Pith. sign in

REVIEW 5 major objections 5 minor 43 references

Critical slowing down in rotor-speed signals forecasts quadrotor loss of control up to 0.9 seconds ahead, using no LOC-event data.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 02:36 UTC pith:SMIK5HDG

load-bearing objection The CSD cascade forecaster is a genuinely useful idea that deserves a careful referee, but the headline '0.9 s' does not match the algorithm's own bound, and the forecasting target is not the labeled LOC event. the 5 major comments →

arxiv 2607.25370 v1 pith:SMIK5HDG submitted 2026-07-28 eess.SY cs.ROcs.SY

Critical slowing down for predicting controller induced loss of control in quadrotors

Jasper J. van Beers , Prashant Solanki , Erik-Jan van Kampen , Coen C. de Visser This is my paper
classification eess.SY cs.ROcs.SY
keywords critical slowing downloss of controlquadrotorearly warning signalslag-1 autocorrelationBayesian detectionflyawaytime-to-LOC forecasting
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

The paper sets out to prove that loss of control in quadrotors—here, the kind caused by an unstable controller loop rather than a rotor failure—can be foreseen using critical slowing down, a generic early-warning phenomenon borrowed from ecology. Its forecaster watches the lag-1 autocorrelation (AC1) of detrended rotor-speed-difference signals; as the closed loop destabilizes, AC1 rises toward 1, and detectors with smaller observation windows register that rise earlier than a detector with a large window calibrated to the moment of loss of control. The sequence of early detections is extrapolated into a time-to-LOC forecast, reaching up to 0.9 seconds of lead time on real flight data from four quadrotors. Because the forecaster can substitute a generic assumed distribution near AC1=1 for actual LOC data, it needs no training set of crashes and no system model. A sympathetic reader would care because this offers a data-light, explainable safety layer for drones that appears to transfer across platforms, controllers, and LOC scenarios without re-parameterization.

Core claim

Controller-induced loss of control in quadrotors is preceded by critical slowing down: the lag-1 autocorrelation of detrended rotor-speed-difference signals climbs toward 1 as the control loop approaches instability. The paper's C-BeFore (Cascaded Bayesian Forecaster) exploits this by running Bayesian LOC detectors on the same early-warning signal at several observation-window lengths. The smallest window detects the AC1 rise first; the largest window is tuned so its detection marks the labeled LOC moment; intermediate detections and the inter-window timing error are combined (with a scaling factor alpha) into a time-to-LOC estimate, Δt_LOC. On 91 real LOC events, the scheme produces forecas

What carries the argument

The C-BeFore cascade: a set of Bayesian detectors B_D = P(L|E_k,D) that all evaluate the same AC1 early-warning signal E_k over different observation-window sizes D (e.g., 150/200, 500, 750 samples), with the largest window aligned to the labeled LOC moment. The smallest window reacts first; the difference in window sizes (Δt = w2 - w1) is the baseline forecast, corrected by α, an inter-window scaling factor computed from the error between the two smallest windows' firing times. Eq. (8) aggregates the individual forecasts to produce Δt_LOC. The key property making it LOC-data-free is the assumed LOC distribution L^A_{E_k}, a constant on [0.9,1.0], justified by CSD theory that AC1 → 1 at tipp

Load-bearing premise

The time-to-LOC forecast assumes the AC1 early-warning signal climbs toward 1 at a roughly constant or mildly accelerating rate, so that a smaller-window detector's firing time can be extrapolated to the largest window using only the difference in window sizes; if AC1 growth stalls, dips, or recovers before the tipping point, the forecast arithmetic produces unreliable or impossible (negative) times.

What would settle it

Assemble a set of near-miss flights—runs where the quadrotor approaches the instability and AC1 rises, then recovers without crossing the ±90° attitude bound—and run the published C-BeFore parameters on them. If the smaller-window detectors fire but the largest-window detector never does, or if AC1 dips after climbing, eq. (8) will issue an imminent-LOC alarm for flights that do not lose control; a systematic near-miss data set would settle whether the monotonic-climb premise holds for real controller instabilities.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • Operators can receive a time-to-LOC estimate of up to 0.9 seconds from rotor-speed telemetry alone, which for a 4 kHz control loop is enough time for a controller to initiate corrective action.
  • The forecaster does not require LOC flight data: using a generic assumed distribution near AC1 = 1 performs nearly as well as using true LOC distributions, removing a major barrier to safety monitoring.
  • A forecaster parameterized on one quadrotor and one LOC scenario detects flyaways on other quadrotors with different flight-control architectures, both indoors and outdoors, with no re-parameterization.
  • False-positive detections are reduced by at least 83% compared to the recurrent-neural-network baselines, and the inference is O(D log D), lighter than an RNN with three or more hidden neurons.
  • Because the approach monitors for the underlying closed-loop instability rather than a specific fault signature, it can be added on top of existing fault-tolerant controllers as a generic safety monitor.

Where Pith is reading between the lines

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

  • If the monotonic-AC1 premise holds generally, the same cascaded-window extrapolation could be applied to other vehicles and machines that lose stability through controller-induced critical transitions—fixed-wing aircraft, ground robots, or robot manipulators with contact instabilities—wherever a scalar CSD indicator can be computed from onboard signals.
  • A concrete extension would be to replace the fixed linear window-difference extrapolation with an adaptive estimator of AC1 growth rate; this could extend lead time beyond 0.9 seconds, but at the cost of the LOC-data-free property that makes the current scheme portable.
  • The paper's own limitation statement notes the monitor does not diagnose the cause of LOC or choose a corrective action; coupling the Δt_LOC output to an online controller that re-tunes gains or engages a safe mode when the forecast shrinks below a threshold is the natural next step, and the forecast horizon here suggests such a loop is feasible.
  • The main usability barrier to adoption is the manual design of the early-warning signal (detrending and AC1 windows) via a parameter sweep; an automated EWS-selection procedure would make the approach a drop-in safety layer on autopilots that log rotor speeds.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit.

Referee Report

5 major / 5 minor

Summary. The paper presents C-BeFore, a two-stage forecasting scheme for controller-induced loss of control (LOC) in quadrotors. In the first stage, lag-1 autocorrelation (AC1) early warning signals are computed from rotor-speed differences and fed into Bayesian detectors with different observation-window sizes. In the second stage, the firing times of the smaller-window detectors are extrapolated to the largest-window detector's firing time, yielding a time-to-LOC estimate Δt_LOC. The method is evaluated on real flight data from four quadrotors: yaw-induced LOC on DataCan75 and CineGo, and flyaway events on SuperKnight5 and DamselFly. The paper claims up to 0.9 s of forewarning, outperformance of recurrent neural network baselines on the DataCan75 dataset, and successful transfer without re-parameterization across platforms, controllers, and LOC scenarios, including a variant that uses an assumed LOC distribution instead of LOC data.

Significance. If the central claims hold, this would be a valuable contribution: a model-free, physically interpretable alternative to black-box LOC predictors, with demonstrated evaluation on a substantial set of real LOC events (91 across four platforms). The manuscript has real strengths: it includes a direct baseline comparison on the same DataCan75 data with explicit numbers (Table 3, Fig. 9), an honest limitations section, and an assumed-distribution variant (Section V.C) that shows little performance loss on CineGo. The empirical material is therefore potentially important for the quadrotor safety community. However, several load-bearing issues need to be resolved before the claims can be accepted as stated.

major comments (5)
  1. [§IV.B.2, §V.A, Table 3] The forecast target is not the LOC event defined by Eq. (1). The text assumes that a detection by the largest-window detector B_Dd 'corresponds to the moment of labeled LOC, t_LOC, and thus represents the ground-truth,' but this is only an assumption. Figure 10 and Table 3 show that detections can occur before the |φ|>90° or |θ|>90° condition is reached (instant forecasts and nonzero detection leeway). Consequently, the reported Δt_LOC values and forecast errors are measured against a detector-alarm time, not against the physical attitude-based LOC label. The RNN baselines were trained to detect Eq. (1); comparing them to a method whose target is a detector firing time is not an equal-footing comparison. Please validate or calibrate B_Dd's firing time against Eq. (1), or explicitly redefine the claim as forecasting the detector alarm rather than LOC.
  2. [§IV.B.2, Eqs. (7)-(8), Algorithm 2] The stated bound 'Δt_LOC∈[0, D_d−D_{d−1}]' is inconsistent with the aggregation in Eq. (8). For the reported window sets [200,500,750] or [150,500,750] at 500 Hz, D_d−D_{d−1}=250 samples = 0.5 s. But if both smaller detectors have active forecasts and α=1, Eq. (8) yields a weighted average that can reach (550+250)/2=400 samples =0.8 s for [200,500,750], or (600+250)/2=425 samples =0.85 s for [150,500,750]. The abstract's 'up to 0.9 seconds' is thus compatible with the algorithm, but the text's claim of a conservative bound of D_d−D_{d−1} is wrong. The bound and the algorithm description need to be reconciled.
  3. [Algorithm 2, line 24] The condition for emitting a forecast, 'if sum(H[i]) > L/2', uses the current-sample detection row H[i], which is reset to zero at every sample in Algorithm 1. In a cascade, the smaller-window detectors fire at different times, so sum(H[i]) will almost never exceed L/2; for L=2 it would require two detectors to fire on the same sample, contradicting the cascade premise. This would prevent forecasts from ever being issued as written. If the implementation used 'sum(A) > L/2' (majority of active forecasts), the pseudocode should be corrected and clearly aligned with the code that produced Table 3.
  4. [§V.D.1 vs. abstract] The abstract and Section V.D claim the forecasters are applied 'without any re-parameterization' to other platforms and LOC scenarios. This is contradicted by the DamselFly description: the EWS computation for the DamselFly uses a moving-average window of 8 samples and an AC1 window of 25 samples, whereas the CineGo parameterization uses 25 and 50 samples respectively. Only the detector windows, thresholds, and assumed LOC distributions are kept fixed. The EWS is part of the forecaster, so the claim of no re-parameterization is overstated. Please either report the DamselFly result as requiring EWS re-tuning, or amend the generality claim.
  5. [Table 3, §V.D.2] For the flyaway generalization cases, the table reports forecast errors of -3.139 s (SuperKnight5) and -0.718 s (DamselFly), but the text only claims successful detection, not forecasting accuracy. The magnitude and sign of these errors are not discussed; a -3.139 s mean forecast error indicates a systematic mismatch between the generated Δt_LOC and the actual event timing, which is a material limitation for any practical use of the forecast in those scenarios. The paper should either report and interpret these forecast errors explicitly, or clearly restrict the flyaway claim to detection rather than time-to-LOC forecasting.
minor comments (5)
  1. [Table 3] The DataCan75 data-driven C-BeFore row lists 48 true positives and 0 false negatives, but the DataCan75 dataset contains 49 LOC flights. One LOC flight is unaccounted for; please correct the count or explain the discrepancy.
  2. [Various] Typos and wording: 'we show that the our approach' (Section I); 'it’s position' (Section I); 'the the approach' (Section VI); 'flyway' vs 'flyaway' used inconsistently throughout; 'shown inaof fig. 10' (Section V.B); 'Indiflight' capitalization.
  3. [Eq. (1)] The variable LM-ATT is defined to be 0 when |φ|>90° or |θ|>90° and 1 otherwise; the text says LOC is 'defined as the moment' the attitude threshold is exceeded. The naming and polarity are confusing; please clarify that the label is 0 during LOC, or rename the variable.
  4. [§IV.B.1, Eqs. (4)-(6)] The relative similarity scores Φ_L and Φ_N are described as 'suitable probability functions' for binary classification. They are normalized inverse Wasserstein distances, not probabilities in a statistical sense. The Bayesian posterior in Eq. (3) should be described as a heuristic scoring rule rather than a formal probabilistic inference, unless a proper generative model is provided.
  5. [§V.A] The RNN comparison selects the best run per architecture ('Run 14', 'Run 31', etc.). This is acceptable as a baseline, but it should be stated explicitly that the comparison uses the best initialization for each RNN, not the average or worst, to avoid appearing to cherry-pick.

Circularity Check

1 steps flagged

Time-to-LOC forecast reduces to predicting the largest-window detector's firing time, which is also the ground truth used for validation.

specific steps
  1. self definitional [Section IV.B.2 (From detections to forecasts), eqs. (7)-(8), Algorithm 2]
    "It is assumed that a detection made by B750 corresponds to the moment of labeled LOC, t_LOC, and thus represents the 'ground-truth' (highlighted in blue in A of fig. 6). The goal of the smaller window detectors is to anticipate when B750 should make a detection."

    The advertised Δt_LOC is computed in Algorithm 2 as i + (D_d − D[w]) and aggregated in eq. (8); i.e., it is a forecast of when B_Dd will fire. The same B_Dd firing time is used as the 'true' value: Algorithm 2 lines 19–20 compute α from the actual B_w2 detection index, and Table 3's forecast errors are measured relative to B_Dd detection. No independent check links B_Dd firing to the eq. (1) |φ|>90/|θ|>90 label. Hence the forecast error measures self-consistency of the detector cascade, not prediction of the LOC event; the derivation is circular by construction.

full rationale

The central forecasting claim is not independent of the detector that defines the forecast target. The paper explicitly assumes that the largest-window Bayesian detector B_Dd fires at the labeled LOC time, then Algorithm 2 and eq. (8) produce a time-to-LOC that is literally the predicted firing time of that same detector. Validation uses B_Dd's actual firing time as ground truth, so the reported forecast errors and 'up to 0.9 s' lead time are not validated against the independent attitude-based LOC definition of eq. (1). This is a genuine, though partial, circularity: the cascade does contain real data-driven signal content and the assumed LOC distribution [0.9, 1.0] is not fitted to the data, so the paper is not wholly tautological. However, because the main headline quantity reduces to an internal detector-alarm forecast, a score of 6 is appropriate. The 0.5 s bound stated in the text (D_d − D_{d−1} = 250 samples) versus the abstract's 0.9 s is an additional inconsistency that supports the concern but is not itself a circular step. Self-citations to [28], [37], and [41] are used for background and baseline data but are not load-bearing for the circularity identified here.

Axiom & Free-Parameter Ledger

6 free parameters · 6 axioms · 1 invented entities

The ledger shows the paper's central claim rests on roughly six free parameters (windows, thresholds, assumed distribution), several benchmark-specific settings, and three ad hoc modeling choices (Φ probabilities, B_Dd as pseudo-ground-truth, linear extrapolation). The strongest non-circular component is the CSD theory and the independence of the assumed L_A from the LOC data. The weakest is the dependence of the 0.9 s forecast on chosen window pairs and on the B_Dd-as-ground-truth assumption.

free parameters (6)
  • Detrending window W (samples) = DataCan75: 10; CineGo: 25; DamselFly: 8
    Chosen via parameter sweep (fig. 8: window set {2,3,5,10,25,50}); a sweep can be done on the very data the forecaster is later evaluated on; DamselFly windows changed without explicit sweep. They control the detrended signal, thus the AC1 levels and all forecasts.
  • AC1 window (samples) = DataCan75: 50; CineGo: 50; DamselFly: 25
    Chosen as part of EWS parameter sweep; governs variance and smoothness of the AC1 series, and hence when it crosses the detector threshold.
  • Detector window sizes D (samples) = DataCan75: [200,500,750]; CineGo: [150,500,750]
    Selected by the authors (via experience and sweep, footnote 16); these three numbers determine the forecast lead time by construction (Δt_hat = w2 - w1, eq. 7), so the 0.9 s claim is sensitive to the largest window pair.
  • Detection threshold λ = DataCan75: 0.89; CineGo: 0.88
    Hand-set probability threshold over [0.65, 1.0); changes true/false positive tradeoff; used by Algorithm 1 to decide detections.
  • Assumed LOC distribution L_A interval = Uniform on [0.9, 1.0] in E_k
    This is a free modeling assumption bounding AC1 near 1 for LOC behavior; it is motivated by CSD theory but the precise interval boundary (0.9) is chosen, not derived.
  • Nominal distribution N_Ek = From nominal flights of each platform (CineGo, SuperKnight5, DamselFly) separately
    Nominal data is treated as available; the reference distribution changes per platform, reducing the claim that no data is needed at all.
axioms (6)
  • domain assumption CSD phenomenon applies to controlled quadrotor closed-loop systems approaching instability (AC1→1 as LOC approaches)
    Core premise of the paper (§II.A, using refs [29,37,38]) imported from complex-systems literature; the authors' own companion work [37] provides the key support, but for this paper it is a background assumption. Not all unstable controller-induced events need exhibit AC1 growth.
  • domain assumption Attitude-based LOC definition (|roll| or |pitch| > 90°) matches the moment of true loss of control
    Eq. (1), inherited from [28]; if the drone remains recoverable past 90° or loses control earlier, the label times shift and all forecast-error statistics change.
  • standard math Pearson lag-1 autocorrelation with moving-average detrending is a valid EWS statistic for these short and non-stationary windows
    Standard statistics, but the detrender is a high-pass filter removing low-frequency drift; its interaction with oscillatory unstable modes is not derived. The paper acknowledges detrending as a major challenge (§V.E).
  • ad hoc to paper Wasserstein-distance relative similarity scores are probabilities for binary LOC classification
    Equations (4)-(5) are introduced ad hoc; the claim that Φ values are suitable probabilities for a binary decision problem is argued rather than proven.
  • ad hoc to paper The largest detector window B_Dd detects at the labeled LOC moment and serves as pseudo-ground-truth
    Section IV.B.2 and fig. 6: 'It is assumed that a detection made by B_750 corresponds to the moment of labeled LOC'. Directly sets the reference times against which forecast errors are measured.
  • ad hoc to paper Linear extrapolation plus a single inter-window correction α captures the approach to LOC
    Eqs. (7)-(8): Δt_hat = w2 - w1, with α correcting the trend between only two window firings; stalling or non-monotonic AC1 trajectories violate it.
invented entities (1)
  • Assumed LOC distribution L_A_Ek independent evidence
    purpose: A synthetic probability distribution over AC1 values assumed to represent near-LOC behavior, used to avoid needing LOC data.
    Uniform on [0.9, 1.0] is testable as a modeling assumption and the paper reports performance when using it; falsifiable in the sense that different intervals give different detection times.

pith-pipeline@v1.3.0-alltime-deepseek · 18329 in / 12079 out tokens · 99779 ms · 2026-08-01T02:36:47.943053+00:00 · methodology

0 comments
read the original abstract

We develop a novel forecasting scheme to anticipate controller induced loss of control (LOC) in quadrotors and evaluate it on real LOC flight data from four different quadrotors. For this, early warning signals of LOC are derived using critical slowing down (CSD), a generic phenomenon shown to precede critical transitions across various complex ecological and biological systems. As such, our early warning indicators are generic in the sense that no system models are needed to facilitate forecasts of LOC. The approach is evaluated on real quadrotor flight data wherein LOC occurs due to unstable controller behavior arising from input-output delays. Our approach achieves a time-to-LOC forecast of up to 0.9 seconds before LOC occurs, outperforming state-of-the-art recurrent neural network quadrotor LOC forecasters in terms of detection accuracy and LOC data reliance. In particular, we leverage insights from CSD to accurately predict LOC without using data of the LOC event itself. Going further, we apply our forecasters without any re-parameterization to anticipate a different LOC scenario, quadrotor flyways, that occur on other quadrotors flying both indoors and outdoors. Despite these differences, our approach successfully detects LOC, demonstrating that it can generalize across controller architectures, quadrotors, and LOC scenarios.

discussion (0)

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

Reference graph

Works this paper leans on

43 extracted references · 14 canonical work pages

  1. [1]

    Safe limits, mindful organizing and loss of control in commercial aviation,

    Oliver, N., Calvard, T., and Potočnik, K., “Safe limits, mindful organizing and loss of control in commercial aviation,”Safety Science,Vol.120,2019,pp.772–780. https://doi.org/https://doi.org/10.1016/j.ssci.2019.08.018,URLhttps://www.sciencedirect. com/science/article/pii/S0925753519301055,

    work page doi:10.1016/j.ssci.2019.08.018 2019

  2. [2]

    Robot application and occupational injuries: Are robots necessarily safer?

    Yang, S., Zhong, Y., Feng, D., Li, R. Y. M., Shao, X.-F., and Liu, W., “Robot application and occupational injuries: Are robots necessarily safer?”Safety Science, Vol. 147, 2022, p. 105623. https://doi.org/https://doi.org/10.1016/j.ssci.2021.105623, URL https://www.sciencedirect.com/science/article/pii/S092575352100463X,

    arXiv 2022

  3. [3]

    Follow-up analysis of mobile robot failures,

    Carlson, J., Murphy, R., and Nelson, A., “Follow-up analysis of mobile robot failures,”IEEE International Conference on Robotics and Automation, 2004. Proceedings. ICRA ’04. 2004, Vol. 5, 2004, pp. 4987–4994 Vol.5. https://doi.org/10.1109/ ROBOT.2004.1302508,

    Pith/arXiv arXiv 2004

  4. [4]

    A Comprehensive Survey of the Recent Studies with UAV for Precision Agriculture in Open Fields and Greenhouses,

    Aslan, M. F., Durdu, A., Sabanci, K., Ropelewska, E., and Gültekin, S. S., “A Comprehensive Survey of the Recent Studies with UAV for Precision Agriculture in Open Fields and Greenhouses,”Applied Sciences, Vol. 12, No. 3, 2022. https://doi.org/10.3390/app12031047, URL https://www.mdpi.com/2076-3417/12/3/1047,

    work page doi:10.3390/app12031047 2022

  5. [5]

    CollaborativeUnmannedVehiclesforInspection, Maintenance, and Repairs of Offshore Wind Turbines,

    Nordin,M.H.,Sharma,S.,Khan,A.,Gianni,M.,Rajendran,S.,andSutton,R.,“CollaborativeUnmannedVehiclesforInspection, Maintenance, and Repairs of Offshore Wind Turbines,”Drones, Vol. 6, No. 6, 2022. https://doi.org/10.3390/drones6060137, URL https://www.mdpi.com/2504-446X/6/6/137,

    work page doi:10.3390/drones6060137 2022

  6. [6]

    Effect of unmanned aerial vehicle (drone) delivery on blood product delivery time and wastage in Rwanda: A retrospective, cross-sectional study and time series analysis,

    Nisingizwe, M. P., Ndishimye, P., Swaibu, K., Nshimiyimana, L., Karame, P., Dushimiyimana, V., Musabyimana, J. P., Musanabaganwa, C., Nsanzimana, S., and Law, M. R., “Effect of unmanned aerial vehicle (drone) delivery on blood product delivery time and wastage in Rwanda: A retrospective, cross-sectional study and time series analysis,”The Lancet Global He...

    2022

  7. [7]

    Safety Control for Quadrotor UAV Against Ground Effect and Blade Damage,

    Guo, K., Zhang, W., Zhu, Y., Jia, J., Yu, X., and Zhang, Y., “Safety Control for Quadrotor UAV Against Ground Effect and Blade Damage,”IEEE Transactions on Industrial Electronics, Vol. 69, No. 12, 2022, pp. 13373–13383. https: //doi.org/10.1109/TIE.2022.3140494

    arXiv 2022

  8. [8]

    QuadcopterTrajectory Tracking inthe Presenceof 4Faulty Actuators: ANonlinear MHE and MPC Approach,

    Eltrabyly, A., Ichalal, D., andMammar, S., “QuadcopterTrajectory Tracking inthe Presenceof 4Faulty Actuators: ANonlinear MHE and MPC Approach,”IEEE Control Systems Letters, Vol. 6, 2022, pp. 2024–2029. https://doi.org/10.1109/LCSYS.2021. 3137099

    work page doi:10.1109/lcsys.2021 2022

  9. [9]

    Estimating the Loss of Effectiveness of UAV Actuators in the Presence of Aerodynamic Effects,

    Madruga, S. P., Nascimento, T. P., Holzapfel, F., and Lima, A. M. N., “Estimating the Loss of Effectiveness of UAV Actuators in the Presence of Aerodynamic Effects,”IEEE Robotics and Automation Letters, Vol. 8, No. 3, 2023, pp. 1335–1342. https://doi.org/10.1109/LRA.2023.3238184. 24

    arXiv 2023

  10. [10]

    Uniform Passive Fault-Tolerant Control of a Quadcopter With One, Two, or Three Rotor Failure,

    Ke, C., Cai, K.-Y., and Quan, Q., “Uniform Passive Fault-Tolerant Control of a Quadcopter With One, Two, or Three Rotor Failure,”IEEE Transactions on Robotics, Vol. 39, No. 6, 2023, pp. 4297–4311. https://doi.org/10.1109/TRO.2023.3297048

    arXiv 2023

  11. [11]

    Upset Recovery Control for Quadrotors Subjected to a Complete Rotor Failure from Large Initial Disturbances,

    Sun, S., Baert, M., van Schijndel, B. S., and de Visser, C. C., “Upset Recovery Control for Quadrotors Subjected to a Complete Rotor Failure from Large Initial Disturbances,”2020 IEEE International Conference on Robotics and Automation (ICRA), 2020, pp. 4273–4279. https://doi.org/10.1109/ICRA40945.2020.9197239

    arXiv 2020

  12. [12]

    Nonlinear MPC for Quadrotor Fault-Tolerant Control,

    Nan, F., Sun, S., Foehn, P., and Scaramuzza, D., “Nonlinear MPC for Quadrotor Fault-Tolerant Control,”IEEE Robotics and Automation Letters, Vol. 7, No. 2, 2022, pp. 5047–5054. https://doi.org/10.1109/LRA.2022.3154033

    arXiv 2022

  13. [13]

    Incremental Nonlinear Fault-Tolerant Control of a Quadrotor With Complete Loss of Two Opposing Rotors,

    Sun, S., Wang, X., Chu, Q., and d. Visser, C., “Incremental Nonlinear Fault-Tolerant Control of a Quadrotor With Complete Loss of Two Opposing Rotors,”IEEE Transactions on Robotics, 2020, pp. 1–15. https://doi.org/10.1109/TRO.2020.3010626,

    arXiv 2020

  14. [14]

    Safety Systems for Emergency Landing of Civilian Unmanned Aerial Vehicles (UAVs)—A Comprehensive Review,

    Farajijalal, M., Eslamiat, H., Avineni, V., Hettel, E., and Lindsay, C., “Safety Systems for Emergency Landing of Civilian Unmanned Aerial Vehicles (UAVs)—A Comprehensive Review,”Drones, Vol. 9, No. 2, 2025. https://doi.org/10.3390/ drones9020141, URL https://www.mdpi.com/2504-446X/9/2/141

    2025

  15. [15]

    A Scalable Safety Critical Control Framework for Nonlinear Systems,

    Gurriet, T., Mote, M., Singletary, A., Nilsson, P., Feron, E., and Ames, A. D., “A Scalable Safety Critical Control Framework for Nonlinear Systems,”IEEE Access, Vol. 8, 2020, pp. 187249–187275. https://doi.org/10.1109/ACCESS.2020.3025248

    arXiv 2020

  16. [16]

    Robust Dynamic Geofencing Attitude Control for Quadrotor Systems,

    Zheng, Z., Su, X., Jiang, T., and Huang, J., “Robust Dynamic Geofencing Attitude Control for Quadrotor Systems,”IEEE Transactions on Industrial Electronics, Vol. 70, No. 2, 2023, pp. 1861–1869. https://doi.org/10.1109/TIE.2022.3159919

    arXiv 2023

  17. [17]

    Onboard Safety Guarantees for Racing Drones: High-Speed Geofencing With Control Barrier Functions,

    Singletary, A., Swann, A., Chen, Y., and Ames, A. D., “Onboard Safety Guarantees for Racing Drones: High-Speed Geofencing With Control Barrier Functions,”IEEE Robotics and Automation Letters, Vol. 7, No. 2, 2022, pp. 2897–2904. https://doi.org/10.1109/LRA.2022.3144777

    arXiv 2022

  18. [18]

    Confidence-aware motion prediction for real-time collision avoidance,

    Fridovich-Keil, D., Bajcsy, A., Fisac, J. F., Herbert, S. L., Wang, S., Dragan, A. D., and Tomlin, C. J., “Confidence-aware motion prediction for real-time collision avoidance,”The International Journal of Robotics Research, Vol. 39, No. 2-3, 2020, pp. 250–265. https://doi.org/10.1177/0278364919859436, URL https://doi.org/10.1177/0278364919859436

    work page doi:10.1177/0278364919859436 2020

  19. [19]

    Safe Drone Flight with Time-Varying Backup Controllers,

    Singletary, A., Swann, A., Rodriguez, I. D. J., and Ames, A. D., “Safe Drone Flight with Time-Varying Backup Controllers,” 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2022, pp. 4577–4584. https://doi.org/10. 1109/IROS47612.2022.9981741

    arXiv 2022

  20. [20]

    Collision-free source seeking control methods for unicycle robots,

    Li, T., and Jayawardhana, B., “Collision-free source seeking control methods for unicycle robots,”IEEE Transactions on Automatic Control, 2024

    2024

  21. [21]

    Quadrotor Safe Flight Envelope Prediction in the High-Speed Regime: A Monte-Carlo Approach,

    Sun, S., and de Visser, C., “Quadrotor Safe Flight Envelope Prediction in the High-Speed Regime: A Monte-Carlo Approach,” AIAA Scitech 2019 Forum, American Institute of Aeronautics and Astronautics, 2019. https://doi.org/10.2514/6.2019-0948, URL https://arc.aiaa.org/doi/abs/10.2514/6.2019-0948, 25

    work page doi:10.2514/6.2019-0948 2019

  22. [22]

    Lossofcontrolpreventionandrecovery: Onboardguidance,control, andsystemstechnologies,

    Belcastro,C., “Lossofcontrolpreventionandrecovery: Onboardguidance,control, andsystemstechnologies,”AIAAGuidance, Navigation, and Control Conference, 2012, p. 4762

    2012

  23. [23]

    Estimating Loss-of-Control: A Data-Based Predictive Control Approach,

    Barlow, J., Stepanyan, V., and Krishnakumar, K., “Estimating Loss-of-Control: A Data-Based Predictive Control Approach,” AIAA Guidance, Navigation, and Control Conference, 2011. https://doi.org/10.2514/6.2011-6408, URL https://arc.aiaa.org/doi/ abs/10.2514/6.2011-6408

    work page doi:10.2514/6.2011-6408 2011

  24. [24]

    Real-time Adaptive Optimal Prediction of Safe Control Spaces and Augmented- Reality Head-Up Displays Towards Aircraft Loss-of-Control Mitigation,

    Rafi, M., Steck, J. E., and Chakravarthy, A., “Real-time Adaptive Optimal Prediction of Safe Control Spaces and Augmented- Reality Head-Up Displays Towards Aircraft Loss-of-Control Mitigation,”AIAA Scitech 2021 Forum, 2021. https://doi.org/10. 2514/6.2021-0757, URL https://arc.aiaa.org/doi/abs/10.2514/6.2021-0757

    work page doi:10.2514/6.2021-0757 2021

  25. [25]

    Flight Envelope Prediction via Optimal Control-Based Reachability Analysis,

    Lu, Z., Hong, H., Gerdts, M., and Holzapfel, F., “Flight Envelope Prediction via Optimal Control-Based Reachability Analysis,” Journal of Guidance, Control, and Dynamics, Vol. 45, No. 1, 2022, pp. 185–195. https://doi.org/10.2514/1.G006219, URL https://doi.org/10.2514/1.G006219

    work page doi:10.2514/1.g006219 2022

  26. [26]

    Hamilton-jacobi reachability: A brief overview and recent advances,

    Bansal, S., Chen, M., Herbert, S., and Tomlin, C. J., “Hamilton-jacobi reachability: A brief overview and recent advances,” 2017 IEEE 56th Annual Conference on Decision and Control (CDC), IEEE, 2017, pp. 2242–2253

    2017

  27. [27]

    A Fault-Tolerant Control Scheme for Fixed-Wing UAVs with Flight Envelope Awareness,

    Zogopoulos-Papaliakos, G., Karras, G. C., and Kyriakopoulos, K. J., “A Fault-Tolerant Control Scheme for Fixed-Wing UAVs with Flight Envelope Awareness,”J. Intell. Robotics Syst., Vol. 102, No. 2, 2021. https://doi.org/10.1007/s10846-021-01393-3, URL https://doi.org/10.1007/s10846-021-01393-3

    work page doi:10.1007/s10846-021-01393-3 2021

  28. [28]

    Loss-of-Control Prediction of a Quadcopter Using Recurrent Neural Networks,

    Altena, A. V., van Beers, J. J., and de Visser, C. C., “Loss-of-Control Prediction of a Quadcopter Using Recurrent Neural Networks,”Journal of Aerospace Information Systems, Vol. 20, No. 10, 2023, pp. 648–659. https://doi.org/10.2514/1.I011231, URL https://doi.org/10.2514/1.I011231

    work page doi:10.2514/1.i011231 2023

  29. [29]

    Early-warning signals for critical transitions,

    Scheffer, M., Bascompte, J., Brock, W. A., Brovkin, V., Carpenter, S. R., Dakos, V., Held, H., Nes, E. H. V., Rietkerk, M., and Sugihara, G., “Early-warning signals for critical transitions,”Nature 2009 461:7260, Vol. 461, 2009, pp. 53–59. https://doi.org/10.1038/nature08227, URL https://www.nature.com/articles/nature08227

    doi:10.1038/nature08227 2009

  30. [30]

    Critical slowing down as a biomarker for seizure susceptibility,

    Maturana, M. I., Meisel, C., Dell, K., Karoly, P. J., D’Souza, W., Grayden, D. B., Burkitt, A. N., Jiruska, P., Kudlacek, J., Hlinka, J., Cook, M. J., Kuhlmann, L., and Freestone, D. R., “Critical slowing down as a biomarker for seizure susceptibility,”Nature Communications 2020 11:1, Vol. 11, 2020, pp. 1–12. https://doi.org/10.1038/s41467-020-15908-3, UR...

    doi:10.1038/s41467-020-15908-3 2020

  31. [31]

    Anticipating infectious disease re-emergence and elimination: A test of early warning signals using empirically based models,

    Tredennick, A. T., O’Dea, E. B., Ferrari, M. J., Park, A. W., Rohani, P., and Drake, J. M., “Anticipating infectious disease re-emergence and elimination: A test of early warning signals using empirically based models,”Journal of the Royal Society Interface, Vol. 19, No. 193, 2022, p. 20220123

    2022

  32. [32]

    Detecting critical slowing down in high-dimensional epidemiological systems,

    Brett,T.,Ajelli,M.,Liu,Q.H.,Krauland,M.G.,Grefenstette,J.J.,Panhuis,W.G.V.,Vespignani,A.,Drake,J.M.,andRohani, P., “Detecting critical slowing down in high-dimensional epidemiological systems,”PLOS Computational Biology, Vol. 16, 26 2020, p. e1007679. https://doi.org/10.1371/JOURNAL.PCBI.1007679, URL https://journals.plos.org/ploscompbiol/article?id= 10.1...

    work page doi:10.1371/journal.pcbi.1007679 2020

  33. [33]

    Forecasting the Onset of Traffic Congestions on Circular Roads,

    Ghadami, A., and Epureanu, B. I., “Forecasting the Onset of Traffic Congestions on Circular Roads,”IEEE Transactions on Intelligent Transportation Systems, Vol. 22, 2021, pp. 1196–1205. https://doi.org/10.1109/TITS.2020.2964021

    arXiv 2021

  34. [34]

    Stability and Resilience of Transportation Systems: Is a Traffic Jam About to Occur?

    Ghadami, A., Doering, C. R., Drake, J. M., Rohani, P., and Epureanu, B. I., “Stability and Resilience of Transportation Systems: Is a Traffic Jam About to Occur?”IEEE Transactions on Intelligent Transportation Systems, Vol. 23, No. 8, 2022, pp. 10803–10814. https://doi.org/10.1109/TITS.2021.3095897

    arXiv 2022

  35. [35]

    Identifyingcriticaltransitionsandinstabilityinhapticsystems,

    Kerr,L.,Hutchison,C.,andKövecses,J.,“Identifyingcriticaltransitionsandinstabilityinhapticsystems,”NonlinearDynamics, Vol. 111, No. 13, 2023, pp. 12607–12623

    2023

  36. [36]

    Warning of a forthcoming collapse of the Atlantic meridional overturning circulation,

    Ditlevsen, P., and Ditlevsen, S., “Warning of a forthcoming collapse of the Atlantic meridional overturning circulation,”Nature Communications, Vol. 14, No. 1, 2023, pp. 1–12

    2023

  37. [37]

    Early warning signals for loss of control in complex systems,

    van Beers, J. J., Scheffer, M., Solanki, P., van de Leemput, I. A., van Nes, E. H., and de Visser, C. C., “Early warning signals for loss of control in complex systems,”Proceedings of the National Academy of Sciences, Vol. 123, No. 27, 2026, p. e2608847123. https://doi.org/10.1073/pnas.2608847123, URL https://www.pnas.org/doi/abs/10.1073/pnas.2608847123

    work page doi:10.1073/pnas.2608847123 2026

  38. [38]

    Emerging signals of declining forest resilience under climate change,

    Forzieri, G., Dakos, V., McDowell, N. G., Ramdane, A., and Cescatti, A., “Emerging signals of declining forest resilience under climate change,”Nature, Vol. 608, No. 7923, 2022, pp. 534–539

    2022

  39. [39]

    Anticipating Critical Transitions,

    Scheffer, M., Carpenter, S. R., Lenton, T. M., Bascompte, J., Brock, W., Dakos, V., van de Koppel, J., van de Leemput, I. A., Levin, S. A., van Nes, E. H., Pascual, M., and Vandermeer, J., “Anticipating Critical Transitions,”Science, Vol. 338, No. 6105, 2012, pp. 344–348. https://doi.org/10.1126/science.1225244, URL https://www.science.org/doi/abs/10.1126...

    doi:10.1126/science.1225244 2012

  40. [40]

    Robustness of variance and autocorrelation as indicators of critical slowing down,

    Dakos, V., Nes, E. H. V., D’Odorico, P., and Scheffer, M., “Robustness of variance and autocorrelation as indicators of critical slowing down,”Ecology, Vol. 93, 2012, pp. 264–271. https://doi.org/10.1890/11-0889. 1, URL https://onlinelibrary.wiley.com/doi/full/10.1890/11-0889.1https://onlinelibrary.wiley.com/doi/abs/10.1890/11-0889. 1https://esajournals.o...

    work page doi:10.1890/11-0889 2012

  41. [41]

    A novel metric for detecting quadrotor loss-of-control,

    Van Beers, J. J., Solanki, P., and De Visser, C. C., “A novel metric for detecting quadrotor loss-of-control,”2024 IEEE International Conference on Robotics and Automation (ICRA), 2024, pp. 15570–15576. https://doi.org/10.1109/ICRA57147. 2024.10610662

    arXiv 2024

  42. [42]

    J., and Davis, R

    Brockwell, P. J., and Davis, R. A.,Introduction to Time Series and Forecasting, Springer International Publishing, 2016. https://doi.org/10.1007/978-3-319-29854-2, URL http://link.springer.com/10.1007/978-3-319-29854-2

    doi:10.1007/978-3-319-29854-2 2016

  43. [43]

    AFaultDetectionandDiagnosisSystemforAutonomousVehiclesBasedon HybridApproaches,

    Fang,Y.,Min,H.,Wang,W.,Xu,Z.,andZhao,X.,“AFaultDetectionandDiagnosisSystemforAutonomousVehiclesBasedon HybridApproaches,”IEEESensorsJournal,Vol.20,No.16,2020,pp.9359–9371. https://doi.org/10.1109/JSEN.2020.2987841. 27

    arXiv 2020