Pith. sign in

REVIEW 4 major objections 6 minor 53 references

An Underwater, Fault-Tolerant, Laser-Aided Robotic Multi-Modal Dense SLAM System for Continuous Underwater In-Situ Observation

T0 review · 4 major / 6 minor · reviewed 2026-08-16 · deepseek-v4-flash

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

Pith's one-line read Water-DSLAM claims continuous dense underwater SLAM through sensor dropouts, with 0.039 m RMSE and roughly tenfold denser maps than compared methods.

desk verdict Serious systems paper with real field experiments; the central claim is plausible but the fault-tolerance story has an untested hole where the trusted DP-INS reference itself drifts. read the letter →

arxiv 2504.21826 v1 pith:IHXPPKCK submitted 2025-04-30 cs.RO

Yaming Ou , Junfeng Fan , Chao Zhou , Pengju Zhang , Zongyuan Shen , Yichen Fu , Xiaoyan Liu , Zengguang Hou This is my paper · ORCID
classification cs.RO MSC 68T40
keywords underwaterSLAMdensemappingstructuredlightmulti-sensorfusionfaulttoleranceESKFfactorgraphin-situobservation
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 claims that underwater SLAM can remain continuous and dense even in darkness, texture-poor water, and structurally degraded scenes, by fusing a high-frequency inertial-acoustic-pressure odometry backbone with a laser-scanning structured-light system and a stereo camera. The authors build a robot, Water-Scanner, with a custom underwater binocular structured-light module, and a SLAM system, Water-DSLAM, whose three parallel subsystems back each other up when any external sensor drops out. If the reported results hold, the payoff is that an underwater robot can keep localizing and mapping for minutes at a time in conditions where current vision- or sonar-based systems either lose track or produce sparse maps. The paper's headline numbers are 0.039 m trajectory RMSE with 100 percent continuity during partial sensor dropout, and roughly tenfold denser point clouds than compared methods.

What carries the argument

The carrying mechanism is the triple-subsystem front-end plus an asynchronous factor graph. UBSL, an underwater binocular structured-light module whose laser line is swept by an oscillating mirror at 2 Hz, produces 70 Hz scans that only become usable sweeps if each scan line is undistorted by the robot's motion; DP-INS supplies the 100 Hz poses that make this undistortion possible. Water-UBSL then tightly couples Sweep-NDT registration with DP-INS in an IESKF, Water-Stereo couples DP-INS with stereo feature optimization, and a multi-modal factor graph adds, interpolates, and removes nodes and edges as constraints arrive at different rates, gated by the fault-detection module.

What would settle it

Run the system in a scene where the DP-INS is deliberately biased (for example, by unmodeled DVL scale error or a strong water current) while the stereo and structured-light sensors are healthy, then check whether the fault-detection gates keep rejecting valid corrections; if valid constraints are rejected and the trajectory diverges, the central assumption fails.

Watch

Extended reading notes

Core claim

Water-DSLAM claims to be the first structured-light-based underwater multi-modal dense SLAM system that runs continuously through partial sensor faults. The central discovery is that a high-frequency ESKF-based DP-INS backbone, fusing IMU, DVL, and pressure data, can carry the system across dropouts of the external perception sensors, while tightly coupled UBSL and stereo subsystems supply relative constraints that correct drift, and a multi-modal factor graph back-end with statistical fault detection selectively accepts those constraints. In experiments, the system reports 0.039 m trajectory RMSE with 100 percent continuity during partial sensor dropout, and point-cloud densities around 6922.4 points per cubic meter over a 750 cubic meter sinkhole volume, roughly ten times denser than the compared fixed-line structured-light and multi-modal baselines.

Load-bearing premise

The whole fault-tolerance scheme trusts the DP-INS backbone as the reference for accepting or rejecting the other sensors' corrections, so if DP-INS itself drifts or its noise model is wrong, the system has no independent check.

Editorial extensions

If this is right

  • If the claims hold, underwater robots can keep estimating pose and building dense maps through partial sensor failures because the DP-INS backbone continues when DVL, pressure, stereo, or structured-light measurements drop out.
  • Dense structured-light mapping at roughly 7000 points per cubic meter makes fine-scale observation possible in darkness and texture-sparse water, such as inspecting cave walls, boulders, and river structures.
  • The fault-detection and asynchronous graph maintenance strategy provides a template for fusing intermittent heterogeneous sensors on other underwater vehicles with different sensor suites.
  • The reported 100 percent continuity under partial dropout, if robust, removes the need for frequent reinitialization, a known failure mode in underwater visual-inertial odometry.

Reading between the lines

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

  • Inference: because the fault-detection gates treat DP-INS as the trusted reference, a natural extension is reciprocal validation, where loop closures or UBSL constraints are used to detect DP-INS drift and trigger re-calibration; the paper does not claim this.
  • Inference: the point-density metric (points per cubic meter) may not be directly comparable across methods unless the surveyed volume and overlap are identical; future comparisons could standardize density over the same surveyed region.
  • Inference: the same fusion pattern of active structured light with inertial and acoustic sensing could plausibly transfer to low-visibility non-underwater settings such as dusty or smoke-filled environments, though the paper does not make that claim.
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 / 6 minor

Summary. The paper presents Water-DSLAM, a multi-modal dense SLAM system for underwater robots, together with a custom sensor platform (Water-Scanner) that includes an IMU, DVL, pressure sensor, stereo camera, and a self-designed scanning underwater binocular structured-light (UBSL) module. The front-end consists of three subsystems: DP-INS (an ESKF fusing IMU, DVL, and pressure data), Water-UBSL (IESKF-based tight coupling of UBSL sweeps with DP-INS), and Water-Stereo (DP-INS-aided stereo visual odometry). The back-end is a factor graph that accepts or rejects subsystem constraints through fault-detection thresholds and maintains asynchronous, intermittently available factors. Experiments in a pool, a dark underwater scene, a 16-meter-deep sinkhole, and a field river report trajectory RMSE of 0.039 m at 100% continuity under partial sensor dropout and dense mapping at 6922.4 points/m^3 in a 750 m^3 volume.

Significance. If the reported results hold, this would be a valuable and rare integrated system: it combines a custom structured-light scanner with inertial-acoustic-visual fusion in a fault-tolerant architecture, and it is evaluated in real underwater environments rather than simulation. The pool experiments use an independent Apriltag-based ground truth, the derivations in Sections V and VI are largely standard and internally consistent, and the hardware and data collection are substantial. The main contributions—tight UBSL/DP-INS coupling, factor-graph maintenance under asynchronous sensor faults, and high-density structured-light mapping in dark underwater scenes—are potentially significant for underwater in-situ observation. However, the central fault-tolerance claim and the quantitative superiority claims are not yet fully supported by the evidence presented.

major comments (4)
  1. [VI.A (Eqs. 43-49), V.A.6, III.B] The fault-detection mechanism treats DP-INS as the trusted reference: UBSL and Stereo constraints are accepted or rejected according to their discrepancy from DP-INS relative poses, with thresholds τt=2.0σ_INS and τR=4.0σ_INS. But DP-INS is not an independent truth source; it fuses DVL and PS, which Section III.B explicitly identifies as external sensors subject to temporary faults. The sliding-window statistical test in V.A.6 can reject large outliers but cannot detect a slowly growing bias in DVL or PS (e.g., from acoustic multipath or pressure drift), because such a bias is consistent within the window. In that regime, valid UBSL and Stereo constraints would be discarded as inconsistent with the biased DP-INS, and the factor graph is initialized and anchored to the same biased reference. None of the experiments inject DVL or PS bias or dropout inside DP-INS; they only test dropout of UBSL and Stereo while DP-INS remains healthy. Thus the central 'uninterrupted, fault-tolerant' claim is not established for the case where the trusted reference itself drifts.
  2. [Tables II-VI; VII.A] Every quantitative result appears to come from a single run per condition. The tables report no standard deviations, no repeated trials, and no statistical comparison, although the pool environments and the free-motion trajectories are likely subject to considerable run-to-run variability. In addition, the baselines were not the original implementations but 'only reproduce core ideas on our platform' (VII.A), which makes the comparisons less definitive. For instance, the headline improvement of 0.039 m over 0.068 m ATE RMSE in Table VI, or the 100% versus ~25% continuity ratios, could be within run-to-run variation or implementation differences. To support the claimed superiority, the authors should provide multiple runs per method with mean and standard deviation (or per-run values), and state exactly which parts of the baseline pipelines were reproduced.
  3. [VII.F, VII.G, Abstract] The '10 times denser than existing methods' claim is based on comparing point densities such as 6922.4 points/m^3 versus 664.2 points/m^3, but point density is not a method-invariant quantity. It depends on the scanning pattern, robot speed, standoff distance, sensor field of view, trajectory, and the volume over which the density is computed. The compared baselines use fixed-line structured light and different trajectories, so a density ratio does not by itself demonstrate a superiority of the mapping algorithm. The authors should either define a controlled density metric (e.g., same trajectory, same scanned volume, same distance range) or present additional metrics such as completeness, map consistency, and error versus an independent scan of the scene.
  4. [V.A.6, VI.A (Eqs. 43-49)] The fault-detection pipeline has several free thresholds—ξ, τt=2.0σ_INS, τR=4.0σ_INS, qthresh, Nthresh, ε—but no sensitivity analysis is provided. Since the entire fault-tolerance and 'uninterrupted operation' claim hinges on these thresholds, the authors should show how the system behaves when they are varied over reasonable ranges, and ideally justify σ_INS calibration. In Eq. (44), Nmatched and the eigenvalues λ3/λ1 are used to flag structural degeneracy, but Nmatched is never defined and the eigenvalue source (which point cloud, what neighborhood) is not specified; without these definitions the structural-fault criterion is not reproducible.
minor comments (6)
  1. [VI.C.8, Eq. (58)] The MAP estimation in Eq. (58) writes min_X while summing over xi∈Y and defining residuals in terms of Y; the optimization variable should be Y, not X.
  2. [VI.B, Eqs. (52), (55), (56)] There are repeated typos: 'ndoe' should be 'node'. Also the UBSL and Stereo factor definitions refer to 'the current node yj and the previous node yj'; it appears one of these should be yi.
  3. [VI.B, Eqs. (55), (56)] The frame notation is inconsistent: the residuals use T^W_yj and T^W_yi but the preceding text defines poses as T^I_yi; the frames should be aligned or explicitly defined.
  4. [VII.D.2] In the mapping evaluation paragraph, the text mentions 'USBL constraints' (line beginning 'Water-Scanner, based on a scanning structured light system') but the system is consistently called UBSL elsewhere; this should be corrected.
  5. [Table V] The table uses ∞* for interrupted runs and color coding, but the meaning of ∞ is not explained in the caption or text; the reader cannot tell whether this is a divergent estimate, a failed optimization, or missing data.
  6. [Abstract and I.B] The claims 'first complete solution' and 'first structured-light-based underwater multi-modal dense SLAM system' are strong and difficult to verify; the authors should soften or qualify them relative to the specific sensor configuration and public evidence.

Circularity Check

1 steps flagged · score 2.0 of 10

Fault-detection gate uses DP-INS as the reference for constraints that were themselves estimated with DP-INS as an input; final trajectory RMSE is still externally anchored.

  1. self definitional [Sec. VI.A.1, Eqs. (43)-(45); Sec. V.B.1, Eq. (32)]
    "First, the discrepancy between point cloud matching and DP-INS estimates is quantified ... ∆t =∥tUBSL− tINS∥2, ∆R =∥ logSO(3)(R⊤UBSLRINS)∨∥2 ... A constraint is flagged as faulty if any of the following conditions are met: F = (1, if (∆t>τ t)∨ (∆R>τ R)∨ (S = 1), 0 otherwise) ... z−h(xk) = [pDP−INS−pk, logSO(3)(R⊤kRDP−INS)∨]⊤."

    The t_UBSL/R_UBSL being checked are not raw UBSL registrations; they are outputs of the Water-UBSL IESKF, whose observation model includes the DP-INS pose as a measurement (Eq. 32). The 'inconsistency' measured in Eq. 43 is therefore the deviation of an estimate that already contains DP-INS from DP-INS itself. A bad UBSL match is pulled toward DP-INS by the filter and can pass the gate, while a biased DP-INS (e.g., from undetected DVL/PS faults) makes good UBSL matches look faulty and discards the corrections needed to fix the bias. The same pattern appears in Sec. VI.A.2 for stereo: the Water-Stereo sliding-window optimization contains the DP-INS factor (Eq. 41) and is then gated against DP-INS (Eqs. 46-49). This is a self-referential validation loop rather than an independent check.

full rationale

Most of the paper's quantitative claims are anchored externally: pool trajectories are evaluated against an Apriltag marker detected by a global camera, DVL and PS are physical sensors providing absolute measurements, and the reported mapping densities are direct point-cloud statistics. The IESKF and factor-graph equations are standard multi-sensor fusion formulations, and the comparisons to VINS-Fusion, DIP-Fusion, Visual-DVL Fusion, and structured-light baselines are against external methods. The one genuinely self-referential element is the fault-detection design: DP-INS is both an input to Water-UBSL and Water-Stereo and the reference against which those subsystems' constraints are accepted or rejected, so the 'fault-tolerant' validation does not cover the case where the trusted reference itself drifts. This is a real but limited self-reference; it does not reduce the headline trajectory RMSE or density numbers to the paper's own outputs. Score 2 reflects one minor load-bearing self-reference, not a fully circular derivation chain.

Assumptions & free parameters 7 free parameters · 6 assumptions · 0 invented entities

The system relies on standard ESKF/IESKF theory and factor graph optimization. The main free parameters are the fault-detection thresholds and noise covariances, none of which are reported with values or sensitivity analysis. The domain assumptions concern the reliability of the IMU, DVL, UBSL model, magnetometer, and static world; if any is violated, the fault-tolerance and mapping claims weaken. No new theoretical entities are postulated; the UBSL hardware is a physical system, not an invented theoretical construct.

free parameters (7)
  • Fault-detection outlier threshold ξ (DP-INS) = 3 (default; no reported value used in experiments)
    Used in Section V.A.6 to reject DVL/PS outliers if ||M-e|| > ξ||s||. Value affects which measurements are accepted; no sensitivity analysis.
  • UBSL fault-detection thresholds τ_t=2.0σ_INS, τ_R=4.0σ_INS = multipliers 2.0 and 4.0
    Section VI.A.1 Eq. 45; multipliers chosen by hand; no ablation reported.
  • UBSL structural-degeneracy thresholds N_thresh and ε = not reported
    Section VI.A.1 Eq. 44; used to flag unreliable point-cloud matches based on matched-point count and planarity eigenratio; values absent.
  • Water-Stereo fault thresholds q_thresh, τ_t, τ_R = not reported
    Section VI.A.2 Eq. 49; q_thresh and residual thresholds are required to decide stereo constraint validity; no values given.
  • NDT voxel size and Gauss-Newton convergence parameters = not reported
    Section V.B; the incremental NDT local map uses voxels, but voxel resolution and iteration counts are not specified, affecting matching quality.
  • Factor graph node rate (5 Hz) = 5 Hz
    Section VI.C.1; downsampled DP-INS odometry used for node creation; design choice that affects graph size and accuracy.
  • DP-INS ESKF noise covariances Q and V = not reported
    Equations 3-5 define process noise Q and measurement noise V; no values are given, so the fusion weights in DP-INS are unspecified.
assumptions (6)
  • standard math IESKF update equations are valid as derived in FAST-LIO [48]; the paper applies them without re-deriving convergence or observability.
    Section V.B Eqs. 26-30 adopt the IESKF formulation directly from [48]; the paper relies on its correctness for the tight coupling between UBSL and DP-INS.
  • domain assumption The IMU provides continuous, valid data; only external sensors fail.
    Section IV states 'the internal sensor IMU continuously provides valid data (although with increasing cumulative errors)'. This is load-bearing for the fault-tolerant architecture; if the IMU fails or saturates, the entire system degrades.
  • domain assumption The DVL velocity observation model v_dvl = R_D^I v is unbiased when the DVL has bottom lock.
    Section V.A.3 uses DVL velocity directly as an observation; signal attenuation is mentioned as a fault, but bias or bottom loss is not modeled.
  • domain assumption The UBSL refraction-based measurement model from [46] is accurate under field conditions; the paper does not re-validate it.
    Section III.B and V.B assume the structured light model produces correct 3D points; if calibration drifts or water turbidity changes refraction, the point clouds inherit bias.
  • domain assumption The environment is static and rigid during mapping.
    NDT registration and loop closure assume a fixed scene; field river experiments include currents, plants, and particles that violate this, but no dynamic-object handling is described.
  • domain assumption The magnetometer provides absolute orientation underwater despite magnetic disturbances.
    Section VI.B.4 adds a magnetometer factor without fault detection or validation; underwater magnetic compasses are commonly disturbed.

how reviews work

0 comments
Cite this review

Pith. "Pith review of An Underwater, Fault-Tolerant, Laser-Aided Robotic Multi-Modal Dense SLAM System for Continuous Underwater In-Situ Observation." pith.science (2026). https://pith.science/paper/IHXPPKCK

@misc{pith2026250421826,
  author       = {Pith},
  title        = {Pith review of: An Underwater, Fault-Tolerant, Laser-Aided Robotic Multi-Modal Dense SLAM System for Continuous Underwater In-Situ Observation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/IHXPPKCK}},
  note         = {Machine review of arXiv:2504.21826}
}
read the original abstract

Existing underwater SLAM systems are difficult to work effectively in texture-sparse and geometrically degraded underwater environments, resulting in intermittent tracking and sparse mapping. Therefore, we present Water-DSLAM, a novel laser-aided multi-sensor fusion system that can achieve uninterrupted, fault-tolerant dense SLAM capable of continuous in-situ observation in diverse complex underwater scenarios through three key innovations: Firstly, we develop Water-Scanner, a multi-sensor fusion robotic platform featuring a self-designed Underwater Binocular Structured Light (UBSL) module that enables high-precision 3D perception. Secondly, we propose a fault-tolerant triple-subsystem architecture combining: 1) DP-INS (DVL- and Pressure-aided Inertial Navigation System): fusing inertial measurement unit, doppler velocity log, and pressure sensor based Error-State Kalman Filter (ESKF) to provide high-frequency absolute odometry 2) Water-UBSL: a novel Iterated ESKF (IESKF)-based tight coupling between UBSL and DP-INS to mitigate UBSL's degeneration issues 3) Water-Stereo: a fusion of DP-INS and stereo camera for accurate initialization and tracking. Thirdly, we introduce a multi-modal factor graph back-end that dynamically fuses heterogeneous sensor data. The proposed multi-sensor factor graph maintenance strategy efficiently addresses issues caused by asynchronous sensor frequencies and partial data loss. Experimental results demonstrate Water-DSLAM achieves superior robustness (0.039 m trajectory RMSE and 100\% continuity ratio during partial sensor dropout) and dense mapping (6922.4 points/m^3 in 750 m^3 water volume, approximately 10 times denser than existing methods) in various challenging environments, including pools, dark underwater scenes, 16-meter-deep sinkholes, and field rivers. Our project is available at https://water-scanner.github.io/.

Figures

Figures reproduced from arXiv: 2504.21826 by the authors.

Figure 1
Figure 1. Mechatronic design, prototype, sensors and test scenarios of underwater multi-sensor robotic system, named Water-Scanner. Note: The picture in the [PITH_FULL_IMAGE:figures/full_fig_p004_1.png] view at source ↗
Figure 2
Figure 2. The proposed underwater, fault-tolerant, laser-aided robotic multi-modal dense SLAM framework for continuous underwater in-situ observation, named [PITH_FULL_IMAGE:figures/full_fig_p005_2.png] view at source ↗
Figure 3
Figure 3. The frequency of sensors in DP-INS Subsystem. [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figures from the paper (15 more)
Figure 4
Figure 4. Figure 4: The designed schematic of Water-UBSL Subsystem. [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]
Figure 5
Figure 5. Figure 5: The designed schematic of Water-Stereo Subsystem. [PITH_FULL_IMAGE:figures/full_fig_p008_5.png]
Figure 6
Figure 6. Figure 6: The schematic of multi-modal factor graph in back-end of Water-DSLAM. [PITH_FULL_IMAGE:figures/full_fig_p009_6.png]
Figure 7
Figure 7. Figure 7: Subsystem fault demonstration. (a) Water-Stereo fault scenario. [PITH_FULL_IMAGE:figures/full_fig_p010_7.png]
Figure 8
Figure 8. Figure 8: The maintenance process of multi-modal factor graph. [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: Experimental scenes. (a) Pool. (b) Dark underwater. (c) 16-meter [PITH_FULL_IMAGE:figures/full_fig_p012_9.png]
Figure 10
Figure 10. Figure 10: Experimental results of the ablation study on the DP-INS subsystem. [PITH_FULL_IMAGE:figures/full_fig_p012_10.png]
Figure 11
Figure 11. Figure 11: Experimental results of the ablation study on the Water-Stereo [PITH_FULL_IMAGE:figures/full_fig_p013_11.png]
Figure 12
Figure 12. Figure 12: Visualization of the point cloud matching process during the ablation [PITH_FULL_IMAGE:figures/full_fig_p014_12.png]
Figure 13
Figure 13. Figure 13: Experimental results of the ablation study on different subsystems of Water-DSLAM. (a) Robot movement process. (b) Trajectories at different [PITH_FULL_IMAGE:figures/full_fig_p015_13.png]
Figure 14
Figure 14. Figure 14: Comparative experimental results of state-of-the-art underwater dense mapping methods: self-stabilizing localized fine-Scale underwater inspection. [PITH_FULL_IMAGE:figures/full_fig_p015_14.png]
Figure 15
Figure 15. Figure 15: Comparative experimental results of state-of-the-art multi-modal underwater SLAM methods: random artificial pool dense mapping. (a) Underwater [PITH_FULL_IMAGE:figures/full_fig_p016_15.png]
Figure 16
Figure 16. Figure 16: Application experiments: underwater dark scene in-situ observation. (a)(b) Dark scene. (c) Self-stabilizing mapping as in [45]. (d) Self-stabilizing [PITH_FULL_IMAGE:figures/full_fig_p017_16.png]
Figure 17
Figure 17. Figure 17: Application experiments: underwater large-scale sinkhole in-situ mapping. (a)-(c) 16-meter-deep sinkhole showcase. (d) Water-Stereo subsystem [PITH_FULL_IMAGE:figures/full_fig_p018_17.png]
Figure 18
Figure 18. Figure 18: Application experiments: Field river in-situ observation. (a) Observation site image derived from Google Earth. (b)-(d) Water-Scanner in field. (e) [PITH_FULL_IMAGE:figures/full_fig_p019_18.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

53 extracted references · 47 canonical work pages

  1. [1]

    Design and validation of a robotic control law for observation of deep-ocean jellyfish,

    J. H. Rife and S. M. Rock, “Design and validation of a robotic control law for observation of deep-ocean jellyfish,” IEEE Transactions on Robotics, vol. 22, no. 2, pp. 282–291, 2006

    work page 2006

  2. [2]

    Mobile robotic platforms for the acoustic tracking of deep-sea demersal fishery re- sources,

    I. Masmitja, J. Navarro, S. Gom ´ariz, J. Aguzzi, B. Kieft, T. O’Reilly, K. Katija, P.-J. Bouvet, C. Fannjiang, M. Vigo et al. , “Mobile robotic platforms for the acoustic tracking of deep-sea demersal fishery re- sources,” Science Robotics, vol. 5, no. 48, p. eabc3701, 2020

    work page 2020

  3. [3]

    An in situ digital synthesis strategy for the discovery and description of ocean life,

    J. A. Burns, K. P. Becker, D. Casagrande, J. Daniels, P. Roberts, E. Orenstein, D. M. V ogt, Z. E. Teoh, R. Wood, A. H. Yin et al., “An in situ digital synthesis strategy for the discovery and description of ocean life,” Science Advances, vol. 10, no. 3, p. eadj4960, 2024

    work page 2024

  4. [4]

    Development of bioinspired multimodal underwater robot “hero-blue

    T. Kim, J. Kim, and S.-C. Yu, “Development of bioinspired multimodal underwater robot “hero-blue” for walking, swimming, and crawling,” IEEE Transactions on Robotics , vol. 40, pp. 1421–1438, 2024

    work page 2024

  5. [5]

    Multi-environment robotic transitions through adaptive morphogenesis,

    R. Baines, S. K. Patiballa, J. Booth, L. Ramirez, T. Sipple, A. Garcia, F. Fish, and R. Kramer-Bottiglio, “Multi-environment robotic transitions through adaptive morphogenesis,” Nature, vol. 610, no. 7931, pp. 283– 289, 2022

    work page 2022

  6. [6]

    Real-time visual slam for autonomous underwater hull inspection using visual saliency,

    A. Kim and R. M. Eustice, “Real-time visual slam for autonomous underwater hull inspection using visual saliency,” IEEE Transactions on Robotics, vol. 29, no. 3, pp. 719–733, 2013

    work page 2013

  7. [7]

    Past, present, and future of simultaneous localization and mapping: Toward the robust-perception age,

    C. Cadena, L. Carlone, H. Carrillo, Y . Latif, D. Scaramuzza, J. Neira, I. Reid, and J. J. Leonard, “Past, present, and future of simultaneous localization and mapping: Toward the robust-perception age,” IEEE Transactions on robotics, vol. 32, no. 6, pp. 1309–1332, 2016

    2016

  8. [8]

    A fast in-motion alignment method for ins/dvl in rough sea conditions,

    H. Qin, X. Wang, G. Wang, M. Hu, Y . Bian, X. Qin, and R. Ding, “A fast in-motion alignment method for ins/dvl in rough sea conditions,” IEEE Transactions on Instrumentation and Measurement , 2024

    work page 2024

Show all 53 references
  1. [9]

    Dual-slam: A framework for robust single camera navigation,

    H. Huang, W.-Y . Lin, S. Liu, D. Zhang, and S.-K. Yeung, “Dual-slam: A framework for robust single camera navigation,” in 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) . IEEE, 2020, pp. 4942–4949

    2020

  2. [10]

    Underwater pose slam using gmm scan matching for a mechanical profiling sonar,

    P. Vial, N. Palomeras, J. Sol `a, and M. Carreras, “Underwater pose slam using gmm scan matching for a mechanical profiling sonar,” Journal of Field Robotics, vol. 41, no. 3, pp. 511–538, 2024. 20

    2024

  3. [11]

    Development, calibration, and image processing of underwater structured light vision system: A survey,

    J. Fan, X. Wang, C. Zhou, Y . Ou, F. Jing, and Z. Hou, “Development, calibration, and image processing of underwater structured light vision system: A survey,” IEEE Transactions on Instrumentation and Measure- ment, vol. 72, pp. 1–18, 2023

    2023

  4. [12]

    Three birds, one stone: Unified laser- based 3-d reconstruction across different media,

    C. Gu, Y . Cong, and G. Sun, “Three birds, one stone: Unified laser- based 3-d reconstruction across different media,” IEEE Transactions on Instrumentation and Measurement , vol. 70, pp. 1–12, 2020

    2020

  5. [13]

    Svin2: A multi-sensor fusion-based underwater slam system,

    S. Rahman, A. Quattrini Li, and I. Rekleitis, “Svin2: A multi-sensor fusion-based underwater slam system,” The International Journal of Robotics Research, vol. 41, no. 11-12, pp. 1022–1042, 2022

    2022

  6. [14]

    A fully-automatic side-scan sonar simultaneous localization and mapping framework,

    J. Zhang, Y . Xie, L. Ling, and J. Folkesson, “A fully-automatic side-scan sonar simultaneous localization and mapping framework,” IET Radar, Sonar & Navigation , vol. 18, no. 5, pp. 674–683, 2024

    2024

  7. [15]

    Experi- mental comparison of open source visual-inertial-based state estimation algorithms in the underwater domain,

    B. Joshi, S. Rahman, M. Kalaitzakis, B. Cain, J. Johnson, M. Xanthidis, N. Karapetyan, A. Hernandez, A. Q. Li, N. Vitzilaios et al. , “Experi- mental comparison of open source visual-inertial-based state estimation algorithms in the underwater domain,” in 2019 IEEE/RSJ Interna...

    2019

  8. [16]

    Vins-mono: A robust and versatile monocular visual-inertial state estimator,

    T. Qin, P. Li, and S. Shen, “Vins-mono: A robust and versatile monocular visual-inertial state estimator,” IEEE transactions on robotics , vol. 34, no. 4, pp. 1004–1020, 2018

    2018

  9. [17]

    Detecting loop closure using enhanced image for underwater vins-mono,

    H. Zhao, R. Zheng, M. Liu, and S. Zhang, “Detecting loop closure using enhanced image for underwater vins-mono,” in Global Oceans 2020: Singapore–US Gulf Coast . IEEE, 2020, pp. 1–6

    2020

  10. [18]

    An open-source, fiducial-based, underwater stereo visual-inertial localization method with refraction correction,

    P. Zhang, Z. Wu, J. Wang, S. Kong, M. Tan, and J. Yu, “An open-source, fiducial-based, underwater stereo visual-inertial localization method with refraction correction,” in 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) . IEEE, 2021, pp. 4331–4336

    2021

  11. [19]

    Sm/vio: Robust underwater state estimation switching between model-based and visual inertial odometry,

    B. Joshi, H. Damron, S. Rahman, and I. Rekleitis, “Sm/vio: Robust underwater state estimation switching between model-based and visual inertial odometry,” in 2023 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2023, pp. 5192–5199

    2023

  12. [20]

    Hybrid visual inertial odometry for robust underwater estimation,

    B. Joshi, C. Bandara, I. Poulakakis, H. G. Tanner, and I. Rekleitis, “Hybrid visual inertial odometry for robust underwater estimation,” in OCEANS 2023-MTS/IEEE US Gulf Coast . IEEE, 2023, pp. 1–7

    2023

  13. [21]

    Acoustic-vins: Tightly coupled acoustic- visual-inertial navigation system for autonomous underwater vehicles,

    J. Song, W. Li, and X. Zhu, “Acoustic-vins: Tightly coupled acoustic- visual-inertial navigation system for autonomous underwater vehicles,” IEEE Robotics and Automation Letters , 2023

    2023

  14. [22]

    Robust underwater visual slam fusing acoustic sensing,

    E. Vargas, R. Scona, J. S. Willners, T. Luczynski, Y . Cao, S. Wang, and Y . R. Petillot, “Robust underwater visual slam fusing acoustic sensing,” in 2021 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2021, pp. 2140–2146

    2021

  15. [23]

    Tightly- coupled visual-dvl fusion for accurate localization of underwater robots,

    Y . Huang, P. Li, S. Yan, Y . Ou, Z. Wu, M. Tan, and J. Yu, “Tightly- coupled visual-dvl fusion for accurate localization of underwater robots,” in 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2023, pp. 8090–8095

    2023

  16. [24]

    Visual- inertial-acoustic sensor fusion for accurate autonomous localization of underwater vehicles,

    Y . Huang, P. Li, S. Ma, S. Yan, M. Tan, J. Yu, and Z. Wu, “Visual- inertial-acoustic sensor fusion for accurate autonomous localization of underwater vehicles,” IEEE Transactions on Cybernetics , 2024

    2024

  17. [25]

    Aqua-slam: Tightly-coupled underwater acoustic-visual-inertial slam with sensor calibration,

    S. Xu, K. Zhang, and S. Wang, “Aqua-slam: Tightly-coupled underwater acoustic-visual-inertial slam with sensor calibration,” IEEE Transactions on Robotics, 2025

    2025

  18. [26]

    An integrated visual odometry system for underwater vehicles,

    Z. Xu, M. Haroutunian, A. J. Murphy, J. Neasham, and R. Norman, “An integrated visual odometry system for underwater vehicles,” IEEE Journal of Oceanic Engineering , vol. 46, no. 3, pp. 848–863, 2020

    2020

  19. [27]

    Sonar visual inertial slam of underwater structures,

    S. Rahman, A. Q. Li, and I. Rekleitis, “Sonar visual inertial slam of underwater structures,” in 2018 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2018, pp. 5190–5196

    2018

  20. [28]

    Svin2: An underwater slam system using sonar, visual, inertial, and depth sensor,

    ——, “Svin2: An underwater slam system using sonar, visual, inertial, and depth sensor,” in 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) . IEEE, 2019, pp. 1861–1868

    2019

  21. [29]

    Absolute scale estimation for underwater monocular visual odometry based on 2-d imaging sonar,

    D. Yang, H. Ai, J. Liu, and B. He, “Absolute scale estimation for underwater monocular visual odometry based on 2-d imaging sonar,” Measurement, vol. 190, p. 110665, 2022

    2022

  22. [30]

    Camera-sonar combination for im- proved underwater localization and mapping,

    A. Cardaillac and M. Ludvigsen, “Camera-sonar combination for im- proved underwater localization and mapping,” IEEE Access, 2023

    2023

  23. [31]

    An open-source bio-inspired solution to underwater slam,

    L. Silveira, F. Guth, P. Drews-Jr, P. Ballester, M. Machado, F. Codevilla, N. Duarte-Filho, and S. Botelho, “An open-source bio-inspired solution to underwater slam,” IFAC-PapersOnLine, vol. 48, no. 2, pp. 212–217, 2015

    2015

  24. [32]

    Underwater caves sonar data set,

    A. Mallios, E. Vidal, R. Campos, and M. Carreras, “Underwater caves sonar data set,” The International Journal of Robotics Research, vol. 36, no. 12, pp. 1247–1251, 2017

    2017

  25. [33]

    Real-time underwater 3d reconstruction using global context and active labeling,

    R. DeBortoli, A. Nicolai, F. Li, and G. A. Hollinger, “Real-time underwater 3d reconstruction using global context and active labeling,” in 2018 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2018, pp. 6204–6211

    2018

  26. [34]

    Wide aperture imaging sonar reconstruction using generative models,

    E. Westman and M. Kaess, “Wide aperture imaging sonar reconstruction using generative models,” in 2019 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2019, pp. 8067–8074

    2019

  27. [35]

    Autonomous 3-d mapping and safe-path planning for underwater terrain reconstruction using multi- level coverage trees,

    Z. Shen, J. Song, K. Mittal, and S. Gupta, “Autonomous 3-d mapping and safe-path planning for underwater terrain reconstruction using multi- level coverage trees,” in OCEANS 2017-Anchorage. IEEE, 2017, pp. 1–6

    2017

  28. [36]

    Ct-cpp: Coverage path planning for 3d terrain reconstruction using dynamic coverage trees,

    ——, “Ct-cpp: Coverage path planning for 3d terrain reconstruction using dynamic coverage trees,” IEEE Robotics and Automation Letters , vol. 7, no. 1, pp. 135–142, 2021

    2021

  29. [37]

    Unified underwater structure-from-motion,

    K. Ichimaru, Y . Taguchi, and H. Kawasaki, “Unified underwater structure-from-motion,” in 2019 International Conference on 3D Vision (3DV). IEEE, 2019, pp. 524–532

    2019

  30. [38]

    Hy- perspectral 3d mapping of underwater environments,

    M. Ferrera, A. Arnaubec, K. Isteni ˇc, N. Gracias, and T. Bajjouk, “Hy- perspectral 3d mapping of underwater environments,” in Proceedings of the IEEE/CVF International Conference on Computer Vision , 2021, pp. 3703–3712

    2021

  31. [39]

    Robust underwater slam using autonomous relocalisation,

    J. S. Willners, Y . Carreno, S. Xu, T. Łuczy ´nski, S. Katagiri, J. Roe, `E. Pairet, Y . Petillot, and S. Wang, “Robust underwater slam using autonomous relocalisation,” IFAC-PapersOnLine, vol. 54, no. 16, pp. 273–280, 2021

    2021

  32. [40]

    High-resolution underwater robotic vision-based mapping and three-dimensional reconstruction for archae- ology,

    M. Johnson-Roberson, M. Bryson, A. Friedman, O. Pizarro, G. Troni, P. Ozog, and J. C. Henderson, “High-resolution underwater robotic vision-based mapping and three-dimensional reconstruction for archae- ology,” Journal of Field Robotics , vol. 34, no. 4, pp. 625–643, 2017

    2017

  33. [41]

    Application of structured light imaging for high resolution mapping of underwater archaeological sites,

    C. Roman, G. Inglis, and J. Rutter, “Application of structured light imaging for high resolution mapping of underwater archaeological sites,” in OCEANS’10 IEEE SYDNEY . IEEE, 2010, pp. 1–9

    2010

  34. [42]

    Generation of high-resolution three-dimensional reconstructions of the seafloor in color using a single camera and structured light,

    A. Bodenmann, B. Thornton, and T. Ura, “Generation of high-resolution three-dimensional reconstructions of the seafloor in color using a single camera and structured light,” Journal of Field Robotics , vol. 34, no. 5, pp. 833–851, 2017

    2017

  35. [43]

    Underwater 3d scanning system for cultural heritage documentation,

    C. Br ¨auer-Burchardt, C. Munkelt, M. Bleier, M. Heinze, I. Gebhart, P. K ¨uhmstedt, and G. Notni, “Underwater 3d scanning system for cultural heritage documentation,” Remote Sensing , vol. 15, no. 7, p. 1864, 2023

    2023

  36. [44]

    Improving self-consistency in underwa- ter mapping through laser-based loop closure,

    T. Hitchcox and J. R. Forbes, “Improving self-consistency in underwa- ter mapping through laser-based loop closure,” IEEE Transactions on Robotics, vol. 39, no. 3, pp. 1873–1892, 2023

    2023

  37. [45]

    Inspection of an underwater structure using point-cloud slam with an auv and a laser scanner,

    A. Palomer, P. Ridao, and D. Ribas, “Inspection of an underwater structure using point-cloud slam with an auv and a laser scanner,” Journal of field robotics , vol. 36, no. 8, pp. 1333–1344, 2019

    2019

  38. [46]

    Binocular structured light 3-d reconstruction system for low-light underwater environments: Design, modeling, and laser-based calibration,

    Y . Ou, J. Fan, C. Zhou, S. Tian, L. Cheng, and M. Tan, “Binocular structured light 3-d reconstruction system for low-light underwater environments: Design, modeling, and laser-based calibration,” IEEE Transactions on Instrumentation and Measurement , vol. 72, pp. 1–14, 2023

    2023

  39. [47]

    Water-mbsl: Un- derwater movable binocular structured light-based high-precision dense reconstruction framework,

    Y . Ou, J. Fan, C. Zhou, L. Cheng, and M. Tan, “Water-mbsl: Un- derwater movable binocular structured light-based high-precision dense reconstruction framework,”IEEE Transactions on Industrial Informatics, vol. 20, no. 4, pp. 6142–6154, 2023

    2023

  40. [48]

    Fast-lio: A fast, robust lidar-inertial odometry package by tightly-coupled iterated kalman filter,

    W. Xu and F. Zhang, “Fast-lio: A fast, robust lidar-inertial odometry package by tightly-coupled iterated kalman filter,” IEEE Robotics and Automation Letters, vol. 6, no. 2, pp. 3317–3324, 2021

    2021

  41. [49]

    A general optimization-based framework for global pose estimation with multiple sensors,

    T. Qin, S. Cao, J. Pan, and S. Shen, “A general optimization-based framework for global pose estimation with multiple sensors,” arXiv preprint arXiv:1901.03642, 2019

    1901 arXiv

  42. [50]

    Hybrid-vins: Underwater tightly coupled hybrid visual inertial dense slam for auv,

    Y . Ou, J. Fan, C. Zhou, P. Zhang, and Z.-G. Hou, “Hybrid-vins: Underwater tightly coupled hybrid visual inertial dense slam for auv,” IEEE Transactions on Industrial Electronics , 2024

    2024

  43. [51]

    Evo: A geometric approach to event-based 6-dof parallel tracking and mapping in real time,

    H. Rebecq, T. Horstsch ¨afer, G. Gallego, and D. Scaramuzza, “Evo: A geometric approach to event-based 6-dof parallel tracking and mapping in real time,” IEEE Robotics and Automation Letters , vol. 2, no. 2, pp. 593–600, 2016

    2016

  44. [52]

    Structured light-based underwater collision-free navigation and dense mapping system for refined exploration in unknown dark environments,

    Y . Ou, J. Fan, C. Zhou, S. Kang, Z. Zhang, Z.-G. Hou, and M. Tan, “Structured light-based underwater collision-free navigation and dense mapping system for refined exploration in unknown dark environments,” IEEE Transactions on Systems, Man, and Cybernetics: Systems , 2024

    2024

  45. [53]

    Extrinsic visual– inertial calibration for motion distortion correction of underwater 3d scans,

    M. Castill ´on, R. Pi, N. Palomeras, and P. Ridao, “Extrinsic visual– inertial calibration for motion distortion correction of underwater 3d scans,” IEEE Access, vol. 9, pp. 93 384–93 398, 2021

    2021

Pith tools

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