Pith. sign in

REVIEW 2 major objections 4 minor 33 references

Blimp-based Crime Scene Analysis

T0 review · 2 major / 4 minor · reviewed 2026-08-16 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read A helium blimp carrying cameras and sensors can document indoor crime-scene evidence while generating essentially no airflow, unlike propeller drones.

desk verdict A modest, honest proof of concept; the feasibility claim is believable but the 'little airflow' headline rests on a single uncalibrated anemometer reading. read the letter →

arxiv 2504.15962 v2 pith:EFSFN6C5 submitted 2025-04-22 cs.RO

Martin Cooney , Fernando Alonso-Fernandez This is my paper · ORCID
classification cs.RO
keywords blimpcrimesceneanalysisindoorroboticstraceevidenceairflowdisturbancethermalimagingphotogrammetrycoveragepathplanning
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

The paper tries to establish that a small helium blimp, used as a drifting 'floating camera,' can document indoor crime scene evidence without the airflow disturbance that makes conventional drones risky. It reports a proof-of-concept blimp equipped with an RGB camera, a thermal camera, three LiDARs, WiFi, and roughly 20 minutes of battery, and argues this is the first small indoor blimp to carry such a sensor set. The load-bearing measurement is that an anemometer at ground level read 0.0 m/s even when the blimp passed about 20 cm from mock evidence, whereas a DJI Ryze Tello drone at 1.2 m produced 0.6–0.8 m/s and scattered the evidence. If correct, blimp-based documentation could preserve trace evidence, blood spatter, and fingerprints while still yielding 3D maps, thermal traces, and automated detection results.

What carries the argument

The carrying object is the blimp itself, used as a 'floating camera' that drifts or hovers without running its propellers, so its wake is minimal. The argument rests on the disturbance comparison in Section 3.2.1: a Mastech MS6252A anemometer placed on the ground beside mock evidence read 0.0 m/s during close passes (about 20 cm), while a DJI Ryze Tello drone at 1.2 m registered 0.6–0.8 m/s and scattered the evidence. This zero-wind measurement is what distinguishes the blimp from prior drone-based crime scene documentation and motivates the rest of the proof-of-concept.

What would settle it

Repeat the pass with a finer anemometer or particle and smoke visualization at the evidence surface, and with the blimp commanded to correct a drift or veer as noted in Section 3.2.2; if the sensor records airflow above a threshold that disturbs a settled fiber or powder, the little-airflow premise is falsified. A simpler check is to measure the wake during a propeller correction burst rather than only during near-free drift.

Watch

Extended reading notes

Core claim

The central claim is that blimps can be used to observe crime scene evidence while generating little airflow. In the paper's own test, the wind sensor consistently read 0.0 m/s even when the blimp passed close, about 20 cm, to the evidence; the comparison drone flying at 1.2 m generated 0.6 to 0.8 m/s and caused most of the mock evidence to scatter. The authors present this as confirmation of their core premise and combine it with demonstrations that the platform can capture most objects in view (5.2 of 7, or 74.3%, on average), generate a 3D textured mesh from its video in about three minutes, reveal recently touched objects as thermal traces, and support automated object detection and bloodstain classification at 74.1% accuracy on a single test image.

Load-bearing premise

The entire disturbance claim rests on one anemometer reading of 0.0 m/s beside the evidence, which assumes the instrument can resolve airflow levels that would matter to trace evidence, that ground-level air beside the evidence represents the flow reaching the evidence, and that the blimp did not need propeller bursts to correct drift during the passes.

Editorial extensions

If this is right

  • Crime scene investigators could document blood spatter, fingerprints, and trace evidence from close range before contamination, since the blimp's presence does not measurably stir the air.
  • A single proof-of-concept platform can carry RGB, thermal, LiDAR, and communication hardware with about 20 minutes of operating time, so multi-modal documentation is feasible indoors.
  • Blimp video can be turned into a 3D textured mesh within minutes, giving investigators an inspectable model rather than hours of footage.
  • Automated evidence detection and bloodstain typing (passive, active, or transfer) are within reach, with a generic foundation model correctly finding all objects in the test image.
  • Coverage path planning for an autonomous blimp can be framed around speed, overlap, navigability, efficiency, interaction, and adaptability, with simple snaking as a baseline.

Reading between the lines

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

  • The zero-wind result likely holds only while the blimp drifts without thrust; any propeller burst for stabilization would add airflow, so a production system would need a control mode that avoids corrections near evidence.
  • The same little-airflow property transfers naturally to other settings where drones are problematic for wind reasons: cleanrooms, munitions or chemical plants, nuclear disaster sites, and habitats of wind- or noise-sensitive animals.
  • The 74.1% bloodstain classification came from color-picking and ellipse geometry on a single heap image; a real dataset and learned model would be needed before treating the three-way typing as validated.
  • The 74.3% object-capture rate suggests manual piloting is workable but imperfect; an autonomy loop that re-visits missed regions could close the gap without adding wind risk.
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

2 major / 4 minor

Summary. arXiv:2504.15962 presents an exploratory proof-of-concept for using a small helium blimp as a non-disruptive 'floating camera' to document indoor crime scenes. The platform carries a Raspberry Pi Zero 2 W, an ESP32-Cam, three TF-Luna LiDARs, and an MLX90640 thermal camera, with roughly 20 minutes of battery life. Experiments address airflow disturbance (anemometer reading 0.0 m/s near evidence vs. 0.6–0.8 m/s for a DJI Tello), manual control (74.3% of objects captured in view across ten trials), 3D mapping via photogrammetry, thermal fingerprinting, generic object detection, a threshold-based bloodstain classifier (reported 74.1% accuracy on 27 contours), and a discussion of coverage path planning. The authors position the work as rapid prototyping and explicitly list limitations, and they release code and video.

Significance. If substantiated, the core claim that a blimp can document evidence while producing negligible airflow addresses a genuine forensic concern, namely downwash from rotary drones disturbing trace evidence. The paper also offers a useful low-cost reference platform and a concrete, falsifiable measurement protocol (the anemometer comparison). The manuscript is honest about its exploratory scope and provides open code/video. However, the key airflow measurement and the bloodstain classifier lack the controls needed to support the paper's stated conclusions; the contribution is currently a promising design exploration rather than a validated system.

major comments (2)
  1. [§3.2.1] The central premise 'that such blimps can be used to observe crime scene evidence while generating little airflow' rests on the statement that the Mastech MS6252A anemometer 'consistently read 0.0 m/s' at ~20 cm from evidence. The manuscript does not report the instrument's specified resolution, accuracy, or starting threshold, nor does it state whether the blimp's propellers were active during the pass. Many vane anemometers have starting thresholds of 0.1–0.3 m/s, so a displayed 0.0 can be a sensor floor rather than true near-zero airflow, and a propeller correction burst during the pass would not be captured by the reported result. The drone comparison (0.6–0.8 m/s) calibrates only the high end of the sensor. Please report the sensor specification, a defined airflow criterion, the motor/PWM state during measurement, and ideally repeated trials with the sensor at multiple positions and evidence types; alternatively, temper the abstract and Section 4 claims to say the blimp can produce lower airflow than a typical drone.
  2. [§3.2.6] The bloodstain classification result of 74.1% (20 of 27 contours) is obtained by tuning color-range, area, and eccentricity thresholds on the same single photograph that is then evaluated. As presented, this is a training-on-test evaluation, so the reported accuracy is not an estimate of how the heuristic would perform on new images; the text even notes 'for classifying bloodstains, only a single image was used' in Section 4.1. The claim in Section 4 that 'the three main kinds of bloodstain were detected and classified with an accuracy of 74.1%' should be rephrased as a demonstration that the heuristic can be made to fit this example, and any accuracy figure should be explicitly labeled as in-sample, or the evaluation should use held-out images.
minor comments (4)
  1. [§3.1.1] In the hardware section, 'Various electronics components where used' should read 'were used'; additionally, Table 1 contains superscripts (16–22) that are never resolved in the submitted text.
  2. [§3.2.2] The 74.3% object-capture metric would benefit from a definition of 'captured in view' and a report of per-trial variability (e.g., min/max or standard deviation), since the text acknowledges variability due to drafts and veering.
  3. [§3.2.3] The 3D mapping result is based on a single video and the free software's 50-frame limit; please clarify what was evaluated (e.g., geometric accuracy or visual completeness) and how 'promising' was assessed.
  4. [References] Reference [14] gives a URL pointing to bmva-archive.org.uk for a War Zone article; this appears to be a typo, and the correct source should be provided. Also, several superscript URLs in the text (e.g., in Sections 1 and 3.1) are not collected into a footnote list in this version.

Circularity Check

1 steps flagged · score 2.0 of 10

The blimp feasibility and low-airflow claims rest on direct empirical measurement, not on fitted parameters or a self-citation chain; only the secondary bloodstain-classification accuracy reduces to an in-sample fit on its single input image.

  1. fitted input called prediction [Section 3.2.6 (Bloodstain Classification); limitation acknowledged in Section 4.1]
    "First, we used color-picking and contours to identify clusters of red pixels. Then, we assessed the area and eccentricity of an ellipse fitted to each detected contour of reasonable size. We assumed that typical transfer stains (e.g., from a hand or foot) could be large compared to individual droplets, that passive drops should appear round (exhibit low eccentricity), and that active drops should appear elongated (have high eccentricity), as illustrated in Fig. 5(d). This resulted in a classification accuracy of 74.1%, with 20 of 27 detected blood contours correctly classified."

    The decision thresholds implicit in 'contour of reasonable size' and in the area/eccentricity criteria are derived from the same single photograph of the evidence heap that is then scored as 20/27 correct. Because no held-out image, cross-validation, or second dataset is used, the 74.1% figure measures how well the hand-designed heuristic fits its own training input, not how well it predicts unseen bloodstains. The paper itself concedes this in Section 4.1: 'for classifying bloodstains, only a single image was used.' Presenting this in-sample performance as 'classification accuracy' therefore equates the reported result with the input that generated the thresholds.

full rationale

The paper's central claims are empirical rather than derivational: the feasibility of the blimp platform, the low-airflow observation, the 74.3% capture rate, the 3D map generation, and the thermal-sensing observations are all direct measurements made on the prototype. The airflow claim in Section 3.2.1 reports that an anemometer consistently read 0.0 m/s; this is an instrument observation whose possible calibration floor is a measurement-validity limitation, not a circular reduction. Likewise, the authors' self-citations in Sections 2 and 3 (previous blimp work, earlier YOLO use, and the prior crime-scene-task overview) provide background and motivation rather than load-bearing evidence: the present feasibility results would stand unchanged without them. The only genuine in-sample circularity is the bloodstain classifier's 74.1% accuracy, which is a secondary 'towards Automation' idea explicitly limited to a single image and not part of the paper's core proof-of-concept contribution. A score of 2 reflects that one minor, non-central fitted-input-as-prediction issue while recognizing that the main claims are independently supported by direct experimentation.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

No derivation is attempted, so the ledger is small. The main fitted content is the bloodstain heuristic, tuned and evaluated on one image. The remaining assumptions are domain-level (representativeness of mock evidence, validity of a single-point airflow reading, ability to drift without propeller bursts), all acknowledged as limitations in Section 4.1. No new physical entities, forces, or conserved quantities are introduced; the floating camera concept is a configuration of existing components.

free parameters (2)
  • Bloodstain contour thresholds (red color range, minimum area, eccentricity cutoffs) = not reported; chosen by inspection of one image
    Section 3.2.6 sets area and eccentricity criteria to separate transfer, passive, and active stains; the 74.1% accuracy is then measured on the same single image, so the thresholds are fitted to the evaluation data.
  • Mock crime scene generation probabilities (evidence types, locations) = not reported; custom probabilities in Python tool
    Section 3.1.2 describes a simulator with customizable probabilities; these parameters shape test layouts but are not load-bearing for the central feasibility or airflow claims.
assumptions (4)
  • domain assumption An anemometer reading of 0.0 m/s at one ground position near evidence establishes negligible disturbance to sensitive evidence.
    Section 3.2.1: single device, single position, no stated resolution or accuracy, and no sampling duration; disturbance to fibers and fine particles may depend on airflow at the evidence surface, not just at ground level.
  • domain assumption Mock-up evidence (paper sheets with red dye E120, a photo of a firearm, two knives, shoes) is representative enough of real crime scene evidence to test disturbance and sensing.
    Sections 3.2 and 4.1: real trace evidence such as fibers and blood degrades and moves differently; the authors acknowledge that field evaluations with real scenes are needed.
  • domain assumption Hovering or slow drifting without propeller use is achievable near evidence in practice.
    Sections 3.2.2 and 4.1: manual control was imperfect, with the blimp susceptible to drafts and imbalances, so recovery maneuvers might engage propellers and generate airflow not captured in the 0.0 m/s reading.
  • standard math Standard computer vision primitives (contour detection, ellipse fitting, pretrained YOLO) behave as documented.
    Sections 3.2.5 and 3.2.6 rely on OpenCV contours, fitted ellipses, eccentricity, and a pretrained YOLOv8l model without re-deriving them.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Blimp-based Crime Scene Analysis." pith.science (2026). https://pith.science/paper/EFSFN6C5

@misc{pith2026250415962,
  author       = {Pith},
  title        = {Pith review of: Blimp-based Crime Scene Analysis},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EFSFN6C5}},
  note         = {Machine review of arXiv:2504.15962}
}
read the original abstract

Crime is a critical problem -- which often takes place behind closed doors, posing additional difficulties for investigators. To bring hidden truths to light, evidence at indoor crime scenes must be documented before any contamination or degradation occurs. Here, we address this challenge from the perspective of artificial intelligence (AI), computer vision, and robotics: Specifically, we explore the use of a blimp as a "floating camera" to drift over and record evidence with minimal disturbance. Adopting a rapid prototyping approach, we develop a proof-of-concept to investigate capabilities required for manual or semi-autonomous operation. Consequently, our results demonstrate the feasibility of equipping indoor blimps with various components (such as RGB and thermal cameras, LiDARs, and WiFi, with 20 minutes of battery life). Moreover, we confirm the core premise: that such blimps can be used to observe crime scene evidence while generating little airflow. We conclude by proposing some ideas related to detection (e.g., of bloodstains), mapping, and path planning, with the aim of stimulating further discussion and exploration.

Figures

Figures reproduced from arXiv: 2504.15962 by the authors.

Figure 1
Figure 1. Basic concept: a "floating camera" could be used to record sensitive evidence without destroying it. scene investigator’s job harder and can seriously damage investigations. Blood spatter patterns and fingerprints can be inadvertently smudged, footprints or tire treads walked on if care is not taken, and trace evidence can be scattered hither and yon by those unaware of its very presence".1 Furthermore, time is a cr… view at source ↗
Figure 2
Figure 2. Hardware: (a) Base of the lower gondola, showing sensor positions (b) gondola with 3 propellers, minicomputer, battery, and sensors on the left, and radio transmitter on the right [PITH_FULL_IMAGE:figures/full_fig_p008_2.png] view at source ↗
Figure 3
Figure 3. User interface: Green buttons control the blimp’s movements, blue buttons start threads to read sensors, and the red button enables recording. The text field shows LiDAR distances, and the videos show thermal (left) and RGB data (right). most search results seemed to point to story-writing tools (e.g., “Murder Scene Generator”).33 This was problematic, since, as the saying goes, "no two crime scenes are the same", b… view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Randomly generated crime scenes: (left) homicide at NFC Villa, (right) evidence heap filling a grid by size—larger items first. 3.2. Exploring Possibilities for Indoor CSA Blimps The developed platform was applied to check the questions identified in reviewing the lite…
Figure 5
Figure 5. Figure 5: Additional exploration: (a) photogrammetry/structure from motion, (b) detecting warm shoes that a criminal might have recently used, (c) detecting objects, (d) detecting blood pattern type (rectangles indicate ground truth, and contour colors indicate predicted class—r…

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

33 extracted references · 33 canonical work pages

  1. [1]

    URL: https://unstats.un.org/sdgs/ report/2023/The-Sustainable-Development-Goals-Report-2023.pdf, accessed: 2025-05-26

    The Sustainable Development Goals Report 2023, 2023. URL: https://unstats.un.org/sdgs/ report/2023/The-Sustainable-Development-Goals-Report-2023.pdf, accessed: 2025-05-26

    work page 2023

  2. [2]

    Hoeffler, What are the costs of violence?, Politics, Philosophy & Economics 16 (2017) 422–445

    A. Hoeffler, What are the costs of violence?, Politics, Philosophy & Economics 16 (2017) 422–445

    work page 2017

  3. [3]

    Sturup, D

    J. Sturup, D. Karlberg, M. Kristiansson, Unsolved homicides in Sweden: A population-based study of 264 homicides, Forensic science international 257 (2015) 106–113

    work page 2015

  4. [4]

    B. L. Benson, W. David, P. A. May, K. Harries, A. M. Rosenthal, A. Lorie, M. L. D. Fridell, G. Fisher-Stewart, J. Pedro, T. M. Saavedra, et al., Without a Trace? Advances in Detecting Trace Evidence, National Institute of Justice 249 (2003)

    work page 2003

  5. [5]

    Rozhok, Modeling of Innovative Lighter-than-Air UAV for Logistics, Surveillance, and Rescue Operations, Ph.D

    A. Rozhok, Modeling of Innovative Lighter-than-Air UAV for Logistics, Surveillance, and Rescue Operations, Ph.D. thesis, Università degli studi di Genova, 2023

    work page 2023

  6. [6]

    Zamfirescu-Pereira, D

    J. Zamfirescu-Pereira, D. Sirkin, D. Goedicke, R. Lc, N. Friedman, I. Mandel, N. Martelaro, W. Ju, Fake it to make it: Exploratory prototyping in HRI, in: Companion of the 2021 ACM/IEEE International Conference on Human-Robot Interaction, 2021, pp. 19–28

    work page 2021

  7. [7]

    Urbanová, M

    P. Urbanová, M. Jurda, T. Vojtíšek, J. Krajsa, Using drone-mounted cameras for on-site body documentation: 3D mapping and active survey, Forensic science international 281 (2017) 52–62

    work page 2017

  8. [8]

    Georgiou, P

    A. Georgiou, P. Masters, S. Johnson, L. Feetham, UAV-assisted real-time evidence detection in outdoor crime scene investigations, Journal of forensic sciences 67 (2022) 1221–1232

    work page 2022

Show all 33 references
  1. [9]

    Araujo, J

    P. Araujo, J. Fontinele, L. Oliveira, Multi-perspective object detection for remote criminal analysis using drones, IEEE Geoscience and Remote Sensing Letters 17 (2019) 1283–1286

    2019

  2. [10]

    Cooney, S

    M. Cooney, S. Ponrajan, F. Alonso-Fernandez, Nano Drone-based Indoor Crime Scene Analysis, 2025 IEEE International Conference on Advanced Robotics and its Social Impacts (ARSO) (accepted April 29, 2025; arXiv preprint arXiv:2502.21019) (2025)

    2025 arXiv

  3. [11]

    Bucknell, T

    A. Bucknell, T. Bassindale, An investigation into the effect of surveillance drones on textile evidence at crime scenes, Science & justice 57 (2017) 373–375

    2017

  4. [12]

    L. W. Murphy, C. Calabrese, J. Stanley, C. Crump, The Future of Drones in America: Law Enforcement and Privacy Considerations, ACLU Statement for the Record for a Senate Judiciary Committee Hearing, American Civil Liberties Union 20 (2013)

    2013

  5. [13]

    Saiki, T

    H. Saiki, T. Fukao, T. Urakubo, T. Kohno, A path following control method under wind disturbances for outdoor blimp robots, in: 2011 IEEE/SICE International Symposium on System Integration (SII), IEEE, 2011, pp. 978–984

    2011

  6. [14]

    Trevithick, Balloon-Launched Drone To Intercept Long Range Kamikaze Drones Emerges In Ukraine, The War Zone (TMZ) (2025)

    J. Trevithick, Balloon-Launched Drone To Intercept Long Range Kamikaze Drones Emerges In Ukraine, The War Zone (TMZ) (2025). URL: https://bmva-archive.org.uk/bmvc/2017/ papers/paper188/index.html

    2025

  7. [15]

    Cooney, F

    M. Cooney, F. Zanlungo, S. Nishio, H. Ishiguro, Designing a flying humanoid robot (FHR): effects of flight on interactive communication, in: 2012 IEEE RO-MAN: The 21st IEEE International Symposium on Robot and Human Interactive Communication, IEEE, 2012, pp. 364–371

    2012

  8. [16]

    St-Onge, C

    D. St-Onge, C. Gosselin, N. Reeves, Dynamic modelling and control of a cubic flying blimp using external motion capture, Proceedings of the Institution of Mechanical Engineers, Part I: Journal of Systems and Control Engineering 229 (2015) 970–982

    2015

  9. [17]

    Yao, Q.-y

    N.-s. Yao, Q.-y. Tao, W.-y. Liu, Z. Liu, Y. Tian, P.-y. Wang, T. Li, F. Zhang, Autonomous flying blimp interaction with human in an indoor space, Frontiers of Information Technology & Electronic Engineering 20 (2019) 45–59

    2019

  10. [18]

    S. U. Ferdous, A. Mohammadi, S. Lakshmanan, Developing a low-cost autonomous blimp with a reduced number of actuators, in: Unmanned Systems Technology XXI, volume 11021, SPIE, 2019, pp. 73–80

    2019

  11. [19]

    Seguin, J

    L. Seguin, J. Zheng, A. Li, Q. Tao, F. Zhang, A deep learning approach to localization for navigation on a miniature autonomous blimp, in: 2020 IEEE 16th International Conference on Control & Automation (ICCA), IEEE, 2020, pp. 1130–1136

    2020

  12. [20]

    H. Q. Pham, S. Singh, M. Garratt, S. Ravi, Controlling a bio-inspired miniature blimp using a depth sensing neural-network camera, Bioinspiration & Biomimetics 19 (2024) 024001

    2024

  13. [21]

    Huang, J

    H.-W. Huang, J. Chen, P. Rupp, C. Ehmke, P. R. Chai, R. Dhar, I. Ballinger, G. Traverso, Cost-Effective Blimp for Autonomous and Continuous Vital Signs Monitoring, in: 2024 IEEE International Conference on Advanced Intelligent Mechatronics (AIM), IEEE, 2024, pp. 1553–1559

    2024

  14. [22]

    M. Xu, J. Shao, Y. Ju, X. Shen, Q. Gao, W. Chen, Q. Zhang, Y. S. Pai, G. Barbareschi, M. Hoppe, et al., Cuddle-Fish: Exploring a Soft Floating Robot with Flapping Wings for Physical Interactions, arXiv preprint arXiv:2504.01293 (2025)

    2025

  15. [23]

    D. Hong, Y. Tanaka, Buoyant choreographies: Harmonies of light, sound, and human connection, in: IEEE International Conference on Robotics and Automation (ICRA) 2025, IEEE, 2025

    2025

  16. [24]

    S. S. Bhat, S. G. Anavatti, M. Garratt, S. Ravi, Review of autonomous outdoor blimps and their applications, Drone Systems and Applications 12 (2024) 1–21

    2024

  17. [25]

    Esposito, F

    M. Esposito, F. Sessa, G. Cocimano, P. Zuccarello, S. Roccuzzo, M. Salerno, Advances in Technologies in Crime Scene Investigation, Diagnostics 2023, 13(20), 3169. DOI org/10.3390/diagnostics13203169 (2023)

    2023 doi

  18. [26]

    J. D. Muñoz, J. Ruiz-Santaquiteria, O. Deniz, G. Bueno, Concealed Weapon Detection Using Thermal Cameras, Journal of Imaging 11 (2025) 72

    2025

  19. [27]

    Cooney, L

    M. Cooney, L. M. W. Klasén, F. Alonso-Fernandez, Designing Robots to Help Women, in: 14th Scandinavian Conference on Artificial Intelligence (SCAI 2024): AI for a better society, June 10-11, 2024, Jönköping, Sweden, IEEE, 2024, pp. 168–177

    2024

  20. [28]

    Bergman, M

    T. Bergman, M. Klöden, J. Dreßler, D. Labudde, Automatic Classification of Bloodstains with Deep Learning Methods, KI-Künstliche Intelligenz 36 (2022) 135–141

    2022

  21. [29]

    Jonnarth, J

    A. Jonnarth, J. Zhao, M. Felsberg, Learning coverage paths in unknown environments with deep reinforcement learning, arXiv preprint arXiv:2306.16978 (2023)

    2023 arXiv

  22. [30]

    Sörme, T

    J. Sörme, T. Edwards, A comparison of path planning algorithms for robotic vacuum cleaners, 2018

    2018

  23. [31]

    W. Moon, B. Park, S. H. Nengroo, T. Kim, D. Har, Path planning of cleaning robot with reinforcement learning, in: 2022 IEEE International Symposium on Robotic and Sensors Environments (ROSE), IEEE, 2022, pp. 1–7

    2022

  24. [32]

    Galanakis, X

    G. Galanakis, X. Zabulis, T. Evdaimon, S.-E. Fikenscher, S. Allertseder, T. Tsikrika, S. Vrochidis, A study of 3D digitisation modalities for crime scene investigation, Forensic sciences 1 (2021) 56–85

    2021

  25. [33]

    Vajgl, P

    M. Vajgl, P. Hurtik, T. Nejezchleba, Dist-YOLO: Fast Object Detection with Distance Estimation, Applied Sciences 12 (2022). URL: https://www.mdpi.com/2076-3417/12/3/1354. doi:10.3390/app12031354

    2022 doi

Pith tools

Reviewed August 16, 2026 · model on record in the stance chip above.