Pith. sign in

REVIEW 4 major objections 5 minor 38 references

Do You Know the Way? Human-in-the-Loop Understanding for Fast Traversability Estimation in Mobile Robotics

T0 review · 4 major / 5 minor · reviewed 2026-08-16 · deepseek-v4-flash

Pith's one-line read A deployed robot can learn where it can drive from a handful of seconds-long human labels, the paper claims.

desk verdict Useful HiL traversability system with solid simulator results; the as-needed novelty gating is under-validated and the SOTA claim outruns the evidence. read the letter →

arxiv 2504.19851 v1 pith:3TVNRQXO submitted 2025-04-28 cs.RO

classification cs.RO
keywords traversabilityestimationhuman-in-the-loopnoveltydetectionDINOv2FeatUponlinelearningmobilerobotnavigationsim-to-realtransfer
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

The paper proposes a human-in-the-loop system, CHUNGUS, that lets a mobile robot learn where it can drive while it is deployed, using only a small number of sparse annotations supplied by a person in seconds. The central claim is that an online-trained traversability head built on frozen DINOv2 and FeatUp features can reach 70 to 100 percent navigation success across simulated forest, warehouse, parking-lot, and dark-lot environments with only about 5 to 26 annotated images per run, and that this matches or beats a self-supervised online baseline that needs a five-minute teleoperation warm-up. The system asks for a human label only when a Faiss nearest-neighbor check on the DINOv2 class token flags an image as novel, which keeps the human effort small. If the claim holds, field robots could adapt to unfamiliar terrain during a mission without a separate training or teleoperation phase.

What carries the argument

The load-bearing object is the novelty detector: DINOv2's 384-dimensional class token is stored in a Faiss index, a new image is novel when its L2 distance to the nearest stored token is at least $\tau_{\text{novelty}} = \mu_{\text{class}} + \alpha \sigma_{\text{class}}$, and only then is a human asked for labels. The second piece is the traversability head, an MLP trained with the LRIZZ pairwise relative loss plus a reconstruction-MSE branch that scores per-pixel uncertainty; the uncertainty score selects which pixels to compare in the new image. Retraining from scratch on the full cached label set lets the system continually adapt without catastrophic forgetting.

What would settle it

Run CHUNGUS in one fixed environment while the lighting fades gradually from day to night and record, frame by frame, the class-token distances and the traversability prediction error against ground truth; if the novelty threshold fails to fire before prediction error rises substantially, the as-needed annotation schedule is not tracking what the paper claims it tracks.

Watch

Extended reading notes

Core claim

CHUNGUS's discovery is that pairwise relative traversability labels, collected only when a cheap image-level novelty detector fires, are sufficient to train a pixel-wise traversability predictor online. The predictor is a small MLP head on frozen DINOv2 patch features upsampled by FeatUp; training from scratch on all cached labels avoids catastrophic forgetting and takes about 0.1 seconds per retraining. In ten-trial simulated navigation experiments, the method succeeded in 70 to 100 percent of runs depending on the novelty threshold, with median labeling time adding roughly 18 to 69 seconds per successful run. On real-world data, a model trained only in the simulator produced traversability maps qualitatively close to those of a model trained on real images, with only a modest rise in human disagreement rate.

Load-bearing premise

The load-bearing premise is that the L2 distance from a new image's DINOv2 class token to the nearest previously seen token is a reliable signal that the traversability head needs new labels; if novelty and traversability error diverge, the annotation schedule stops protecting the robot.

Editorial extensions

If this is right

  • A robot can start a deployment with an empty training set and still navigate a new environment, because the first annotations come from the novelty detector rather than from a teleoperation phase.
  • Human cost is a few seconds per label: about 1.5 seconds per smart label and roughly 5 to 26 annotated images per successful run, far below the five-minute teleoperation warm-up used by the baseline.
  • The method outperforms WVN without teleoperation and performs comparably to WVN with five minutes of teleoperation, while also running inference at 7.7 Hz versus 5.7 Hz.
  • A model trained purely in the high-fidelity simulator transfers to real images with only modest degradation, suggesting simulation can supply pre-training data for the traversability head.

Reading between the lines

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

  • The novelty threshold is defined only when the Faiss index holds more than one token; a principled cold-start rule for the first images would make the method fully deployable without hand-tuning.
  • Because the detector uses only the class token, the same query-and-retrain loop could be attached to other dense prediction heads, such as affordance or risk maps, by swapping the MLP head and keeping the novelty trigger.
  • A natural stress test is to expose the robot to gradual domain shift, such as dusk lighting or camera exposure changes within one environment; if class-token distance does not track prediction error there, the annotation schedule would need a fallback trigger.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 5 minor

Summary. The paper proposes CHUNGUS, an online human-in-the-loop traversability estimation system. A frozen DINOv2+FeatUp feature extractor feeds a small MLP head that is trained from scratch on sparse pairwise relative traversability labels. A Faiss-based novelty detector, operating on the DINOv2 class token with an L2 distance and a threshold recomputed from the current index, decides when to pause the robot and request a human annotation. The system is evaluated in four photorealistic simulated environments for navigation success against Blind, LiDAR, WVN, WVN with 5 min teleoperation, and an offline Big CHUNGUS variant. Additional experiments include an ablation over annotation pixel selection and the novelty threshold alpha, and a sim-to-real transfer evaluation of Big CHUNGUS using human disagreement rate.

Significance. If the claims hold, the main value of the paper is a practical system that learns traversability during deployment without teleoperation, using only a handful of quickly collected human labels. The paper ships code, uses 10 trials per navigation condition, and demonstrates that a model trained solely in the proposed simulator transfers to real imagery with only a modest increase in disagreement rate. These are tangible strengths. However, the headline claim of state-of-the-art traversability prediction is not supported by the experiments, and the specific contribution of the novelty detector in the actual small-sample online regime is not isolated by any ablation. The central HiL idea is defensible, but the manuscript overreaches in its framing and leaves a load-bearing component under-validated.

major comments (4)
  1. [Abstract and Section IV] The abstract and contribution list claim 'state-of-the-art traversability prediction performance,' but the evaluation does not compare traversability prediction quality against any other traversability prediction method. Table I compares navigation success against Blind, LiDAR, WVN, and Big CHUNGUS, while Table IV reports human disagreement rate only for Big CHUNGUS variants. No comparison to W-RIZZ, V-STRONG, WayFAST, or similar methods is provided, and no common benchmark metric for traversability prediction is reported. The state-of-the-art claim should be removed or substantiated with a direct prediction-quality comparison.
  2. [Sections III-B, IV-B, and Table III] The 'as-needed' annotation claim rests on the novelty detector described in Section III-B. The detector is validated only as an offline four-environment classifier with roughly 1000 training tokens per environment in Section IV-B and Table II. In deployment, the Faiss index contains about 5-26 tokens and the threshold is recomputed after each annotation; no evidence shows that L2 class-token distance in this small-sample online regime tracks the traversability head's prediction error. Moreover, Table III compares pixel-selection strategies (Random vs Smart) while holding the Faiss-driven image-selection schedule fixed; there is no ablation with a fixed or random image-annotation schedule matched for label count. Without such an ablation, the navigation results do not isolate the contribution of the novelty detector to the reported success rates.
  3. [Sections III-B and IV-A] The cold-start behavior of the system is unspecified. Section III-B defines the novelty threshold only for a Faiss index containing more than one class token, yet Section IV-A states that HiL CHUNGUS starts each run with an empty training dataset. The manuscript does not state how the first image is selected for annotation, what threshold is used when the index has zero or one token, or how the first cross-image label is obtained when the set of labeled images is empty. These details affect every navigation trial and must be stated for the method to be reproducible.
  4. [Section IV-A, Table I] The claim that HiL CHUNGUS 'significantly' outperforms WVN is based on 10 trials per condition, and no confidence intervals or statistical tests are reported. Large gaps such as 90% vs 10% are suggestive, but comparisons such as WVN (5 min) at 80% vs CHUNGUS (alpha=1) at 100% in the warehouse are within the resolution of 10 trials. Reporting exact binomial confidence intervals or increasing the number of trials would make the comparison more robust.
minor comments (5)
  1. [Section III-A, Eq. (1)] The reconstruction-loss weight gamma is fixed at 0.1, but no sensitivity analysis or justification is provided. Since gamma is a free parameter in the loss, a brief ablation or a note on its role would improve reproducibility.
  2. [Section IV-D] The real-world evaluation uses only the offline Big CHUNGUS model; the online HiL procedure is not demonstrated on physical hardware. The text should clarify that the real-world data support the sim-to-real transfer claim but not the online HiL loop itself.
  3. [Section IV-C, Table III] The 'Manual' condition reports a per-label time of about 3.3 s versus about 1.5 s for the other conditions, but the total navigation time includes annotation pauses. Separating annotation time from navigation time in Table III would make the comparison of labeling efficiency clearer.
  4. [Section III-B] There is a typo in the phrase 'classifed as novel'; it should read 'classified as novel'.
  5. [Table III] The column header 'Map Labeling Method' is ambiguous because the rows mix annotation pixel-selection strategies with the novelty hyperparameter alpha. Renaming the columns or splitting the table would improve readability.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the claimed predictions are supported by independent evaluations and the same-author W-RIZZ citation is a published prior loss, not a forced premise.

full rationale

The paper's derivation chain does not reduce its outputs to its inputs. The traversability head is trained with L_CHUNGUS = L_RIZZ + 0.1*(MSE of reconstructed DINOv2 features), but the reconstruction term is an auxiliary objective, not a relabeling of the traversability target; the pairwise ordinal labels come from the human annotator, so the predictions are not recomputed from a fitted parameter renamed as a prediction. The novelty detector is a separate component: a Faiss L2 nearest-neighbor test on DINOv2 class tokens, with threshold tau = mu_class + alpha*sigma_class computed from unlabeled within-index distances. It is validated in Table II on held-out images in a four-environment anomaly detection task, which is independent of the navigation success metric; alpha is an explored hyperparameter rather than a value fit to the headline results. The only same-author dependency is W-RIZZ [15], which supplies the L_RIZZ loss and pairwise labeling interface; that is a published prior result, and the paper's contributions are evaluated against external baselines (WVN, LiDAR, Blind) rather than by citing the prior work as an unverified uniqueness theorem. The identified weaknesses - undefined cold-start threshold for the empty Faiss index, lack of validation in the small-index online regime, and no fixed-schedule ablation for the novelty detector - are empirical gaps that could weaken the as-needed labeling claim, but they are not instances of a prediction being equivalent to its input by construction. Therefore no circularity is established.

Assumptions & free parameters 2 free parameters · 7 assumptions · 0 invented entities

The main burden falls on domain assumptions about feature transferability and novelty proxies, rather than on fitted constants. The two free hyperparameters, gamma and alpha, are small in number and reported transparently; they are not fit to hide the result.

free parameters (2)
  • gamma (reconstruction loss weight) = 0.1
    Used in Eq. (1) to weight the reconstruction MSE branch; fixed by hand for all experiments, not fitted.
  • alpha (novelty threshold multiplier) = 1, 2, and 3 (explored); 2 used for AD eval, 1 and 2 for navigation, 3 in ablation
    Sets tau_novelty = mu_class + alpha sigma_class in Sec. III-B. Treated as a user-chosen aggressiveness knob; results are reported for multiple values, so it is not hidden, but it is a free hyperparameter of the method.
assumptions (7)
  • domain assumption DINOv2 pretrained features are informative and transferable for traversability.
    Invoked in Sec. III-A, where DINOv2 and FeatUp features are treated as a frozen backbone that supports linear and MLP traversability heads across sim and real domains.
  • domain assumption FeatUp upsampling preserves edges better than bilinear interpolation.
    Sec. III-A relies on FeatUp to produce pixel-level features that retain critical details like edges; this is a design assumption from the cited FeatUp paper, not re-derived here.
  • domain assumption The LRIZZ pairwise relative loss from W-RIZZ is a valid training signal for sparse traversability labels.
    Sec. III-A adopts LRIZZ and the sparse labeling strategy from Schreiber et al. [15]; the current paper does not justify the loss beyond citing prior work.
  • ad hoc to paper Class-token nearest-neighbor distance is a proxy for when the traversability network needs new supervision.
    Sec. III-B defines the entire annotation trigger on this assumption; validated only on four discrete environment shifts, not on gradual domain drift.
  • ad hoc to paper Reconstruction MSE of the upsampled DINOv2 feature is a proxy for pixel-level novelty.
    Sec. III-C uses the reconstruction error branch to choose which pixels the human labels; the paper states this was found to improve performance but gives no independent justification.
  • domain assumption The custom Unreal Engine simulator has sufficient visual fidelity for sim-to-real transfer of traversability predictions.
    Sec. III-E and Sec. IV-D assume that simulator images are realistic enough that models trained on them transfer; only HDR on one real dataset is reported, with no robotic deployment.
  • domain assumption A small MLP head can be trained from scratch on a few pairwise labels to produce dense traversability predictions.
    Sec. III-A and Sec. IV-C assume the frozen feature representation is strong enough that the sparse online dataset, roughly 5 to 26 images per run, is sufficient for dense prediction; the ablation supports this in simulation but it remains an assumption about the feature space.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Do You Know the Way? Human-in-the-Loop Understanding for Fast Traversability Estimation in Mobile Robotics." pith.science (2026). https://pith.science/paper/3TVNRQXO

@misc{pith2026250419851,
  author       = {Pith},
  title        = {Pith review of: Do You Know the Way? Human-in-the-Loop Understanding for Fast Traversability Estimation in Mobile Robotics},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/3TVNRQXO}},
  note         = {Machine review of arXiv:2504.19851}
}
read the original abstract

The increasing use of robots in unstructured environments necessitates the development of effective perception and navigation strategies to enable field robots to successfully perform their tasks. In particular, it is key for such robots to understand where in their environment they can and cannot travel -- a task known as traversability estimation. However, existing geometric approaches to traversability estimation may fail to capture nuanced representations of traversability, whereas vision-based approaches typically either involve manually annotating a large number of images or require robot experience. In addition, existing methods can struggle to address domain shifts as they typically do not learn during deployment. To this end, we propose a human-in-the-loop (HiL) method for traversability estimation that prompts a human for annotations as-needed. Our method uses a foundation model to enable rapid learning on new annotations and to provide accurate predictions even when trained on a small number of quickly-provided HiL annotations. We extensively validate our method in simulation and on real-world data, and demonstrate that it can provide state-of-the-art traversability prediction performance.

Figures

Figures reproduced from arXiv: 2504.19851 by the authors.

Figure 1
Figure 1. Our proposed human-in-the-loop (HiL) traversability prediction [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of our proposed human-in-the-loop traversability prediction method. RGB camera images are passed through a DINOv2 [25] feature extractor [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Images from our simulator, showing the forest (top left), warehouse [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figures from the paper (1 more)
Figure 4
Figure 4. Figure 4: Traversability predictions using Big CHUNGUS, showing color [PITH_FULL_IMAGE:figures/full_fig_p007_4.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

38 extracted references · 31 canonical work pages

  1. [1]

    Embedded high precision control and corn stand counting algorithms for an ultra-compact 3d printed field robot,

    E. Kayacan, Z. Zhang, and G. Chowdhary, “Embedded high precision control and corn stand counting algorithms for an ultra-compact 3d printed field robot,” in Robotics: Science and Systems , 2018

  2. [2]

    CERBERUS in the DARPA subterranean challenge,

    M. Tranzatto, T. Miki, M. Dharmadhikari, L. Bernreiter, M. Kulkarni, F. Mascarich, O. Andersson, S. Khattak, M. Hutter, R. Siegwart, and K. Alexis, “CERBERUS in the DARPA subterranean challenge,”Science Robotics, vol. 7, no. 66, p. eabp9742, 2022

  3. [3]

    Terrain traversability analysis methods for unmanned ground vehicles: A survey,

    P. Papadakis, “Terrain traversability analysis methods for unmanned ground vehicles: A survey,” Engineering Applications of Artificial In- telligence, vol. 26, no. 4, pp. 1373–1385, 2013

  4. [4]

    High resolution maps from wide angle sonar,

    H. Moravec and A. Elfes, “High resolution maps from wide angle sonar,” in IEEE International Conference on Robotics and Automation , 1985, pp. 116–121

  5. [5]

    Thrun, W

    S. Thrun, W. Burgard, and D. Fox, Probabilistic Robotics. MIT Press, Aug. 2005

  6. [6]

    Siegwart, I

    R. Siegwart, I. R. Nourbakhsh, and D. Scaramuzza, Introduction to Autonomous Mobile Robots , 2nd ed. MIT Press, Feb. 2011

  7. [7]

    Real-time semantic mapping for autonomous off-road navigation,

    D. Maturana, P.-W. Chou, M. Uenoyama, and S. Scherer, “Real-time semantic mapping for autonomous off-road navigation,” in International Conference on Field and Service Robotics , 2017, pp. 335–350

  8. [8]

    Where should I walk? Predicting terrain properties from images via self-supervised learning,

    L. Wellhausen, A. Dosovitskiy, R. Ranftl, K. Walas, C. Cadena, and M. Hutter, “Where should I walk? Predicting terrain properties from images via self-supervised learning,” IEEE Robotics and Automation Letters, vol. 4, no. 2, pp. 1509–1516, 2019

Show all 38 references
  1. [9]

    A RUGD dataset for autonomous navigation and visual perception in unstructured outdoor environments,

    M. Wigness, S. Eum, J. G. Rogers, D. Han, and H. Kwon, “A RUGD dataset for autonomous navigation and visual perception in unstructured outdoor environments,” in IEEE/RSJ International Conference on Intel- ligent Robots and Systems , 2019, pp. 5000–5007

  2. [10]

    BADGR: An autonomous self- supervised learning-based navigation system,

    G. Kahn, P. Abbeel, and S. Levine, “BADGR: An autonomous self- supervised learning-based navigation system,” IEEE Robotics and Au- tomation Letters, vol. 6, no. 2, pp. 1312–1319, 2021

  3. [11]

    WayFAST: Navigation with predictive traversability in the field,

    M. V . Gasparino, A. N. Sivakumar, Y . Liu, A. E. B. Velasquez, V . A. H. Higuti, J. Rogers, H. Tran, and G. Chowdhary, “WayFAST: Navigation with predictive traversability in the field,” IEEE Robotics and Automation Letters , vol. 7, no. 4, pp. 10 651–10 658, 2022

  4. [12]

    Self-supervised traversability prediction by learning to reconstruct safe terrain,

    R. Schmid, D. Atha, F. Sch ¨oller, S. Dey, S. Fakoorian, K. Otsu, B. Ridge, M. Bjelonic, L. Wellhausen, M. Hutter et al., “Self-supervised traversability prediction by learning to reconstruct safe terrain,” in IEEE/RSJ International Conference on Intelligent Robots and Systems...

  5. [13]

    How does it feel? Self-supervised costmap learning for off-road vehicle traversability,

    M. G. Castro, S. Triest, W. Wang, J. M. Gregory, F. Sanchez, J. G. Rogers, and S. Scherer, “How does it feel? Self-supervised costmap learning for off-road vehicle traversability,” in IEEE International Con- ference on Robotics and Automation , 2023, pp. 931–938

  6. [14]

    Weakly-supervised traversability estimation for mobile robots using sparse point annotation,

    A. M. Schreiber, “Weakly-supervised traversability estimation for mobile robots using sparse point annotation,” Master’s thesis, University of Illinois at Urbana-Champaign, 2023

  7. [15]

    W-RIZZ: A weakly-supervised framework for relative traversability estimation in mobile robotics,

    A. Schreiber, A. N. Sivakumar, P. Du, M. V . Gasparino, G. Chowdhary, and K. Driggs-Campbell, “W-RIZZ: A weakly-supervised framework for relative traversability estimation in mobile robotics,” IEEE Robotics and Automation Letters , vol. 9, no. 6, pp. 5623–5630, 2024

  8. [16]

    Fast Traversability Estimation for Wild Visual Navigation,

    J. Frey, M. Mattamala, N. Chebrolu, C. Cadena, M. Fallon, and M. Hutter, “Fast Traversability Estimation for Wild Visual Navigation,” in Robotics: Science and Systems , 2023

  9. [17]

    Wild Visual Navigation: Fast Traversabil- ity Learning via Pre-Trained Models and Online Self-Supervision,

    M. Mattamala, J. Frey, L. Piotr, N. Chebrolu, C. Cadena, G. Martius, M. Hutter, and M. Fallon, “Wild Visual Navigation: Fast Traversabil- ity Learning via Pre-Trained Models and Online Self-Supervision,” arXiv:2404.07110, 2024

  10. [18]

    V-STRONG: Visual self-supervised traversability learning for off-road navigation,

    S. Jung, J. Lee, X. Meng, B. Boots, and A. Lambert, “V-STRONG: Visual self-supervised traversability learning for off-road navigation,” in IEEE International Conference on Robotics and Automation , 2024, pp. 1766–1773

  11. [19]

    Velociraptor: Leveraging visual foundation models for label-free, risk- aware off-road navigation,

    S. Triest, M. Sivaprakasam, S. Aich, D. Fan, W. Wang, and S. Scherer, “Velociraptor: Leveraging visual foundation models for label-free, risk- aware off-road navigation,” in Conference on Robot Learning , 2024

  12. [20]

    Traversability-aware legged naviga- tion by learning from real-world visual data,

    H. Zhang, Z. Li, X. Zeng, L. Smith, K. Stachowicz, D. Shah, L. Yue, Z. Song, W. Xia, S. Levine et al., “Traversability-aware legged naviga- tion by learning from real-world visual data,” arXiv:2410.10621, 2024

  13. [21]

    WayFASTER: a self-supervised traversability prediction for increased navigation aware- ness,

    M. V . Gasparino, A. N. Sivakumar, and G. Chowdhary, “WayFASTER: a self-supervised traversability prediction for increased navigation aware- ness,” in IEEE International Conference on Robotics and Automation , 2024, pp. 8486–8492

  14. [22]

    IMOST: In- cremental memory mechanism with online self-supervision for continual traversability learning,

    K. Ma, Z. Sun, C. Xiong, Q. Zhu, K. Wang, and L. Pei, “IMOST: In- cremental memory mechanism with online self-supervision for continual traversability learning,” arXiv:2409.14070, 2024

  15. [23]

    RELLIS-3D dataset: Data, benchmarks and analysis,

    P. Jiang, P. Osteen, M. Wigness, and S. Saripalli, “RELLIS-3D dataset: Data, benchmarks and analysis,” arXiv:2011.12954, 2020

  16. [24]

    Adaptive robot traversability estimation based on self-supervised online continual learning in unstructured environments,

    H.-S. Yoon, J.-H. Hwang, C. Kim, E. I. Son, S.-W. Yoo, and S.-W. Seo, “Adaptive robot traversability estimation based on self-supervised online continual learning in unstructured environments,” IEEE Robotics and Automation Letters , vol. 9, no. 6, pp. 4902–4909, 2024

  17. [25]

    DINOv2: Learning robust visual features without supervision,

    M. Oquab, T. Darcet, T. Moutakanni, H. V . V o, M. Szafraniec, V . Khali- dov, P. Fernandez, D. Haziza, F. Massa, A. El-Nouby, R. Howes, P.-Y . Huang, H. Xu, V . Sharma, S.-W. Li, W. Galuba, M. Rabbat, M. Assran, N. Ballas, G. Synnaeve, I. Misra, H. Jegou, J. Mairal, P. Labatu...

  18. [26]

    FeatUp: A model-agnostic framework for features at any resolution,

    S. Fu, M. Hamilton, L. E. Brandt, A. Feldmann, Z. Zhang, and W. T. Freeman, “FeatUp: A model-agnostic framework for features at any resolution,” in International Conference on Learning Representations , 2024

  19. [27]

    Emerging properties in self-supervised vision transformers,

    M. Caron, H. Touvron, I. Misra, H. J ´egou, J. Mairal, P. Bojanowski, and A. Joulin, “Emerging properties in self-supervised vision transformers,” in IEEE/CVF International Conference on Computer Vision , 2021, pp. 9630–9640

  20. [28]

    Segment anything,

    A. Kirillov, E. Mintun, N. Ravi, H. Mao, C. Rolland, L. Gustafson, T. Xiao, S. Whitehead, A. C. Berg, W.-Y . Loet al., “Segment anything,” in IEEE/CVF International Conference on Computer Vision , 2023, pp. 3992–4003

  21. [29]

    Learning off-road terrain traversability with self-supervisions only,

    J. Seo, S. Sim, and I. Shim, “Learning off-road terrain traversability with self-supervisions only,” IEEE Robotics and Automation Letters , vol. 8, no. 8, pp. 4617–4624, 2023

  22. [30]

    An empirical investigation of catastrophic forgetting in gradient-based neural networks,

    I. J. Goodfellow, M. Mirza, D. Xiao, A. Courville, and Y . Bengio, “An empirical investigation of catastrophic forgetting in gradient-based neural networks,” arXiv:1312.6211, 2013

  23. [31]

    Safe robot navigation via multi-modal anomaly detection,

    L. Wellhausen, R. Ranftl, and M. Hutter, “Safe robot navigation via multi-modal anomaly detection,” IEEE Robotics and Automation Letters, vol. 5, no. 2, pp. 1326–1333, 2020

  24. [32]

    Proactive anomaly detection for robot navigation with multi-sensor fusion,

    T. Ji, A. N. Sivakumar, G. Chowdhary, and K. Driggs-Campbell, “Proactive anomaly detection for robot navigation with multi-sensor fusion,” IEEE Robotics and Automation Letters , vol. 7, no. 2, pp. 4975– 4982, 2022

  25. [33]

    An attentional recurrent neural network for occlusion-aware proactive anomaly detection in field robot navigation,

    A. Schreiber, T. Ji, D. L. McPherson, and K. Driggs-Campbell, “An attentional recurrent neural network for occlusion-aware proactive anomaly detection in field robot navigation,” in IEEE/RSJ International Conference on Intelligent Robots and Systems , 2023, pp. 8038–8045

  26. [34]

    The Faiss library,

    M. Douze, A. Guzhva, C. Deng, J. Johnson, G. Szilvasy, P.-E. Mazar´e, M. Lomeli, L. Hosseini, and H. J ´egou, “The Faiss library,” arXiv:2401.08281, 2024

  27. [35]

    Elevation mapping for locomotion and navigation using GPU,

    T. Miki, L. Wellhausen, R. Grandia, F. Jenelten, T. Homberger, and M. Hutter, “Elevation mapping for locomotion and navigation using GPU,” in IEEE/RSJ International Conference on Intelligent Robots and Systems, 2022, pp. 2273–2280

  28. [36]

    MEM: Multi-modal elevation mapping for robotics and learning,

    G. Erni, J. Frey, T. Miki, M. Mattamala, and M. Hutter, “MEM: Multi-modal elevation mapping for robotics and learning,” in IEEE/RSJ International Conference on Intelligent Robots and Systems , 2023, pp. 11 011–11 018

  29. [37]

    Information theoretic MPC for model-based reinforcement learning,

    G. Williams, N. Wagener, B. Goldfain, P. Drews, J. M. Rehg, B. Boots, and E. A. Theodorou, “Information theoretic MPC for model-based reinforcement learning,” in IEEE International Conference on Robotics and Automation, 2017, pp. 1714–1721

  30. [38]

    A framework for self-training perceptual agents in simulated photorealistic environments,

    P. Mania and M. Beetz, “A framework for self-training perceptual agents in simulated photorealistic environments,” in IEEE International Conference on Robotics and Automation , 2019, pp. 4396–4402

Pith tools

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