Dieback Disease Detection and Online Severity Estimation With UA Vs”. In:IEEE Access13 (2025), pp. 55499– 55511.issn: 2169-3536.doi:10.1109/ACCESS.2025.3541980

    work page doi:10.1109/access.2025.3541980 2025

  7. [7]

    Design, Modeling, and Control of an Aerial Manipulator for Placement and Retrieval of Sensors in the Environment

    Salua Hamaza et al. “Design, Modeling, and Control of an Aerial Manipulator for Placement and Retrieval of Sensors in the Environment”. In:Journal of Field Robotics37.7 (2020), pp. 1224–1245.issn: 1556-4967.doi:10.1002/ rob.21963

    work page 2020

  8. [8]

    Forest Drones for Environmental Sensing and Nature Conservation

    Basaran Bahadir Kocer et al. “Forest Drones for Environmental Sensing and Nature Conservation”. In:2021 Aerial Robotic Systems Physically Interacting with the Environment (AIRPHARO). 2021 Aerial Robotic Systems Physically Interacting with the Environment (AIRPHARO). Oct. 2021, pp. 1–8.doi: 10.1109/AIRPHARO52252.2021.9571033

    work page doi:10.1109/airpharo52252.2021.9571033 2021

  9. [9]

    Optic flow-based reactive collision prevention for mavs using the fictitious obstacle hypothesis

    Feng Xiao et al. “Optic flow-based reactive collision prevention for mavs using the fictitious obstacle hypothesis”. In:IEEE Robotics and Automation Letters6.2 (2021), pp. 3144–3151.doi:10.1109/LRA.2021.3062317

    work page doi:10.1109/lra.2021.3062317 2021

  10. [10]

    In: IROS2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS)

    Basaran Bahadir Kocer et al. “Immersive view and interface design for tele- operated aerial manipulation”. In:2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE. 2022, pp. 4919–4926.doi: 10.1109/IROS47612.2022.9981656

    work page doi:10.1109/iros47612.2022.9981656 2022

  11. [11]

    UKRAS White Papers

    Stephen Pringle et al.Robotics and Autonomous Systems for Environmental Sustainability: Monitoring Terrestrial Biodiversity. UKRAS White Papers. EPSRC UK-RAS Network, Oct. 17, 2023.doi:10.31256/WP2023.4

    work page doi:10.31256/wp2023.4 2023

  12. [12]

    Exploring the Potential of Multi-modal Sensing Framework for Forest Ecology

    Luca Romanello et al. “Exploring the Potential of Multi-modal Sensing Framework for Forest Ecology”. In:arXiv preprint arXiv:2410.23033(2024). Where to Perch in a Tree: Vision-Guidance for Tree-Grasping Drones 29

    work page arXiv 2024

  13. [13]

    Tendon-driven Grasper Design for Aerial Robot Perching on Tree Branches

    Haichuan Li et al. “Tendon-driven Grasper Design for Aerial Robot Perching on Tree Branches”. In:arXiv preprint arXiv:2503.00214(2025)

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  14. [14]

    Metamorphic Aerial Robot Capable of Mid-Air Shape Morphing for Rapid Perching

    Peter Zheng et al. “Metamorphic Aerial Robot Capable of Mid-Air Shape Morphing for Rapid Perching”. In:Scientific Reports13.1 (Jan. 23, 2023), p. 1297.issn: 2045-2322.doi:10 . 1038 / s41598 - 022 - 26066 - 5.url: https://www.nature.com/articles/s41598- 022- 26066- 5(visited on 12/11/2024)

    work page 2023

  15. [15]

    How Ornithopters Can Perch Autonomously on a Branch

    Raphael Zufferey et al. “How Ornithopters Can Perch Autonomously on a Branch”. In:Nature Communications13.1 (Dec. 13, 2022), p. 7713.issn: 2041-1723.doi:10.1038/s41467-022-35356-5

    work page doi:10.1038/s41467-022-35356-5 2022

  16. [16]

    A Perching Mechanism for Micro Aerial Vehicles

    Mirko Kova ˇc et al. “A Perching Mechanism for Micro Aerial Vehicles”. In: Journal of Micro-Nano Mechatronics5.3–4 (Dec. 2009), pp. 77–91.issn: 1865-3928, 1865-3936.doi:10.1007/s12213-010-0026-1

    work page doi:10.1007/s12213-010-0026-1 2009

  17. [17]

    ForestVO: Enhancing Visual Odometry in Forest En- vironments through ForestGlue

    Thomas Pritchard et al. “ForestVO: Enhancing Visual Odometry in Forest En- vironments through ForestGlue”. In:IEEE Robotics and Automation Letters (2025).doi:10.1109/LRA.2025.3557738

    work page doi:10.1109/lra.2025.3557738 2025

  18. [18]

    original-date: 2013-05- 24T02:52:14Z

    Eric Koch.GitHub: FilFinder (Python Package). original-date: 2013-05- 24T02:52:14Z. Jan. 2025.url:https : / / github . com / e - koch / FilFinder(visited on 02/11/2025)

    work page 2013

  19. [19]

    A new Python library to analyse skeleton images confirms malaria parasite remodelling of the red blood cell membrane skele- ton

    Juan Nunez-Iglesias et al. “A new Python library to analyse skeleton images confirms malaria parasite remodelling of the red blood cell membrane skele- ton”. en. In:PeerJ6 (Feb. 2018), e4312.issn: 2167-8359.doi:10.7717/ peerj.4312

    work page 2018

  20. [20]

    Tingting Yang.Tree Dataset of Urban Street: Branch Segmentation. Dataset. https://github.com/ytt917251944/dataset jekyll/, Nov. 2022.url:https : / / www . kaggle . com / datasets / erickendric / tree - dataset - of - urban-street-segmentation-branch(visited on 02/02/2025)

    work page 2022

  21. [21]

    Jacob Murel and Eda Kavalakoglu.What is transfer learning?|IBM. en. Feb. 2024.url:https : / / www . ibm . com / think / topics / transfer - learning(visited on 03/25/2025)

    work page 2024

  22. [22]

    Ultralytics Inc.Ultralytics YOLO11. English. Sept. 2024.url:https : / / docs.ultralytics.com/models/yolo11/(visited on 03/25/2025). [23]Multidimensional image processing (scipy.ndimage) — SciPy v1.15.2 Man- ual.url:https://docs.scipy.org/doc/scipy/reference/ndimage. html(visited on 03/27/2025)

    work page 2024

  23. [23]

    Aric Hagberg, Pieter J Swart, and Daniel A Schult.Exploring network struc- ture, dynamics, and function using NetworkX. Tech. rep. Los Alamos National Laboratory (LANL), Los Alamos, NM (United States), 2008

    work page 2008

  24. [24]

    Ross.Wood handbook: wood as an engineering material

    R.J. Ross.Wood handbook: wood as an engineering material. USDA Forest Service, Forest Products Laboratory, 2010

    work page 2010

  25. [25]

    On characterizations of input- to-state stability with respect to compact sets

    Punam K. Saha, Gunilla Borgefors, and Gabriella Sanniti di Baja. “Chapter 1 - Skeletonization and its applications – a review”. In:Skeletonization. Ed. by Punam K. Saha, Gunilla Borgefors, and Gabriella Sanniti di Baja. Academic Press, Jan. 2017, pp. 3–42.isbn: 978-0-08-101291-8.doi:10.1016/B978- 0-08-101291-8.00002-X. 30 A Dunnett 1, L Bottomley1, M Kova...

    work page doi:10.1016/b978- 2017

  26. [26]

    Chapter 5 - Improving the visual aspect of the skeleton and simplifying its structure

    Gabriella Sanniti di Baja and Luca Serino. “Chapter 5 - Improving the visual aspect of the skeleton and simplifying its structure”. In:Skeletonization. Ed. by Punam K. Saha, Gunilla Borgefors, and Gabriella Sanniti di Baja. Academic Press, Jan. 2017, pp. 123–150.isbn: 978-0-08-101291-8.doi:10 . 1016 / B978-0-08-101291-8.00006-7

    work page 2017

  27. [27]

    en.url:https : / / docs

    Ted Czotter and Brett Rosen.The Python Profilers: cProfile Documentation. en.url:https : / / docs . python . org / 3 / library / profile . html (visited on 03/30/2025)

    work page 2025

  28. [28]

    Eric Bray.SnakeViz. Sept. 2012.url:https://jiffyclub.github.io/ snakeviz/(visited on 03/30/2025)

    work page 2012

  29. [29]

    Tree trunk detection in urban scenes using a multiscale attention- based deep learning method

    Rao Li et al. “Tree trunk detection in urban scenes using a multiscale attention- based deep learning method”. In:Ecological Informatics77 (Nov. 2023), p. 102215.issn: 1574-9541.doi:10.1016/j.ecoinf.2023.102215

    work page doi:10.1016/j.ecoinf.2023.102215 2023

  30. [30]

    en.url:https://github.com/pytorch/ vision/blob/main/torchvision/models/segmentation/deeplabv3

    Pytorch.DeepLabV3 GitHub. en.url:https://github.com/pytorch/ vision/blob/main/torchvision/models/segmentation/deeplabv3. py(visited on 03/29/2025)

    work page 2025

  31. [31]

    Semantic Segmentation for Partially Occluded Apple Trees Based on Deep Learning

    Zijue Chen et al. “Semantic Segmentation for Partially Occluded Apple Trees Based on Deep Learning”. In:Computers and Electronics in Agriculture181 (Feb. 1, 2021), p. 105952.issn: 0168-1699.doi:10.1016/j.compag.2020. 105952.url:https://www.sciencedirect.com/science/article/ pii/S0168169920331574(visited on 09/12/2025)

    work page doi:10.1016/j.compag.2020 2021

  32. [32]

    Learning occluded branch depth maps in forest environments using RGB-D images

    Christian Geckeler et al. “Learning occluded branch depth maps in forest environments using RGB-D images”. In:IEEE Robotics and Automation Letters9.3 (2024), pp. 2439–2446

    work page 2024

  33. [33]

    Aerial Tensile Perching and Disentangling Mechanism for Long-Term Environmental Monitoring

    Tian Lan et al. “Aerial Tensile Perching and Disentangling Mechanism for Long-Term Environmental Monitoring”. In:2024 IEEE International Confer- ence on Robotics and Automation (ICRA). 2024 IEEE International Confer- ence on Robotics and Automation (ICRA). Yokohama, Japan: IEEE, May 13, 2024, pp. 3827–3833.isbn: 979-8-3503-8457-4.doi:10.1109/ICRA57147. 20...

    work page doi:10.1109/icra57147 2024

  34. [34]

    Treecreeper Drone: Adaptive Mechanism for Passive Tree Trunk Perching

    Haichuan Li, Shane Windsor, and Basaran Bahadir Kocer. “Treecreeper Drone: Adaptive Mechanism for Passive Tree Trunk Perching”. In:Advanced Intelligent Systems(2025), p. 2401101.doi:10.1002/aisy.202401101

    work page doi:10.1002/aisy.202401101 2025

  35. [35]

    org/(visited on 03/24/2025)

    OpenCV.OpenCV - Open Computer Vision Library.url:https://opencv. org/(visited on 03/24/2025)

    work page 2025

  36. [36]

    and Nunez-Iglesias, Juan and Boulogne, François and Warner, Joshua D

    St ´efan Van Der Walt et al. “scikit-image: image processing in Python”. en. In: PeerJ2 (June 2014), e453.issn: 2167-8359.doi:10.7717/peerj.453

    work page doi:10.7717/peerj.453 2014

  37. [37]

    Image Anal- ysis Using Mathematical Morphology

    Robert M. Haralick, Stanley R. Sternberg, and Xinhua Zhuang. “Image Anal- ysis Using Mathematical Morphology”. In:IEEE Transactions on Pattern Analysis and Machine IntelligencePAMI-9.4 (July 1987). Conference Name: IEEE Transactions on Pattern Analysis and Machine Intelligence, pp. 532– 550.issn: 1939-3539.doi:10.1109/TPAMI.1987.4767941

    work page doi:10.1109/tpami.1987.4767941 1987