{"id":"2e6e6a2e-e47d-4601-b30c-d946f72b0a07","arxiv_id":"2412.11436","paper_version":2,"verdict":"CONDITIONAL","confidence":"HIGH","novelty_score":5.0,"correctness_risk":"medium","formal_verification":"none","parameter_count":2,"one_line_summary":"Event-based cameras can track a laser spot's tilt to about 1 pixel (15 arcsec) error in the lab, but the milliarcsecond accuracy required for laser guide star tip-tilt remains untested.","lead":"This paper tests an event-based camera, which records only brightness changes, for measuring tip-tilt of a laser guide star spot in adaptive optics. In lab tests it tracked a moving laser spot down to about one pixel of error, but that is far from the milliarcsecond accuracy the technique really needs.","discovery_kind":"new_application","skeptic_critique":{"model":"deepseek-v4-flash","headline":"The 15-arcsec lab accuracy is measured against the DM command signal, which Sec. 3.2.3 admits the DM did not actually reproduce, so the headline error may be an artifact of an unreliable reference.","rationale":"The most load-bearing assumption is not simply the future step from 15 arcsec to milliarcseconds; it is that the 15-arcsec number itself is a valid measurement. The paper's own Fig. 10 shows phase mismatch and the text blames imperfect DM motion, yet Fig. 11 labels the commanded set tilt as the true value. Since no independent measurement of the actual spot motion is provided, the central quantitative claim is not established. The qualitative demonstration that background level and threshold control the event count is plausible and supported by the event-rate figures, so this is not grounds for rejection. Rather, the paper should be accepted only on the condition that the accuracy result is re-derived using an independent reference, such as a frame-based centroid on the same spot. The reader's conditional verdict is therefore unchanged, but for a sharper reason than the arcsecond-to-milliarcsecond extrapolation alone. The absence of public data and code makes this impossible to check post hoc.","tokens_in":13487,"tokens_out":7035,"duration_ms":64726,"concrete_test":"Modify the test bench by inserting a 50/50 beam splitter after the deformable mirror so the same laser spot is imaged simultaneously on the event camera and on a calibrated frame-based camera with matched plate scale; record centroids from the frame camera at 1 kHz as an independent ground truth, then compare the event-tracker output over the same 200-Hz, 200-arcsec, high-background runs used in Fig. 11. If the RMS difference between the event tracker and the frame-reference centroid is ≤1 pixel (15 arcsec), the claimed accuracy is confirmed; if not, the Fig. 11 error must be recomputed against the frame reference and the accuracy claim revised.","verdict_should_be":"UNCHANGED","load_bearing_attack":"Section 3.2.3 states that the DM-induced tilt had 'observed inconsistencies... due to signal processing issues' and 'an imperfect sinusoidal signal reaching the actuators,' and Fig. 10's caption explicitly says the blue line is 'the signal sent to the DM rather than the true tilt induced by the DM.' Despite this, the error analysis in Fig. 11 and the derived n=0.339 in Eq. (9) treat the commanded DM angle as the 'true value.' If the DM's actual angular amplitude or phase differed from the command at the tested frequencies, the reported 'error of 1 pixel or 15 arcsec' is not the detector's error with respect to true tilt; it is the difference between the tracker and a known-imperfect reference. The claim could be worse or better than stated; the data as presented cannot tell. This is the load-bearing premise of the central lab claim, and it also undermines the extrapolation to milliarcsecond LGS time-delay sensing. In addition, Eq. (9) requires the aperture radius a, which is never specified, so the n=0.339 wavefront-error value cannot be independently checked.","agreement_with_reader":"partial"},"referee_report":{"model":"deepseek-v4-flash","summary":"The paper reports a laboratory characterization of an event-based camera as a tip-tilt sensor for laser guide star adaptive optics. The authors describe the logarithmic response of the sensor, present a theoretical model for event triggering under background illumination, and validate qualitative predictions with experiments using a deformable mirror to induce known tilts on a 589-nm laser spot. They measure event rates versus laser power and background level, characterize noise sources (shot noise and leak events), and apply an exponential moving-average tracker to estimate spot position. The headline result is that, under high background illumination, the tilt measurement error can be reduced to about 1 pixel (15 arcsec), and that by tuning the event threshold, comparable accuracy can be achieved across different background levels. The paper concludes that event-based detectors are a strong contender for LGS tip-tilt sensing via the time-delay method, with future work planned to reduce the plate scale toward milliarcsecond accuracy.","tokens_in":13653,"tokens_out":2924,"duration_ms":26717,"significance":"If the experimental claims hold, this is a useful proof-of-concept: it demonstrates, with real hardware, that an event-based sensor can track a moving spot at high speed with an error of about one pixel and that its response can be matched across background levels by adjusting the threshold. The noise characterization (Sec. 3.2.2) provides quantitative data that will be of value to others working on event-based wavefront sensing, and the paper is careful to cite prior event-based vision literature. However, the significance is tempered by two gaps. First, the accuracy claim relies on a reference signal that the paper itself describes as imperfect (Sec. 3.2.3, Fig. 10), so the numerical error values are not established to the claimed precision. Second, the gap between 15 arcsec and the milliarcsecond level required for the LGS time-delay method is explicitly acknowledged as untested future work (Sec. 4), meaning the paper's central conclusion that the sensor is 'a strong contender' for LGS tip-tilt sensing is not yet directly supported by the data. These are fixable in a revision, but they are load-bearing for the paper's main claim.","major_comments":[{"comment":"The error analysis treats the DM command signal as the 'true value' of the tilt, yet the paper states that the DM had 'observed inconsistencies... due to signal processing issues' and that 'an imperfect sinusoidal signal' reached the actuators, with Fig. 10's caption explicitly noting that the blue line is 'the signal sent to the DM rather than the true tilt induced by the DM.' If the DM's actual angular amplitude or phase differed from the command, the reported error of 1 pixel or 15 arcsec is not the detector's error relative to true tilt but the difference between the tracker and an imperfect reference. This undermines the central quantitative claim. The authors should either re-analyze the data against a calibrated reference (e.g., a simultaneous frame-based measurement or an interferometric verification of DM motion) or, at minimum, quantify the uncertainty in the DM's actual tilt and propagate it into the reported error bars.","section":"Sec. 3.2.3, Fig. 10 and Fig. 11"},{"comment":"The wavefront error calculation yielding n = 0.339 cannot be independently checked because the aperture radius a is never specified in the experimental setup (Sec. 3.1). The equation Δω_tilt = 2 a y / f = n λ requires a, f, and y; while f = 200 mm and y is presumably the measured position error, a is absent from the setup description. Please provide the value of a used and show the intermediate steps leading to n = 0.339.","section":"Sec. 3.2.3, Eq. (9)"},{"comment":"The paper's conclusion that event-based detectors are 'a strong contender' for LGS tip-tilt sensing goes beyond what the experiments demonstrate. The measured accuracy is 15 arcsec at a plate scale of 15 arcsec/pixel, while the time-delay method requires milliarcsecond-level accuracy, as stated in Sec. 1.1. The authors acknowledge in Sec. 4 that reducing the plate scale to reach milliarcsecond accuracy is future work. As written, the conclusion should be tempered to state that the sensor has demonstrated the required qualitative behavior at arcsecond scales and that the milliarcsecond capability remains to be tested. This is a scope issue rather than a technical error, but it affects the paper's main claim.","section":"Sec. 4 and Sec. 3.3"},{"comment":"The tracker weighting parameter m was 'optimized empirically based on the drift of the position estimate with respect to the true spot position' (Sec. 3.1), and m = 0.8 is used throughout the accuracy analysis. If the 'true spot position' used for this optimization is the same DM command signal that is later acknowledged to be imperfect, then the tracker may be tuned to an unreliable reference. Please report how m was optimized, whether the results are sensitive to m, and re-evaluate the optimization if the reference is corrected.","section":"Sec. 3.1 and Sec. 3.2.3"}],"minor_comments":[{"comment":"The caption correctly notes that the blue line is the signal sent to the DM, but this caveat should also be stated in the main text where Fig. 10 is discussed, so that readers do not mistake the command for the true induced tilt.","section":"Fig. 10 caption"},{"comment":"The plate scale is given as 15 arcsec/pixel, but the procedure for calibrating this plate scale is not described. Please provide the calibration method (e.g., known DM displacement or ruler measurement) so that the pixel-to-angle conversion is reproducible.","section":"Sec. 3.1"},{"comment":"The meaning of I1 and I2 in Eq. (5) is clear from context, but a brief definition (e.g., 'the intensities of two successive samples') would improve readability.","section":"Sec. 2, Eqs. (5)-(6)"},{"comment":"The superscript citation '33' appears at the end of the Fig. 11 caption, which is an unusual placement; consider moving the citation to the main text or using a standard figure-caption format.","section":"Sec. 3.2.3, Fig. 11"},{"comment":"The paper states that data and code are not publicly available. Given the empirical nature of the work, making at least the processed data (event counts, tracker outputs, and error values) available would strengthen reproducibility; at minimum, please clarify what can be obtained upon request.","section":"Code and Data Availability"},{"comment":"The phrase 'immunity to a constant background can be achieved' is slightly misleading given the earlier discussion in Sec. 2 that a constant background does affect the response. Consider rewording to 'the effect of a constant background can be compensated' to avoid apparent contradiction.","section":"Sec. 3.2.1"}],"recommendation":"major_revision","confidential_remarks":"The paper is within the scope of Optical Engineering and presents a useful characterization study. However, the central accuracy claim is weakened by the unreliable reference signal (DM command vs. actual tilt), and the extrapolation to milliarcsecond LGS sensing is explicitly deferred. Both issues are addressable in revision, but the authors need to either provide a trustworthy reference measurement or substantially soften the claims. The novelty is incremental rather than breakthrough, but the experimental data on event-based sensors for this application are valuable to the AO community."},"author_rebuttal":null,"desk_editor":{"model":"deepseek-v4-flash","letter":"Colleague,\n\nThe paper is a straightforward laboratory characterization of an event-based camera for laser guide star tip-tilt sensing. The new part is the application: first lab test of this sensor class for LGS tip-tilt, with a clear story about how background illumination and event threshold interact. The noise measurements and the threshold-matched conditions across backgrounds are useful and consistent with the logarithmic-response model. I believe the qualitative behavior: higher background reduces shot noise and leak events, and threshold tuning can equalize response.\n\nThe soft spot is the reference truth. The stress-test note is on target: the errors in Fig. 11 and the n=0.339 value treat the DM command signal as true tilt, while the paper itself says the DM had signal processing issues and the commanded sinusoid was imperfect. The caption of Fig. 10 even says the blue line is the signal sent to the DM, not the true tilt. So the reported 1-pixel / 15-arcsec error is relative to a reference the authors know to be unreliable. That doesn't kill the paper, but it means the central accuracy claim is not yet established. The authors are transparent about the DM issue, which I credit, but the analysis still uses the command as ground truth, so the numbers should be taken with a grain of salt.\n\nTwo smaller issues: Eq. (9) requires the aperture radius a, which is never given, so the λ/3 result can't be independently checked. And the abstract calls the sensor a 'strong contender' for LGS tip-tilt, but the demonstrated accuracy is 15 arcsec while the time-delay method needs milliarcseconds; the authors explicitly defer that to future work. That's a mismatch between abstract and body.\n\nIf the DM reference issue can be addressed—by measuring actual DM motion or acknowledging the uncertainty in the error bars—the paper would be a solid contribution. As it stands, it's a well-written feasibility study that points in a promising direction but doesn't yet support the application-level claim. I'd send it to peer review with a request for revision; a good referee can help tighten the claims and the error analysis.\n\nFor you: worth a skim if you work in wavefront sensing, but not a must-read. I would not cite it as evidence of milliarcsecond capability, but might cite it for the background/noise characterization.","headline":"A credible lab feasibility study for event-based tip-tilt sensing, but the headline accuracy is measured against a known-imperfect DM reference and the leap to milliarcseconds is openly deferred.","tokens_in":14218,"tokens_out":2379,"would_cite":false,"duration_ms":20061,"reading_group":"maybe","serious_thinker":"yes","would_accept_peer_review":true},"rs_alignment":null,"lean_confirmation":null,"pith_extraction":{"msc":[],"pacs":[],"model":"deepseek-v4-flash","headline":"Event-based camera measures laser-spot tilt to 15 arcseconds","keywords":["event-based detectors","tip-tilt sensing","laser guide star","adaptive optics","wavefront sensing","time-delay method","noise characterization"],"falsifier":"A direct test would be to re-run the same experiment with a smaller plate scale, for example 1 arcsecond per pixel or finer, and check whether the measured tilt error still reaches one pixel; if the spot then triggers too few events to keep the tracker locked, or the error scales worse than pixel size, the route to milliarcsecond sensing fails. An independent high-speed frame camera could also verify that the deformable mirror's true tilt matches the commanded signal, since the paper reports processing issues that made the mirror's motion imperfect.","tokens_in":13217,"feed_emoji":"🔭","tokens_out":6217,"duration_ms":51725,"temperature":0.7,"pith_summary":"Event-based detectors—cameras that emit asynchronous pixel events only when brightness changes—are proposed here as the missing sensor for laser guide star tip-tilt sensing. In laboratory tests with a 589-nm laser spot moved by a deformable mirror, an event-based camera recovered the tilt with error down to about one pixel, 15 arcseconds, under high background illumination, corresponding to roughly $\\lambda/3$ of wavefront tilt error. The authors show that tuning the contrast threshold reproduces the same measurement accuracy at lower background levels, effectively making the detector immune to a constant background. This matters because the time-delay method for laser guide stars needs milliarcsecond-scale differential tilt measurements, and frame-based detectors have not delivered them.","feed_headline":"Event-based camera measures laser-spot tilt to 15 arcsec","feed_subtitle":"Threshold tuning keeps accuracy stable across background levels, a step toward laser guide star adaptive optics.","key_machinery":"The central object is the event-based sensor, whose pixels compare the logarithm of intensity against stored levels and emit an event when the log-intensity change $\\Delta L(\\mathbf{x}_k,t_k)$ reaches a contrast threshold $C$, with polarity $p_k\\in\\{-1,+1\\}$. Because the difference is taken after the logarithm, a constant background $b$ enters the response as $\\log(I_1+b)-\\log(I_2+b)$, so it is not automatically cancelled—but the paper shows the threshold can be tuned to match event rates across backgrounds. The measurements are carried by an average position tracker, $\\hat{x}_{t_i}=m\\hat{x}_{t_{i-1}}+(1-m)x_i$, which updates the spot position event by event, and by the wavefront error relation $\\Delta\\omega_{\\text{tilt}}=2ay/f=n\\lambda$, which converts a tracked lateral displacement into a tilt error in waves.","core_discovery":"The central claim is that an event-based detector can serve as a laser guide star tip-tilt sensor: with the event threshold set appropriately, it tracks a moving laser spot to about one pixel of error even under bright background light, and the same accuracy is reachable at other background levels by changing the threshold. In the best configuration the residual tilt error gives $n=0.339$ in $\\Delta\\omega_{\\text{tilt}}=2ay/f=n\\lambda$, meaning the tilt is measured to roughly a third of a wave. Higher background illumination is not a liability: shot-noise-triggered events drop as illumination rises, and the remaining leak events are localized near the bright spot, which is why the brightest conditions gave the most accurate measurements.","pith_inferences":["If the plate scale is reduced without losing event rate, the same tracker could in principle reach the milliarcsecond regime, but the number of events per oscillation will drop; the paper's own log-background equations suggest that raising background or lowering threshold would be needed to compensate—a testable trade-off.","The matched-condition result hints at a general calibration procedure: record event counts at two backgrounds and interpolate the threshold offset, which could be automated for on-sky operation.","Because the spot's interior is silent in event data, centroid accuracy depends on edge events; a tracker that weights ON/OFF pairs or uses temporal bins may outperform the exponential average used here.","The noise characterization implies that in dark-sky conditions shot noise dominates, so nighttime astronomy would need an artificial background or bias adjustment; daytime laser-communication links avoid this naturally."],"forward_implications":["An event-based wavefront sensor could measure laser guide star tip-tilt without relying on a natural guide star, enabling the time-delay method and increasing sky coverage.","Daytime adaptive optics, including free-space optical communications, would benefit because higher sky background improves tip-tilt measurement accuracy instead of degrading it.","Microsecond-scale asynchronous readout removes frame-rate limits on sensing speed, so faster corrections are possible with lower data bandwidth.","Threshold tuning acts as a calibration knob that keeps tip-tilt accuracy stable as background illumination changes.","The demonstrated 15-arcsecond error is a proof of concept; reducing the plate scale is the stated path toward the milliarcsecond differential tip-tilt needed for the time-delay method."],"supporting_citations":[{"why":"Proposes the time-delay method that motivates the whole measurement; supplies the integration-time and sky-coverage equations used in the introduction.","marker":"[8]"},{"why":"Provides the average position tracker algorithm and the event-based wavefront sensing approach that the lab experiment adapts.","marker":"[12]"},{"why":"Defines the event-based sensor model and the log-intensity event equations used in the analysis.","marker":"[11]"},{"why":"Characterizes intensity-dependent shot noise, the basis for the finding that bright backgrounds improve accuracy.","marker":"[29]"},{"why":"Explains leak events from capacitor decay, used to interpret noise at the laser spot pixels.","marker":"[30]"},{"why":"Supplies the wavefront aberration formula converting spot displacement to tilt in waves.","marker":"[34]"}],"fun_headline_variants":["Event-based camera measures laser-spot tilt to 1/3 wave","Bright background sharpens event-based tip-tilt sensing","Threshold tuning enables event-based tilt sensing under glare","Event-based detector tracks laser guide star tilt to 1/3 wave"],"cache_read_input_tokens":3200,"weakest_assumption_plain":"The load-bearing premise is that a tracking accuracy of about 15 arcseconds demonstrated in the laboratory can be scaled to the milliarcsecond differential-tilt signals of the time-delay method by changing the plate scale, and that the deformable mirror's commanded signal is a reliable ground truth for tilt.","fun_headline_variants_meta":{"raw":{"variants":["Event-based camera measures laser-spot tilt to 1/3 wave","Bright background sharpens event-based tip-tilt sensing","Threshold tuning enables event-based tilt sensing under glare","Event-based detector tracks laser guide star tilt to 1/3 wave"]},"model":"deepseek-v4-flash","effort":"low","cost_usd":0.000499,"raw_usage":{"total_tokens":2376,"prompt_tokens":813,"completion_tokens":1563,"prompt_tokens_details":{"cached_tokens":384},"prompt_cache_hit_tokens":384,"prompt_cache_miss_tokens":429,"completion_tokens_details":{"reasoning_tokens":1493}},"tokens_in":429,"tokens_out":1563,"duration_ms":13674,"temperature":1.0,"reasoning_tokens":1493,"cache_read_input_tokens":384,"cache_creation_input_tokens":0},"cache_creation_input_tokens":0},"created_at":"2026-08-11T14:56:01.195873+00:00","model_set":{"reader":"deepseek-v4-flash"},"falsifier":"A direct test would be to re-run the same experiment with a smaller plate scale, for example 1 arcsecond per pixel or finer, and check whether the measured tilt error still reaches one pixel; if the spot then triggers too few events to keep the tracker locked, or the error scales worse than pixel size, the route to milliarcsecond sensing fails. An independent high-speed frame camera could also verify that the deformable mirror's true tilt matches the commanded signal, since the paper reports processing issues that made the mirror's motion imperfect.","supporting_citations":[{"cited_title":"Propagation delay of a laser beacon as a tool to retrieve absolute tilt measurements,","cited_arxiv_id":null,"evidence_quote":"Proposes the time-delay method that motivates the whole measurement; supplies the integration-time and sky-coverage equations used in the introduction."},{"cited_title":"Shack-Hartmann wavefront sensing using spatial-temporal data from an event-based image sensor,","cited_arxiv_id":null,"evidence_quote":"Provides the average position tracker algorithm and the event-based wavefront sensing approach that the lab experiment adapts."},{"cited_title":"Event-based vision: a survey,","cited_arxiv_id":null,"evidence_quote":"Defines the event-based sensor model and the log-intensity event equations used in the analysis."},{"cited_title":"Unraveling the paradox of intensity-dependent DVS pixel noise","cited_arxiv_id":"2109.08640","evidence_quote":"Characterizes intensity-dependent shot noise, the basis for the finding that bright backgrounds improve accuracy."},{"cited_title":"Shining light on the DVS pixel: a tutorial and discussion about biasing and optimization,","cited_arxiv_id":null,"evidence_quote":"Explains leak events from capacitor decay, used to interpret noise at the laser spot pixels."},{"cited_title":"Basic wavefront aberration theory for optical metrology,","cited_arxiv_id":null,"evidence_quote":"Supplies the wavefront aberration formula converting spot displacement to tilt in waves."}],"review_version":1}