3D Gamma-ray and Neutron Mapping in Real-Time with the Localization and Mapping Platform from Unmanned Aerial Systems and Man-Portable Configurations

A. Haefner; A. Moran; B.J. Quiter; E. Suzuki; J.W. Cates; K. Vetter; R. Barnowski; R. Pavlovsky; T.H.Y. Joshi; V. Negut

REVIEW 4 major objections 5 minor 1 cited by

3D Gamma-ray and Neutron Mapping in Real-Time with the Localization and Mapping Platform from Unmanned Aerial Systems and Man-Portable Configurations

T0 review · 4 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read A compact detector platform called NG-LAMP establishes that neutron sources can be localized in three dimensions and in real time from a drone or in hand, even when a much stronger gamma-ray source is nearby and even when the neutron…

desk verdict A real engineering first—real-time 3D neutron mapping on a drone—with the usual field-demo soft spots: no error bars, no ground-truth table, and no quantitative look at how clean the neutron channel really is. read the letter →

arxiv 1908.06114 v1 pith:FSUWDPBW submitted 2019-08-16 physics.ins-det

R. Pavlovsky , J.W. Cates , W.J. Vanderlip , T.H.Y. Joshi , A. Haefner , E. Suzuki , R. Barnowski , V. Negut

show 3 more authors

A. Moran K. Vetter B.J. Quiter

This is my paper · ORCID

classification physics.ins-det

keywords 3Dradiationmappingneutronlocalizationgamma-rayspectroscopyCLLBCscintillatorunmannedaerialsystemscenedatafusionspecialnuclearmaterialsreal-timereconstruction

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 reports a system that maps gamma-ray and neutron sources in three dimensions, in real time, from a small drone or a hand-carried unit. The authors integrate the existing LAMP scene-data-fusion pipeline with four CLLBC scintillator crystals that detect both gamma rays and neutrons, calling the result NG-LAMP. They claim first demonstrations: a neutron source localized while a much stronger gamma-ray source sat nearby; simultaneous spectroscopic localization of three gamma sources plus a neutron source; and localization of a shielded plutonium surrogate by neutron signal alone, with the gamma-ray reconstruction flat as background. If true, this extends 3D scene-data-fusion radiation mapping from gamma-only to dual-particle search, which matters for finding shielded special nuclear materials.

What carries the argument

The load-bearing mechanism is the fusion of two established pieces: the scene-data-fusion pipeline, which couples a simultaneous localization and mapping (SLAM) engine with maximum-likelihood expectation-maximization (MLEM) over voxelized space, and the CLLBC scintillator, a crystal that responds to both gamma rays and thermalized neutrons with distinct pulse shapes and pulse heights. For each detected event, pulse-shape and pulse-height discrimination assigns it to a gamma-ray or neutron channel, and angular-dependent detection efficiencies per detector module enter the MLEM update. This is what lets the same reconstructed volume separate a neutron hotspot from a nearby gamma-ray hotspot using only particle discrimination, and lets photopeak windowing run separate 3D reconstructions for each isotopic gamma signature.

What would settle it

Run NG-LAMP with a strong gamma-ray source and no neutron source, process the data through the neutron-only reconstruction, and look for a false hotspot. A quantitative version would measure the gamma rejection ratio of the CLLBC neutron channel and require that the reconstructed neutron intensity at the gamma source stay within background noise before accepting the neutron-localization claim.

Watch

Extended reading notes

Core claim

The central claim is that a single lightweight platform, carrying four monolithic CLLBC (Cs2LiLa(Br,Cl)6:Ce) scintillators and an onboard computer, can perform 3D maximum-likelihood reconstruction of both gamma-ray and neutron sources in real time, using only particle-type or photopeak selection to separate the channels. The paper reports three demonstrations: an eight-minute UAS flight that localized a 500 microcurie cobalt-60 source and a one-significant-quantity plutonium surrogate, with the neutron source shielded by 14 cm of lead, in the same reconstruction volume; a walk-through of four rooms that localized three photopeak-windowed gamma sources and a PuBe neutron source at about 2 Hz update rate; and a UAS flight around a vehicle whose shielded plutonium surrogate emitted a gamma dose rate indistinguishable from background, where the neutron reconstruction peaked at the source while the gamma reconstruction stayed nearly uniform. The authors conclude that NG-LAMP localizes neutron point sources in 3D and in real time in measurements under ten minutes, without changing reconstruction parameters except energy or particle cuts.

Load-bearing premise

The claim depends on CLLBC's pulse-shape and pulse-height discrimination cleanly separating neutron events from gamma-ray events; if strong gamma-ray counts leak into the neutron channel, the apparent neutron hotspot could actually be the gamma source.

Editorial extensions

If this is right

NG-LAMP can localize a neutron source in the presence of a far more active gamma-ray source from a single sUAS flight of about eight minutes, using only particle discrimination between channels.
The same reconstruction parameters, with only energy or particle cuts changed, produce gamma-ray maps that remain background-like when the gamma signal is shielded, giving an internal consistency check on the neutron-only localization.
Spectroscopic photopeak windowing supports near-real-time (about 2 Hz update rate) simultaneous localization of multiple isotopic gamma sources plus a neutron source in a handheld survey.
Both fission-neutron (Pu surrogate) and alpha-n (PuBe) sources were localized, so the method is not tied to one neutron energy spectrum.
Switching between handheld, ground-vehicle, and sUAS mounts takes simple mechanical changes under five minutes, so the same system can serve search and response roles.

Reading between the lines

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

Because the gamma-ray reconstruction is used as a null check when the source is heavily shielded, the same dual-channel data could be used to estimate shielding thickness or residual gamma attenuation by comparing the two reconstructions.
The demonstrated sensitivity is limited by neutron count rate and CLLBC's low fast-neutron sensitivity; a natural next test is counting-rate-limited detection-distance curves for one significant quantity of plutonium under realistic shielding.
The architecture should extend to distributed neutron sources such as contamination by keeping MLEM rather than point-source fitting, though the paper only demonstrates point-like neutron sources.
Neutron-only localization to a specific vehicle compartment suggests that wide-area aerial neutron search could be practical, but the false-alarm rate from gamma leakage into the neutron channel remains the key quantity to characterize.

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.

Desk Editor's Note

read the letter

Colleague,

What you should know: this is the first believable demonstration of real-time 3D neutron localization from a small UAS and handheld platform, using the LAMP architecture they've built over the past few years. It extends their gamma-ray SDF work by swapping in CLLBC detectors that see both gammas and neutrons, and the field tests are real: a concealed-source exercise with 60Co and a shielded Pu surrogate, a four-room spectroscopic survey, and a shielded 1-SQ surrogate in a vehicle. The paper is an engineering report, not a physics breakthrough, but the integration and the field results look genuine.

What's good: the multi-source spectroscopic localization (Figures 4/5) is a nice demonstration—photo-peak windowing on three gamma sources and neutron capture simultaneously, with ~2 Hz updates. The shielded-SNM scenario is the one that matters for the application, and the reconstruction showing a neutron hotspot with gamma consistent with background is exactly the right data product. I also appreciate that they state clearly that pulse-height discrimination was used for most results, rather than hiding that fact.

Where I'd push back: the evidence is qualitative. There are no localization error numbers, no ground-truth coordinates, no statistical comparison of reconstructed source positions to known positions. The "near uniform" gamma background is asserted, not quantified. And the biggest gap: the neutron/gamma separation quality is never characterized. With a 500 μCi 60Co source at ~1 m standoff, the gamma count rate in CLLBC is high. They used a pulse-height cut for Figure 3, and they don't report gamma rejection ratio or a control run with 60Co present and the neutron source absent. The stress-test scenario—pile-up leaking into the neutron window—is plausible in principle. But note: if that leakage were the dominant neutron signal, the reconstructed hotspot would sit on the 60Co truck, not on the SUV ten meters away where they claim the Pu surrogate. So I don't think the Figure 3 claim is likely an artifact; still, the missing control measurement makes the confidence lower than it should be.

Bottom line: this is a solid, useful engineering demonstration that deserves peer review. I'd send it out, with the reasonable request that the authors add a small set of quantitative validations: table of reconstructed vs known source positions, a leakage measurement for the neutron channel under high gamma rate, and a statistical description of the background reconstruction. For those who work in applied nuclear detection, this is worth a read.

Recommendation: accept for review; expect a conditional decision pending those additions.

Referee Report

4 major / 5 minor

Summary. The paper describes Neutron Gamma LAMP (NG-LAMP), an extension of the LAMP platform that combines four CLLBC scintillators with real-time 3D scene data fusion to localize gamma-ray and neutron sources simultaneously. Three demonstrations are reported: a UAS flight that simultaneously localized a 500 μCi 60Co source (gamma rays) and a 1 SQ Pu surrogate (neutrons) placed about 10 m apart; a handheld survey that spectroscopically localized three gamma-ray sources and one PuBe neutron source; and a UAS flight around a vehicle containing a heavily shielded Pu surrogate, where the gamma-ray reconstruction is said to be consistent with a near-uniform background and the neutron reconstruction localized the source. The central claims are that this is the first real-time 3D neutron-localization capability from a small UAS/handheld platform, and that neutron signatures alone can locate shielded special nuclear material.

Significance. If the claims hold, the work is significant for nuclear security and emergency response: it would be the first demonstrated real-time 3D dual-particle mapper from a small UAS and handheld configurations, extending prior LAMP gamma-ray work to neutron signatures. The paper benefits from field-scale demonstrations, independent scenarios, and the use of existing, published MLEM/SLAM components rather than any circular fitting to the results. The manuscript is generally clear about what was measured, and the reconstructions appear consistent with the described source placements. However, the central claims rest on an unquantified neutron-selection purity and on qualitative statements about localization and background uniformity, which are not yet supported by the evidence presented.

major comments (4)

[Methods and Approach; Results A, Figure 3] The neutron-selection purity is not quantified, and this is load-bearing for the headline claim of neutron localization in the presence of a strong gamma-ray source. The manuscript states in Methods that 'Pulse height discrimination is used as a matter of convenience for most results that follow, due to low fast neutron sensitivity in CLLBC,' and Figure 3's caption says the neutron map was produced with 'only particle discrimination' applied. If the Figure 3 discrimination is a pulse-height cut rather than pulse-shape discrimination, then gamma-ray pile-up from the 500 μCi 60Co source could leak into the neutron window, especially near the truck. The paper reports no neutron-window energy spectrum, no gamma rejection ratio, no count rates in the neutron channel as a function of position, and no control measurement with the 60Co present and the neutron source absent. Without these, the possibility that the apparent neutron hotspot is biased or even caused by gamma leakage is not excluded. Please report the measured gamma rejection into the neutron channel and the neutron-window count rates along the flight path, or otherwise demonstrate that the neutron-only map is not contaminated.
[Results A, B, and C (Figures 3, 4, 6)] No quantitative localization accuracy or ground-truth comparison is provided. The text says sources were 'correctly localize[d]' and that the neutron source was localized 'to the vehicle' or 'in the back of a vehicle,' but no coordinates, distances to known source positions, or localization-error metrics are given. For a paper whose central claim is 3D real-time localization, at least a table of true versus reconstructed source positions (or a distance error per source) is needed to substantiate the claim. Without this, the demonstrations show that hot spots appear in plausible locations, but not that the system localizes with any quantified accuracy. Please add quantitative comparisons for each demonstration.
[Results C, Figure 6] The claim that the gamma-ray reconstruction is 'consistent with a near uniform distribution' is unsupported. The manuscript provides no measure of uniformity (e.g., a chi-square statistic, a flatness metric, or a comparison against a background-only measurement), no expected background model, and no uncertainty. Since the final scenario explicitly relies on gamma-ray data being 'consistent with background,' this qualitative statement is load-bearing. Please provide a quantitative uniformity test or the reconstructed gamma-ray source distribution with error bars, and state the acceptance criterion used to judge it 'nearly uniform.'
[Methods and Approach; Conclusion] The reconstruction parameters are underspecified, which limits reproducibility and the strength of the 'we did not alter gamma-ray or neutron reconstruction parameters' claim. The paper states that three MLEM iterations were used for the Figure 4 data, but does not report iteration counts, voxel sizes, time binning, or the angular-detection-efficiency model for the other reconstructions. Since MLEM iteration count and the efficiency model directly affect the reconstructed distributions and the claimed 'same reconstruction parameters' comparison in Figure 3, please specify these parameters for every demonstration or state where they are defined in the cited LAMP references.

minor comments (5)

[Abstract and Introduction] The phrase '3.5% FWHM at 661.7 keV' would be clearer as 'energy resolution of about 3.5% FWHM at 661.7 keV.' Also, the sentence in the Introduction that begins 'This is accomplished with four monolithic radiation detectors...' would read better if split into two sentences.
[Figure 4 caption] The caption says 'Three iterations of MLEM were used to estimate source locations,' but the body text in Section B does not mention the iteration count for the other figures. Please state the iteration count and all reconstruction parameters in one place, or in each caption.
[Figure 6 caption] The color coding is described in the body text as 'neutron source in red' and 'gamma-ray data... in blue,' but the caption in the provided text does not reproduce this color description. Please make the caption self-contained so the figure is interpretable without the body text.
[References] Reference [1] is to an arXiv preprint, but the published version or a more stable citation would be preferable. Reference [6] for RMD lists only the company name and city; a product page or datasheet would be more useful.
[Throughout] There are typographical inconsistencies, such as 'Simultaneous and Mapping (SLAM)' in the Methods section, which should be 'Simultaneous Localization and Mapping,' and 'K. Vetter' appearing with lowercase 'k' in Reference [1]. A careful proofreading pass is recommended.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the paper is an experimental systems demonstration whose results come from independent field measurements, not from a fitted parameter or self-referential derivation.

full rationale

The work reported in arXiv:1908.06114 is an experimental demonstration, not a derivation chain. The central claims are that NG-LAMP can localize neutron sources in 3D in real time, including in the presence of a strong gamma-ray source, and that it can simultaneously localize multiple gamma-ray and neutron sources spectroscopically. These claims are supported by independent measurements: a UAS flight around a 500 uCi 60Co source and a shielded 1 SQ Pu surrogate separated by about 10 m (Figure 3), a handheld four-room survey with three 10 uCi gamma-ray sources and a shielded 1 Ci PuBe neutron source (Figures 4 and 5), and a separate sUAS flight around a heavily shielded 1 SQ Pu surrogate with gamma-ray emissions consistent with background (Figure 6). The reconstruction pipeline uses MLEM with a citation to the authors' prior SDF work, but that citation supplies an already-published algorithmic tool rather than the target result; the present paper's contribution is the CLLBC-based dual-particle detector integration and the new measurement outcomes. No parameter is fitted to the demonstration data and then renamed as a prediction: the paper states that gamma-ray and neutron reconstructions use the same MLEM parameters except for particle and energy cuts. The pulse-height-limited neutron/gamma separation is a potential measurement-quality limitation, since gamma pileup from the strong 60Co source could in principle leak into the neutron selection window, but that is an experimental artifact risk, not a circularity: the neutron hotspot is not algebraically constructed from the gamma-ray map or from the claim under test. The self-citations to LAMP and SDF are contextual and load-bearing only in the ordinary sense of citing methods, and they are not used to forbid alternatives or to justify a fitted quantity. Accordingly, the paper is self-contained against its external benchmarks and warrants a circularity score of 0.

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

The central claim rests on the MLEM/SDF framework from prior work, the SLAM pose estimates, and the assumption that neutron/gamma discrimination is clean. There are no new physical entities or fitted constants introduced. The main hidden parameters are the detector efficiency model, energy windows, MLEM iterations, and voxel grid.

free parameters (4)

MLEM iteration count = 3
Three iterations of MLEM are used for the reconstructions in Figures 4 and 5; this choice affects convergence and the sharpness of the source distribution.
Angular detection efficiency model
Angular-dependent gamma-ray detection efficiencies are incorporated into the MLEM solution, but the text does not specify whether they come from simulation or calibration. This model directly influences the reconstructed source location.
Energy windows and particle cuts = Photo-peaks of 137Cs, 60Co, 241Am/133Ba, and neutron capture region
Spectral windows are chosen for each source to create separate reconstructions; the choices affect which counts contribute to each map and are not justified quantitatively.
Reconstruction voxel grid and time binning
The 3D voxel grid and the integration time bins for MLEM are not specified in the text; both are user-selected parameters that affect resolution and update rate.

assumptions (4)

standard math MLEM converges to a plausible source distribution given accurate poses and a correct detector response model.
The MLEM algorithm is a standard expectation-maximization method; the paper relies on its properties without proving them.
standard math Google Cartographer SLAM provides platform poses accurate enough for radiation mapping.
The SLAM solution is used to assign radiation counts to 3D positions; if pose errors are large, the reconstruction is degraded.
domain assumption CLLBC pulse-height and pulse-shape discrimination cleanly separates neutron from gamma-ray events.
This assumption is essential for the neutron-only reconstructions, but no discrimination curve or quantitative cross-talk measurement is provided.
domain assumption The surrogate Pu source emits neutrons at a rate equivalent to 1 significant quantity of Pu, and the 14 cm lead/tungsten shielding reduces gamma dose to background levels.
These are experimental descriptions used to define the source terms; they are plausible but not independently verified in the paper.

how reviews work

0 comments

Cite this review

Pith. "Pith review of 3D Gamma-ray and Neutron Mapping in Real-Time with the Localization and Mapping Platform from Unmanned Aerial Systems and Man-Portable Configurations." pith.science (2026). https://pith.science/paper/FSUWDPBW

@misc{pith2026190806114,
  author       = {Pith},
  title        = {Pith review of: 3D Gamma-ray and Neutron Mapping in Real-Time with the Localization and Mapping Platform from Unmanned Aerial Systems and Man-Portable Configurations},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/FSUWDPBW}},
  note         = {Machine review of arXiv:1908.06114}
}

abstract

Nuclear Scene Data Fusion (SDF), implemented in the Localization and Mapping Platform (LAMP) fuses three-dimensional (3D), real-time volumetric reconstructions of radiation sources with contextual information (e.g. LIDAR, camera, etc.) derived from the environment around the detector system. This information, particularly when obtained in real time, may be transformative for applications, including directed search for lost or stolen sources, consequence management after the release of radioactive materials, or contamination avoidance in security-related or emergency response scenarios. 3D reconstructions enabled by SDF localize contamination or hotspots to specific areas or objects, providing higher resolution over larger areas than conventional 2D approaches, and enabling more efficient planning and response, particularly in complex 3D environments. In this work, we present the expansion of these gamma-ray mapping concepts to neutron source localization. Here we integrate LAMP with a custom $Cs_2LiLa(Br,Cl)_6:Ce$ (CLLBC) scintillator detector sensitive to both gamma-rays and neutrons, which we dub Neutron Gamma LAMP (NG-LAMP). NG-LAMP enables simultaneous neutron and gamma-ray mapping with high resolution gamma-ray spectroscopy. We demonstrate the ability to detect and localize surrogate Special Nuclear Materials (SNM) in real-time and in 3D based on neutron signatures alone, which is critical for the detection of heavily shielded SNM, when gamma-ray signatures are attenuated. In this work, we show for the first time the ability to localize, in 3D and realtime, a neutron source in the presence of a strong gamma-ray source, simultaneous and spectroscopic localization of three gamma-ray sources and a neutron source, and finally the localization of a surrogate SNM source based on neutron signatures alone, where gamma-ray data are consistent with background.

Figures

Figures reproduced from arXiv: 1908.06114 by the authors.

**Figure 1.** (Left) Picture of the NG-LAMP system, and (right) NG-LAMP mounted on sUAS (DJI Matrice 600). NG-LAMP performs 3D SDF onboard in real-time. SDF requires a Simultaneous and Mapping (SLAM) engine, which is implemented in the LAMP software framework using Google Cartographer [9] to provide position and orientation estimates. Maximum likelihood expectation maximization (MLEM) [2] is performed in 3D for all poses, as illu… view at source ↗

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

Surrogate distributed radiological sources III: quantitative distributed source reconstructions
physics.ins-det 2024-12 conditional novelty 6.0 of 10

Quantitative aerial reconstructions of eight surrogate distributed gamma-ray sources achieved shape agreement (structure coefficient up to 0.94) and total activity within about 15% after applying a fitted calibration factor.

Reference graph

Works this paper leans on

3 extracted references · 2 canonical work pages · cited by 1 Pith paper

[6]

Radiation Monitoring Devices, Watertown MA 024724

work page
[7]

Velodyne Lidar, San Jose CA 95138

work page
[8]

Real-time loop closure in 2D LIDAR SLAM,

DJI Matrice 600. https://www.dji.com/matrice600. 2019 [9] W. Hess, D. Kohler, H. Rapp and D. Andor, "Real-time loop closure in 2D LIDAR SLAM," 2016 IEEE International Conference on Robotics and Automation (ICRA) , Stockholm, 2016, pp. 1271-1278. doi: 10.1109/ICRA.2016.7487258 [10] “IAEA International Safeguards Glossary 2001,” Accessed Aug 15 2019. ht...

arXiv 2019

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [6]

Radiation Monitoring Devices, Watertown MA 024724

work page

[2] [7]

Velodyne Lidar, San Jose CA 95138

work page

[3] [8]

Real-time loop closure in 2D LIDAR SLAM,

DJI Matrice 600. https://www.dji.com/matrice600. 2019 [9] W. Hess, D. Kohler, H. Rapp and D. Andor, "Real-time loop closure in 2D LIDAR SLAM," 2016 IEEE International Conference on Robotics and Automation (ICRA) , Stockholm, 2016, pp. 1271-1278. doi: 10.1109/ICRA.2016.7487258 [10] “IAEA International Safeguards Glossary 2001,” Accessed Aug 15 2019. ht...

arXiv 2019