Pith. sign in

REVIEW 4 major objections 5 minor 27 references

Miniature multihole airflow sensor for lightweight aircraft over wide speed and angular range

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

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

Pith's one-line read A 9 g multihole probe estimates airspeed, angle of attack, and sideslip from 3 to 27 m/s and ±35° using a third-degree polynomial calibration.

desk verdict A genuinely lightweight 9 g multihole probe with a systematic design study; the headline wind-tunnel MAEs are probably optimistic because the test set includes interpolated points, but the engineering contribution is solid and deserves careful peer review. read the letter →

arxiv 2505.03331 v1 pith:YG5QZRWB submitted 2025-05-06 cs.RO

Lukas Stuber , Simon Jeger , Raphael Zufferey , Dario Floreano This is my paper · ORCID
classification cs.RO
keywords multiholepressureprobeairspeedestimationangleofattacksideslipstalldetectionlightweightUAVdifferentialsensorspolynomialregression
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 tries to close a gap in small-drone instrumentation: existing multihole pressure probes are either heavier than 30 g or need airspeeds above 20 m/s, while lightweight fixed-wing UAVs often fly below 15 m/s and are most at risk near stall. The authors propose a 9 g single-component probe that measures pressure differences between paired holes with five differential sensors, calibrate it with a third-degree multivariate polynomial over 3-27 m/s and ±35° angles, and report mean absolute errors of 0.44 m/s for airspeed and 0.16° for angle. They also test conical versus spherical tips and four hole spacings, concluding that cone tips give better resolution and lower noise while hole spacing has no significant effect. If the claim is right, a lightweight UAV can carry one cheap, public-domain sensor instead of a pitot tube plus two vanes, and stall detection becomes feasible at low speeds.

What carries the argument

The load-bearing object is the five-channel differential pressure pattern: four pairs of peripheral holes angled at 45° plus one static-to-tip pair, each read by a differential pressure sensor, so paired holes halve the required sensor count. The key identity is the scaling factor $q=\sqrt{\Delta P_1^2+\Delta P_2^2+\Delta P_3^2+\Delta P_4^2+\Delta P_5^2}$; dividing each pressure difference by $q$ removes most of the airspeed dependence, and three separate third-degree polynomial regressions then map the scaled pressures to airspeed, angle of attack, and sideslip. This $q$-scaling is what lets a model trained over 3-27 m/s and ±35° generalize across the tested flight envelope, and the design comparison is what justifies the final cone-tip, 1.2 mm-spacing configuration.

What would settle it

Recalibrate all three polynomials on only the 17 measured wind-tunnel configurations, without the interpolated artificial points, then evaluate on newly measured wind-tunnel points across the same 3-27 m/s and ±35° grid; if the mean absolute error remains near 0.44 m/s and 0.16°, the central claim stands, and if it rises substantially, the headline accuracy is an artifact of the interpolation.

Watch

Extended reading notes

Core claim

The paper's central claim is that a 9 g, single-component, differentially sensed multihole probe can recover airspeed, angle of attack, and sideslip across 3-27 m/s and ±35° using only a calibrated multivariate polynomial, with a mean absolute error of 0.44 m/s and 0.16° in wind-tunnel tests and consistent behavior during outdoor stall and yaw maneuvers. The authors establish this through a design comparison showing conical tips outperform spherical tips while hole spacing has no significant effect, a preprocessing step that scales five pressure differences by a common factor to remove airspeed dependence, and third-degree polynomial models selected by an elbow method. Outdoor flights against an autopilot-based estimate, rather than a true ground truth, show the probe tracks airspeed, angle of attack, and sideslip, and it keeps working during maneuvers where a pitot tube alone degrades.

Load-bearing premise

The reported accuracy rests on the assumption that adding artificial calibration points created by linear interpolation before splitting the data does not make the test error look better than it truly is.

Editorial extensions

If this is right

  • Small fixed-wing UAVs can replace a pitot tube plus two angular vanes with one 9 g probe and still obtain three airflow quantities, reducing weight and mounting complexity.
  • Stall detection becomes practical at low flight speeds because the angle-of-attack estimate is accurate enough to indicate approach to stall before lift is lost.
  • The design result tells builders that cone tips are worth using, while hole spacing can be chosen by manufacturing convenience rather than accuracy.
  • Because calibration is only a third-degree polynomial, the sensor can run on modest autopilot hardware without requiring a neural network.
  • The public release of design and calibration files allows other groups to reproduce the sensor and extend it to higher angles or other airframes.

Reading between the lines

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

  • The reported wind-tunnel accuracy may be optimistic because linearly interpolated artificial points are added before the train/test split, and artificial points near the zero-angle regime are nearly combinations of measured training points; retraining on the 17 measured configurations alone would reveal the honest holdout error.
  • The $q$-scaling preprocessing is probe-agnostic, so other multihole probes that use differential pressure sensors could adopt it, potentially making low-speed polynomial calibration a standard workflow rather than a bespoke solution.
  • The 50 Hz measurement rate and the polynomial inverse map could be used online to estimate turbulence intensity, but the paper only demonstrates quasi-steady calibration, so that extension remains untested.
  • If the sensor truly operates at 3 m/s, it may also serve as a calibration or airflow tool for very slow indoor drones, a regime the paper does not explore.
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 paper presents a 9 g multihole pressure probe with five differential pressure sensors integrated into a single 3D-printed structure, avoiding external tubing. The authors compare eight hardware variants (two tip shapes and four hole spacings) in an open wind tunnel, select a conical-tip, 1.2 mm spacing design, calibrate it with a degree-3 multivariate polynomial regression over 3-27 m/s airspeed and +/-35 degrees angle of attack and sideslip, and validate the sensor in outdoor flights on a fixed-wing UAV. The headline results are a mean absolute error of 0.44 m/s for airspeed and 0.16 degrees for angles in the wind-tunnel test set, with larger errors in outdoor flight, and the authors argue the sensor fills a gap for lightweight UAVs flying near the stall regime.

Significance. If the reported wind-tunnel accuracy is reliable, this is a useful contribution: a lightweight, open-source multihole probe that operates at low airspeeds would address a real gap, since existing miniature probes typically require airspeeds above 20 m/s. The hardware design study (eight variants), the public release of the design and data repository, and the outdoor flight validation are all strengths. However, the central accuracy claim is weakened by the calibration protocol: linearly interpolated artificial points are added before the train/test split, which can artificially lower the reported test MAE. The outdoor validation also relies on non-ground-truth references. The contribution is promising, but the evidence for the headline accuracy numbers needs revision.

major comments (4)
  1. [Section III-B, Fig. 7A] The reported test MAE (0.44 m/s, 0.163 degrees, 0.156 degrees) is computed after adding linearly interpolated artificial points to the regression dataset before the 70/30 shuffled split. Because each artificial point is a convex combination of nearby measured points, an artificial point assigned to the test set can be predicted almost exactly from its parent measured points if those parents are in the training set. The test error therefore partly measures how well the model reproduces linearly interpolated targets rather than generalization to unseen flow conditions. Please recompute and report the MAE and RMSE using only the original measured points for testing, and state explicitly whether any augmented points are excluded from the test subset.
  2. [Section III-B, Fig. 7B] The manuscript does not specify the unit of the regression dataset used for the 70/30 split: whether each sample is the averaged static portion of a 20 s measurement for one (speed, AoA, AoS) configuration, or whether the 504,900 individual filtered time samples are used. If individual time samples from the same static segment appear in both training and test sets, the split is not independent and the reported errors are biased downward regardless of the interpolation issue. Please clarify the number and granularity of samples entering the regression and, if time samples are used, perform the split at the configuration level.
  3. [Section IV, Eqs. (2)-(3)] The outdoor validation references are not ground truth. The AoA and AoS estimates are derived from the Pixhawk autopilot's body-frame velocity estimate, which combines IMU, GPS, and pitot data and assumes zero wind. The statement in Section IV that the results 'demonstrate that our sensor and its calibration perform well in an unseen, applied outdoor setting and did not overfit' is stronger than the evidence supports. Please either validate against an independent reference (for example, a calibrated wind vane, motion capture, or a known maneuver) or explicitly qualify the outdoor results as relative comparisons rather than accuracy validation.
  4. [Section V] The conclusion states that the probe 'achieves an RMSE of 0.22 degrees for angle estimation and 3.6% for speed' and compares these numbers with literature, but no RMSE values or their computation basis are reported in Section III. The text reports only MAE for the test set. Please provide the RMSE values together with the same test-set definition used for the MAE, or remove the unsupported comparison.
minor comments (5)
  1. [Throughout] The term 'UA V' appears with an odd space in the abstract and introduction; use 'UAV' consistently.
  2. [Section II-A and Section V] The sensor is described as having 13 holes in Section II-A but as a 'nine-hole' multihole pressure probe in the conclusion; please reconcile the terminology for the number of holes.
  3. [Section III-B, Fig. 7C] The y-axis of the elbow plot is not labeled; it is unclear whether the plotted quantity is MAE, RMSE, or another error metric. Please label the axis and state the metric in the caption.
  4. [Table I] The notation in Table I (avg/std reductions over five dimensions) is difficult to parse. Consider presenting the four metrics as explicit equations or a clearer table with row and column headers.
  5. [References] References [2] and [3] cite Fandom and Wikiwand pages rather than archival accident reports or peer-reviewed sources; please replace them with authoritative references.

Circularity Check

1 steps flagged · score 6.0 of 10

Wind-tunnel MAE partly computed on linearly interpolated test points, so the headline 0.44 m/s and 0.16° accuracies are optimistically biased.

  1. fitted input called prediction [Section III-B, Fig. 7A and 7C]
    "The data exhibit linear behaviour at small angles of attack (AoA) and sideslip (AoS), which we reinforce by generating artificial data through linear interpolation (Fig. 7A) before fitting the multivariate polynomial regression model (Fig. 7B). The dataset is first shuffled and subsequently divided into 70% training and 30% testing subsets. ... When evaluated on the test subset the model achieved a mean absolute error (MAE) of 0.163◦ and 0.156◦ for the AoA and AoS estimation respectively and 0.44 m/s for air speed estimation."

    The test-set evaluation is performed after augmenting the data with linearly interpolated points and then randomly splitting the combined set. Each artificial point is a convex combination of nearby measured points; since the split is 70/30, an interpolated point in the test set almost certainly has its measured parents in the training set. A degree-3 polynomial fitted to those parents reproduces the linear combination almost exactly, especially in the near-zero-angle region the paper itself describes as linear. Thus the reported MAEs are partly computed on test points whose values are constructed from training data—the model is effectively being asked to predict its own interpolants.

full rationale

The core sensing pipeline is not circular: the scaling factor q is computed from the sensor's own differential pressures (Eq. 1), not from ground-truth labels, and the multivariate polynomial regression maps measured pressure features to wind-tunnel reference values, which is a standard supervised calibration. There is no load-bearing self-citation: the authors' prior work is cited only for morphing-drone context and for the CAD/data repository, not for any uniqueness theorem or modeling ansatz that constrains the result. The one genuine circularity-adjacent flaw is the holdout protocol in Section III-B: linearly interpolated artificial points are added before the 70/30 split, so the test set can contain points that are convex combinations of training measurements. Evaluating on such points does not measure generalization and likely deflates the headline MAEs. The outdoor flight validation is independent in spirit, but it uses autopilot estimates that the authors explicitly say do not provide ground truth, and its errors are larger. Score reflects partial circularity in the central accuracy claim, not a fully self-referential derivation.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The paper's central claim is empirical: a fitted polynomial plus a normalization scheme on measured pressures. The free parameters are the regression coefficients, the filter cutoff, and the unspecified interpolation augmentation. The key domain assumption is that q-normalization decouples speed from angle. No new physical entities are introduced.

free parameters (3)
  • Multivariate polynomial regression coefficients (three degree-3 models: airspeed, AoA, AoS) = Least-squares estimates (not listed in paper; deposited in Zenodo)
    These map the five normalized pressure differences to the three output quantities; they are the core fitted parameters.
  • Butterworth low-pass filter cutoff frequency = 10 Hz
    Chosen by hand in Section III-A for noise suppression; a different cutoff could change the calibration data.
  • Artificial linear interpolation points near zero angle = Not specified (count, spacing, and insertion rule are not given)
    Added in Section III-B before model fitting; their presence in the test set affects the reported MAE and is a hidden data-choice parameter.
assumptions (4)
  • domain assumption The Euclidean norm q of the five differential pressures (Eq. 1) fully removes the airspeed influence from the normalized pressure ratios.
    Invoked in Section III-A; if the decoupling is incomplete, especially at high angles or low speeds, the polynomial model will mix speed and angle errors.
  • domain assumption A degree-3 multivariate polynomial can approximate the normalized pressure to airspeed/angle mapping across 3-27 m/s and ±35°.
    Selected by elbow method in Section III-B; no physical model justifies this functional form, so it is an empirical assumption.
  • domain assumption Wind-tunnel measurements are representative of outdoor flight, and the Pixhawk/pitot references are accurate enough for validation in no-wind conditions.
    Underpins the outdoor validation in Section IV; the no-wind requirement is rarely satisfied in real operations and the references are not ground truth.
  • standard math Standard statistical methods (least squares, t-tests, Butterworth filtering, random shuffling) apply to this data.
    Used throughout Sections II and III without justification, which is acceptable for an engineering paper but still a background assumption.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Miniature multihole airflow sensor for lightweight aircraft over wide speed and angular range." pith.science (2026). https://pith.science/paper/YG5QZRWB

@misc{pith2026250503331,
  author       = {Pith},
  title        = {Pith review of: Miniature multihole airflow sensor for lightweight aircraft over wide speed and angular range},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/YG5QZRWB}},
  note         = {Machine review of arXiv:2505.03331}
}
read the original abstract

An aircraft's airspeed, angle of attack, and angle of side slip are crucial to its safety, especially when flying close to the stall regime. Various solutions exist, including pitot tubes, angular vanes, and multihole pressure probes. However, current sensors are either too heavy (>30 g) or require large airspeeds (>20 m/s), making them unsuitable for small uncrewed aerial vehicles. We propose a novel multihole pressure probe, integrating sensing electronics in a single-component structure, resulting in a mechanically robust and lightweight sensor (9 g), which we released to the public domain. Since there is no consensus on two critical design parameters, tip shape (conical vs spherical) and hole spacing (distance between holes), we provide a study on measurement accuracy and noise generation using wind tunnel experiments. The sensor is calibrated using a multivariate polynomial regression model over an airspeed range of 3-27 m/s and an angle of attack/sideslip range of +-35{\deg}, achieving a mean absolute error of 0.44 m/s and 0.16{\deg}. Finally, we validated the sensor in outdoor flights near the stall regime. Our probe enabled accurate estimations of airspeed, angle of attack and sideslip during different acrobatic manoeuvres. Due to its size and weight, this sensor will enable safe flight for lightweight, uncrewed aerial vehicles flying at low speeds close to the stall regime.

Figures

Figures reproduced from arXiv: 2505.03331 by the authors.

Figure 1
Figure 1. Miniature integrated 3D airflow sensor: (A) [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Comparison of multihole pressure probes on mass and [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Design of the multihole pressure probe: (A) [PITH_FULL_IMAGE:figures/full_fig_p002_3.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Design parameter optimisation: (A) Four different hole spacings between 0.4 mm and 1.2 mm are explored with cone and sphere tip shapes, resulting in eight different designs. (B) All probes were manufactured through Stereolithography resin printing. (C) A robotic arm wa…
Figure 5
Figure 5. Figure 5: Results of the design parameter optimisation: (A) [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 7
Figure 7. Figure 7: Calibration method: (A) The regression dataset includes 17 measured points (blue) obtained in a closed wind tunnel across nine different speeds ranging from 3 m/s to 27 m/s. To enhance numerical stability, additional linearly interpolated points (red) are added near th…
Figure 8
Figure 8. Figure 8: Validation through outdoor flights: (A) Experimental setup featuring an EasyGlider4 fixed-wing UAV equipped with a Pixhawk autopilot, a pitot tube and our MPP. (B) Three different manoeuvres were performed. (C) Estimations of airspeed, AoA, and AoS from our MPP (red), …

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

27 extracted references · 27 canonical work pages

  1. [1]

    Science, technology and the future of small autonomous drones,

    D. Floreano and R. J. Wood, “Science, technology and the future of small autonomous drones,” nature, vol. 521, no. 7553, pp. 460–466, 2015

    work page 2015

  2. [2]

    Birgenair flight 301,

    P. C. Wiki, “Birgenair flight 301,” https: //planecrash.fandom.com/wiki/Birgenair Flight 301, 2024

    work page 2024

  3. [3]

    Northwest orient airlines flight 6231,

    Wikiwand, “Northwest orient airlines flight 6231,” https: //www.wikiwand.com/en/articles/Northwest Orient Airlines Flight 6231, 2024

    work page 2024

  4. [4]

    Failure analysis for an unmanned aerial vehicle using safe path planning,

    C. E. Lin and P.-C. Shao, “Failure analysis for an unmanned aerial vehicle using safe path planning,” Journal of Aerospace Information Systems, vol. 17, no. 7, pp. 358–369, 2020

    work page 2020

  5. [5]

    Synthetic wind estimation for small fixed-wing drones,

    A. Sharma, G. F. Laupr ´e, P. Longobardi, and J. Skaloud, “Synthetic wind estimation for small fixed-wing drones,” Atmosphere, vol. 15, no. 11, p. 1339, 2024

    work page 2024

  6. [6]

    On estimation of wind velocity, angle-of-attack and sideslip angle of small uavs using standard sensors,

    T. A. Johansen, A. Cristofaro, K. Sørensen, J. M. Hansen, and T. I. Fossen, “On estimation of wind velocity, angle-of-attack and sideslip angle of small uavs using standard sensors,” in 2015 International Conference on Unmanned Aircraft Systems (ICUAS) . IEEE, 2015, pp. 510–519

    work page 2015

  7. [7]

    On estimation of wind velocity, angle-of-attack and sideslip angle of small uavs using standard sensors,

    T. A. Johansen, A. Cristofaro, K. Sørensen, J. M. Hansen, and T. I. Fossen, “On estimation of wind velocity, angle-of-attack and sideslip angle of small uavs using standard sensors,” Journal of Aircraft , 2019

    work page 2019

  8. [8]

    Dis- tributed pressure sensing–based flight control for small fixed-wing unmanned aerial systems,

    K. T. Wood, S. Araujo-Estrada, T. Richardson, and S. Windsor, “Dis- tributed pressure sensing–based flight control for small fixed-wing unmanned aerial systems,” Journal of Aircraft , 2019

    work page 2019

Show all 27 references
  1. [9]

    The compressible cali- bration of miniature multi-hole probes,

    E. S. Johansen, O. K. Rediniotis, and G. Jones, “The compressible cali- bration of miniature multi-hole probes,” Journal of Fluids Engineering , vol. 123, no. 1, pp. 128–138, 09 2000

    2000

  2. [10]

    An approach to minimize aircraft motion bias in multi-hole probe wind measurements made by small unmanned aerial systems,

    L. Al-Ghussain and S. C. C. Bailey, “An approach to minimize aircraft motion bias in multi-hole probe wind measurements made by small unmanned aerial systems,” Atmos. Meas. Tech. , vol. 14, p. 173–184,

  3. [11]

    A miniature seven-hole probe air data system and its application on spin prediction,

    S. Li, Y . Gu, J. Han, and Q. Wang, “A miniature seven-hole probe air data system and its application on spin prediction,” Internationa journal of micro air vehicles , 2016

    2016

  4. [12]

    Development of an unmanned aerial vehicle for atmo- spheric turbulence measurement,

    B. M. Witte, “Development of an unmanned aerial vehicle for atmo- spheric turbulence measurement,” University of Kentucky , 2016

    2016

  5. [13]

    The oxford probe: an open access five- hole probe for aerodynamic measurements,

    B. F. Hall and T. Povey, “The oxford probe: an open access five- hole probe for aerodynamic measurements,” Measurement Science and Technology, 2017

    2017

  6. [14]

    Miniature multihole pressure probes and their neural-network-based calibration,

    O. K. Rediniotis and R. Vijayagopal, “Miniature multihole pressure probes and their neural-network-based calibration,” AIAA Journal, 1999

    1999

  7. [15]

    Influence of tip shape on reynolds number sensitivity for a seven hole pressure probe,

    J. Crawford and A. Michael Birk, “Influence of tip shape on reynolds number sensitivity for a seven hole pressure probe,” Journal of engi- neering for gas turbines and power , vol. 135, no. 9, p. 091602, 2013

    2013

  8. [16]

    The use of pressure fluctuations on the nose of an aircraft for measuring air motion,

    E. N. Brown, C. A. Friehe, and D. H. Lenschow, “The use of pressure fluctuations on the nose of an aircraft for measuring air motion,” National Center for Atmospheric Research , 1982

    1982

  9. [17]

    Angle of attack measurement using low-cost 3d printed five hole probe for uav applications,

    L. Sankaralingam and C. Ramprasadh, “Angle of attack measurement using low-cost 3d printed five hole probe for uav applications,” Elsevier, 2020

    2020

  10. [18]

    An unmanned aerial system for measuring profiles and turbulence in the atmospheric boundary layer,

    S. Alaoui-Sosse, P. Durand, P. Medina, P. Pastor, M. Lothon, and I. Cernov, “An unmanned aerial system for measuring profiles and turbulence in the atmospheric boundary layer,” Senors, 2019

    2019

  11. [19]

    Design of pitot-static tube shapes and their influence on airspeed measurement accuracy,

    S. Kusmirek, V . Socha, D. Urban, L. Hanakova, F. Kripsky, R. Walton, and P. Pecho, “Design of pitot-static tube shapes and their influence on airspeed measurement accuracy,” Journal of Aerospace Engineering , 2024

    2024

  12. [20]

    Miniature multihole airflow sensor for lightweight aircraft over wide speed and angular range,

    L. Stuber, S. Jeger, R. Zufferey, and D. Floreano, “Miniature multihole airflow sensor for lightweight aircraft over wide speed and angular range,” 11 2024. [Online]. Available: https://doi.org/10.5281/zenodo.15254763 8

    2024 doi

  13. [21]

    Multivariate polynomial regression in data mining: method- ology, problems and solutions,

    P. Sinha, “Multivariate polynomial regression in data mining: method- ology, problems and solutions,” Int. J. Sci. Eng. Res , vol. 4, no. 12, pp. 962–965, 2013

    2013

  14. [22]

    The determination of cluster number at k-mean using elbow method and purity evaluation on headline news,

    D. Marutho, S. H. Handaka, E. Wijaya et al. , “The determination of cluster number at k-mean using elbow method and purity evaluation on headline news,” in 2018 international seminar on application for technology of information and communication . IEEE, 2018, pp. 533– 538

    2018

  15. [23]

    Agile perching maneuvers in birds and morphing-wing drones,

    V . W ¨uest, S. Jeger, M. Feroskhan, E. Ajanic, F. Bergonti, and D. Flo- reano, “Agile perching maneuvers in birds and morphing-wing drones,” Nature Communications, vol. 15, no. 1, p. 8330, 2024

    2024

  16. [24]

    Adaptive morphing of wing and tail for stable, resilient, and energy-efficient flight of avian- informed drones,

    S. L. Jeger, V . W¨uest, C. Toumieh, and D. Floreano, “Adaptive morphing of wing and tail for stable, resilient, and energy-efficient flight of avian- informed drones,” arXiv preprint arXiv:2403.08598 , 2024

    2024 arXiv

  17. [25]

    A twist of the tail in turning maneuvers of bird-inspired drones,

    H.-V . Phan and D. Floreano, “A twist of the tail in turning maneuvers of bird-inspired drones,” Science Robotics , vol. 9, no. 96, p. eado3890, 2024. [Online]. Available: https://www.science.org/doi/abs/10.1126/scirobotics.ado3890

    2024 doi

  18. [26]

    Bird-inspired reflexive morphing enables rudderless flight,

    E. Chang, D. D. Chin, and D. Lentink, “Bird-inspired reflexive morphing enables rudderless flight,” Science Robotics , vol. 9, no. 96, p. eado4535, 2024. [Online]. Available: https://www.science.org/doi/abs/10.1126/scirobotics.ado4535

    2024 doi

  19. [2021]

    Available: https: //doi.org/10.5194/amt-14-173-2021

    [Online]. Available: https: //doi.org/10.5194/amt-14-173-2021

    2021 doi

Pith tools

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