Pith. sign in

REVIEW 3 major objections 7 minor 50 references

PixelNav: Towards Model-based Vision-Only Navigation with Topological Graphs

T0 review · 3 major / 7 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read PixelNav claims that a modular model-based vision-only navigation system can match or beat end-to-end learned policies in real-world indoor goal reaching, with greater interpretability and robustness to unseen obstacles.

desk verdict A coherent modular vision-only navigation pipeline with honest experiments, but the empirical support is thin and Eq. 8 has a likely typo that currently undermines the safety claim. read the letter →

arxiv 2507.20892 v1 pith:JMYVF7HF submitted 2025-07-28 cs.RO

Sergey Bakulin , Timur Akhtyamov , Denis Fatykhov , German Devchich , Gonzalo Ferrer This is my paper · ORCID
classification cs.RO MSC 68T4068T4593C85
keywords vision-onlynavigationtopologicalgraphmodelpredictivecontrolMPPItraversabilityestimationvisualplacerecognitioninverseperspectivemappinggoalreaching
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

PixelNav is a modular, vision-only navigation system that tries to show that a robot can reach goals as reliably as end-to-end learned navigation policies without being trained end-to-end itself. Instead of one network mapping image to action, it uses a topological graph of images as a map, visual place recognition to localize, a traversability segmentation network to mark safe ground, and a sampling-based model predictive controller (MPPI) to steer toward a subgoal pixel while avoiding projected obstacle points. The authors report that in two real indoor spaces PixelNav reaches the goal more often than the ViNT and NoMaD baselines, and that it reacts to previously unseen obstacles while the baseline policies tend to overfit the expert paths. Because the pipeline is modular, each component's errors are attributable and fixable; the paper itself identifies traversability misclassification of white walls and an imprecise inverse perspective mapping as the main failure sources.

What carries the argument

The load-bearing object is the MPPI cost made of two geometric terms that link pixels to the robot's body. The traversal mask $T_t=\tau(O_t)$ marks safe ground; the subgoal pixel $I p^{\mathrm{sg}}_t$ is picked by matching the current image to the subgoal image, recovering a yaw $\alpha_t$ from the essential matrix, and tracing a ray through the mask. The controller then minimizes $q(\mathbf{x}_k,u_k)=w_{\mathrm{obst}} q_{\mathrm{obst}}(\mathbf{x}_k)+w_{\mathrm{sg}} q_{\mathrm{sg}}(\mathbf{x}_k)+u_k^\top Q_{\mathrm{ctrl}}u_k$, where $q_{\mathrm{sg}}=\|P(\mathbf{x}_k)-I p^{\mathrm{sg}}_t\|_2$ keeps the forward projection $P$ of future robot positions near the subgoal pixel in image space, and $q_{\mathrm{obst}}$ counts, via the inverse perspective map $P^{-1}$, how many sampled obstacle contour points fall within a safety radius $r_{\mathrm{safe}}$ of the robot in the ground plane. All geometric quantities depend on a fixed camera height $h_{\mathrm{cam}}$ and known intrinsics; the assumptions that the camera is rigid, orthogonal to the ground, and at known height are the premises that make $P$ and $P^{-1}$ well-defined.

What would settle it

Place a white, textureless wall or box at a known location in the robot's path and check whether the traversability mask labels it as traversable and whether PixelNav's planned trajectories cross it; a direct collision with that target obstacle, reproducible across trials, would falsify the central robustness claim. More quantitatively, compare the IPM-projected obstacle positions against a LiDAR or metric ground-truth map and measure whether the discrepancy exceeds the robot's safety radius $r_{\mathrm{safe}}$.

Watch

Extended reading notes

Core claim

The central claim is that a model-based, hierarchical pipeline can be competitive with state-of-the-art end-to-end vision-only navigation in the real world. PixelNav achieves a goal-reaching rate of 0.9 in Space 1 and 0.8 in Space 2, against 0.7 and 0.0 for NoMaD and 0.2 and 0.0 for ViNT, and a lower target-obstacle direct collision rate than NoMaD in both spaces. The authors interpret this as evidence that explicit pixel-space planning over a traversability mask gives robustness to obstacles missing from the topological graph, because the controller computes collision costs from the current observation rather than from a memorized trajectory. They also claim the architecture is more interpretable and permits identification and repair of the bottleneck component, citing failures in traversability estimation and IPM precision as the current limits.

Load-bearing premise

The whole safety argument rests on the accuracy of the traversability mask and of the inverse perspective mapping, so if either mislabels or misprojects the scene, the controller plans against false geometry and the collision-avoidance claim breaks.

Editorial extensions

If this is right

  • A modular, model-based pipeline can serve as a practical alternative to end-to-end navigation when interpretability or certification matters.
  • Because failures are attributable to specific modules, improving traversability segmentation and IPM precision should directly raise goal-reaching and lower collision rates.
  • The system's explicit reaction to unseen obstacles suggests that per-frame traversability reasoning generalizes better than memorized expert trajectories to novel geometry.
  • The frozen behavior in the harder space implies that a simple rollback or recovery policy can be bolted on without retraining the navigation model.

Reading between the lines

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

  • If the traversability mask and IPM were improved to metric accuracy, this architecture might also close the gap on indirect collisions, since those stem from the robot losing sight of the obstacle—a problem that a short memory buffer or obstacle-velocity estimate could address.
  • The pixel-space subgoal cost is essentially a bearing and azimuth heuristic; testing it against a visual servoing baseline would clarify whether the ray-traced two-thirds-length pixel choice is doing the work.
  • The fixed-camera-height and camera-orthogonal-to-ground assumptions restrict deployment; using an onboard IMU or learned camera pose could extend the same pipeline to handheld or tilted-camera robots.
  • A natural stress test would be placing an obstacle after the camera has passed it, to see whether the lack of temporal memory causes repeated indirect collisions even with perfect traversability.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 7 minor

Summary. The paper proposes PixelNav, a hierarchical vision-only navigation system that combines a topological graph, visual place recognition (AnyLoc), a SegFormer-based traversability segmentation, SuperPoint/SuperGlue-based subgoal pixel selection, and an MPPI low-level controller. The method is evaluated in two indoor spaces against the end-to-end baselines ViNT and NoMaD, with 9 runs per method per space, and the authors report higher goal-reaching rates and lower target-obstacle collision rates for PixelNav. The central claim is that a modular, model-based pipeline can match or surpass learned end-to-end policies in real-world vision-only navigation while offering greater interpretability.

Significance. If the claims hold, PixelNav provides a valuable data point that model-based navigation with learned perceptual components can compete with modern end-to-end policies, and its modular design could facilitate diagnosis and future certification. The paper includes real-world comparisons, uses established components, and states that code will be released. However, the empirical support is thin: only 9 runs per method per space with no statistical analysis, and the formal presentation of the inverse perspective mapping contains an undefined symbol. The authors themselves identify imprecise IPM and traversability errors as failure sources, which tempers the robustness claim. The contribution is interesting but requires stronger evidence and a corrected formal derivation.

major comments (3)
  1. [Section 3.5, Eq. (8)] The inverse perspective mapping in Eq. (8) uses an undefined focal-length symbol f_h in the lateral coordinate. The forward projection in Eqs. (5)-(6) implies R_y = -(I_u - c_x) f_y h_cam / (f_x (I_v - c_y)), with f_y appearing in both coordinates; the printed f_h is never defined in Section 3.5, Table 1, or elsewhere. Because the collision cost q_obst in Eq. (10) relies on P^{-1} to place obstacle points in the robot frame, this ill-specified mapping directly affects the formal validity of the collision-avoidance claim. Please correct the typo to f_y or define f_h explicitly.
  2. [Section 4.2, Tables 2-3] The comparative claims are based on only 3 trials per condition and 9 runs per method per space, with no variance reporting or significance tests. For example, in Space 1 the difference in goal-reaching rate between PixelNav (0.9) and NoMaD (0.7) corresponds to one run, and in Table 3 the 0.8 GRR for PixelNav is 7 out of 9 runs. The statement that PixelNav 'significantly outperforms' is not supported by the data. Please provide confidence intervals or a significance test, increase the number of runs, or hedge the comparative claims accordingly.
  3. [Section 4.4] The failure analysis explicitly lists 'Imprecise IPM' and 'Errors in traversability estimation' as major sources of failure. Since the MPPI collision cost in Eq. (10) depends on the IPM and the traversability mask, these admissions imply that the geometric basis for collision avoidance is only heuristic. This is in tension with the concluding statement that PixelNav shows 'robustness to the unseen obstacles.' To make the robustness claim supportable, please quantify how often these failure modes affected the reported runs, or restrict the claim to the specific perturbation scenarios where the method succeeded.
minor comments (7)
  1. [Section 3.2, Eq. (1)] Equation (1) writes arctan2 of a vector divided by its norm; this should be expressed as atan2 of the y- and x-components of the displacement vector.
  2. [Table 1] Several parameters mentioned in the text (ρ, φmax, lsg, MPPI horizon, number of samples) are not listed in Table 1; please include them for reproducibility.
  3. [Section 3.4] The subgoal pixel selection heuristic is under-specified; please define precisely how the ray is traced through the traversable mask, what 'upper border' means, and how the 2/3 rule is applied.
  4. [Section 4.2] The metric definitions should clarify whether TDCR is averaged over perturbation trials only, and how manual interventions after freezes affect the reported GRR and collision counts.
  5. [Abstract and Section 1] The claim of a 'significantly higher level of interpretability' is not supported by any analysis; please define interpretability in this context and discuss how the modular architecture enables it.
  6. [Introduction] There is a typo: 'the born of a research branch' should be 'the birth of a research branch'; also, Eq. (10) contains a duplicated '=' sign.
  7. [Section 3.3] The traversability model is trained on the EgoWalk dataset, whose authors overlap with the current paper's authors; this overlap should be disclosed in the main text, and ideally an analysis of the model's generalization to environments outside the dataset distribution should be provided.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: PixelNav's navigation performance is an empirical benchmark result, not a derivation from its own outputs.

full rationale

I walked the derivation chain from graph construction (Eqs. 1-3), traversability mask τ(O_t), subgoal pixel selection, MPPI cost (Eqs. 9-11), to the real-world metrics. No step defines a quantity in terms of the outcome it is used to predict. The MPPI collision cost Eq. 10 is a standard indicator cost over back-projected obstacle points; it is not fitted to the reported goal-reaching rate or TDCR. The traversability model is trained on EgoWalk, which is cited from an overlapping author group, but that citation is a data source, not an unverified uniqueness theorem or an ansatz smuggled in by citation; the paper's load-bearing claim is the real-world comparison against ViNT and NoMaD using official weights. The hand-tuned parameters (wobst, wsg, Qctrl, rsafe, segmentation threshold, 2/3 ray fraction) are standard controller tuning and do not turn the goal-reaching metric into a fit. The undefined f_h in Eq. 8 is an internal formal or typographical error that undermines the precision of the IPM, and the paper itself concedes that the IPM is imprecise and acts as a heuristic; this is a correctness and safety concern, not a circularity, because no result is equivalent to its inputs by construction. No circular step can be exhibited, so the score is 0.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

The system relies on a set of hand-tuned parameters, a self-authored traversability dataset (EgoWalk), and several geometric assumptions about camera placement and ground-plane projection. No new physical entities are introduced. The central empirical comparison is grounded in external baselines, but the lack of sensitivity analysis on the tuned parameters and the admitted failure modes in traversability and IPM make these assumptions load-bearing.

free parameters (7)
  • graph edge distance coefficient rho = not stated
    Tunable coefficient in Euclidean edge criterion (Eq. 2); affects graph connectivity and path planning.
  • graph angular threshold phi_max = not stated
    Tunable threshold in angular edge criterion (Eq. 3); limits feasible transitions for the unicycle model.
  • subgoal node offset lsg = not stated
    Offset index selecting the subgoal along the planned path, re-used over multiple time steps (Section 3.1).
  • traversability segmentation threshold = 0.95
    Binary segmentation threshold from Table 1; determines which pixels are treated as traversable.
  • MPPI costs weights wobst, wsg, Qctrl = wobst=10, wsg=10, Qctrl=diag(1,100)
    Cost weights in Eq. 11 tuned during real-world debugging, as stated in Section 3.5.
  • collision safety radius rsafe = 2 meters
    Threshold distance in Eq. 10, set to approximate robot radius and safety level; the authors note IPM imprecision justifies the high value.
  • subgoal pixel ray fraction = approximately 2/3
    Heuristic in Section 3.4 to pick a safe point along the ray through the traversable mask, avoiding mask edges.
assumptions (5)
  • domain assumption Camera height is fixed and known, and the camera plane is orthogonal to the ground surface (Section 3.5 assumptions 1-3).
    The inverse perspective mapping in Eq. 8 and the obstacle cost in Eq. 10 depend on these geometric assumptions. If violated, the projection and back-projection are biased.
  • domain assumption The traversability model trained on EgoWalk generalizes to the two test spaces and to unseen obstacles.
    The collision avoidance cost relies on the binary mask being correct; the paper reports failures on white walls and textureless regions in Section 4.4.
  • standard math The essential matrix decomposition from SuperPoint/SuperGlue matches gives a correct relative yaw angle.
    Subgoal pixel selection traces a ray at the estimated yaw angle alpha_t; if matching or decomposition fails, the target pixel is wrong.
  • domain assumption The binary traversability mask is a complete model of the obstacle-free ground surface.
    Obstacle points are sampled from mask contours, so any unlabeled obstacle or mislabeled wall becomes invisible to the planner.
  • domain assumption The unicycle kinematic model in Eq. 4 adequately describes the AgileX Tracer robot.
    MPPI plans over this model; systematic model error would degrade the closed-loop behavior.

how reviews work

0 comments
Cite this review

Pith. "Pith review of PixelNav: Towards Model-based Vision-Only Navigation with Topological Graphs." pith.science (2026). https://pith.science/paper/JMYVF7HF

@misc{pith2026250720892,
  author       = {Pith},
  title        = {Pith review of: PixelNav: Towards Model-based Vision-Only Navigation with Topological Graphs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JMYVF7HF}},
  note         = {Machine review of arXiv:2507.20892}
}
read the original abstract

This work proposes a novel hybrid approach for vision-only navigation of mobile robots, which combines advances of both deep learning approaches and classical model-based planning algorithms. Today, purely data-driven end-to-end models are dominant solutions to this problem. Despite advantages such as flexibility and adaptability, the requirement of a large amount of training data and limited interpretability are the main bottlenecks for their practical applications. To address these limitations, we propose a hierarchical system that utilizes recent advances in model predictive control, traversability estimation, visual place recognition, and pose estimation, employing topological graphs as a representation of the target environment. Using such a combination, we provide a scalable system with a higher level of interpretability compared to end-to-end approaches. Extensive real-world experiments show the efficiency of the proposed method.

Figures

Figures reproduced from arXiv: 2507.20892 by the authors.

Figure 1
Figure 1. Overview of the method belong to the projection of the safe and obstacle-free re￾gions of the environment’s surface. This mask is obtained with a deep traversability estimation model τ : Tt = τ (Ot). The subgoal pixel Ip sg t , where I stands for the image plane coordinate system, is selected using heuristics described in the sections below. Finally, the MPC-based low-level planner (controller) uses the previously o… view at source ↗
Figure 3
Figure 3. Subgoal pixel selection algorithm illustration. The se [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 2
Figure 2. Example of the traversability prediction. The green mask [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figures from the paper (4 more)
Figure 4
Figure 4. Figure 4: Example of the contour points sampling and IPM [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 6
Figure 6. Figure 6: Direct and indirect collisions examples Since existing vision-only approaches have limited tem￾poral and spatial reasoning capabilities, we argue that it is not fair to expect perfect zero-collision performance. Thus, to make an evaluation more fair, we propose to spli…
Figure 5
Figure 5. Figure 5: Spaces selected for the evaluation For each location and each method, we first evaluate the performance of the method in no perturbation setting: the state of the environment is similar to the one from the recorded topological graph. The goal of the model is to fol￾low…
Figure 7
Figure 7. Figure 7: Example of the traversability estimation failures in the [PITH_FULL_IMAGE:figures/full_fig_p007_7.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

50 extracted references · 42 canonical work pages

  1. [1]

    Vision-only robot navigation in a neural ra- diance world

    Michal Adamkiewicz, Timothy Chen, Adam Caccavale, Rachel Gardner, Preston Culbertson, Jeannette Bohg, and Mac Schwager. Vision-only robot navigation in a neural ra- diance world. IEEE Robotics and Automation Letters, 7(2): 4606–4613, 2022. 2

    work page 2022

  2. [2]

    Egowalk: A multimodal dataset for robot navigation in the wild

    Timur Akhtyamov, Mohamad Al Mdfaa, Javier Anto- nio Ramirez, Sergey Bakulin, German Devchich, De- nis Fatykhov, Alexander Mazurov, Kristina Zipa, Malik Mohrat, Pavel Kolesnik, et al. Egowalk: A multimodal dataset for robot navigation in the wild. arXiv preprint arXiv:2505.21282, 2025. 3

    arXiv 2025

  3. [3]

    Netvlad: Cnn architecture for weakly supervised place recognition

    Relja Arandjelovic, Petr Gronat, Akihiko Torii, Tomas Pa- jdla, and Josef Sivic. Netvlad: Cnn architecture for weakly supervised place recognition. In Proceedings of the IEEE conference on computer vision and pattern recognition , pages 5297–5307, 2016. 6

    work page 2016

  4. [4]

    Navigation world models

    Amir Bar, Gaoyue Zhou, Danny Tran, Trevor Darrell, and Yann LeCun. Navigation world models. In Proceedings of the Computer Vision and Pattern Recognition Conference , pages 15791–15801, 2025. 2

    2025

  5. [5]

    Risk-aware off-road navigation via a learned speed dis- tribution map

    Xiaoyi Cai, Michael Everett, Jonathan Fink, and Jonathan P How. Risk-aware off-road navigation via a learned speed dis- tribution map. In 2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 2931–2937. IEEE, 2022. 2

    work page 2022

  6. [6]

    Probabilistic traversability model for risk-aware motion planning in off-road environ- ments

    Xiaoyi Cai, Michael Everett, Lakshay Sharma, Philip R Osteen, and Jonathan P How. Probabilistic traversability model for risk-aware motion planning in off-road environ- ments. In 2023 IEEE/RSJ International Conference on In- telligent Robots and Systems (IROS) , pages 11297–11304. IEEE, 2023

    work page 2023

  7. [7]

    Evora: Deep evidential traversability learning for risk-aware off- road autonomy

    Xiaoyi Cai, Siddharth Ancha, Lakshay Sharma, Philip R Os- teen, Bernadette Bucher, Stephen Phillips, Jiuguang Wang, Michael Everett, Nicholas Roy, and Jonathan P How. Evora: Deep evidential traversability learning for risk-aware off- road autonomy. IEEE Transactions on Robotics, 2024. 2

    work page 2024

  8. [8]

    Splat-nav: Safe real-time robot navigation in gaussian splatting maps

    Timothy Chen, Ola Shorinwa, Joseph Bruno, Aiden Swann, Javier Yu, Weijia Zeng, Keiko Nagami, Philip Dames, and Mac Schwager. Splat-nav: Safe real-time robot navigation in gaussian splatting maps. IEEE Transactions on Robotics,

    work page

Show all 50 references
  1. [9]

    Prescribed safety performance imitation learning from a single expert dataset

    Zhihao Cheng, Li Shen, Miaoxi Zhu, Jiaxian Guo, Meng Fang, Liu Liu, Bo Du, and Dacheng Tao. Prescribed safety performance imitation learning from a single expert dataset. IEEE transactions on pattern analysis and machine intelli- gence, 45(10):12236–12249, 2023. 1

    2023

  2. [10]

    Mobility vla: Multimodal instruction navigation with long-context vlms and topological graphs

    Hao-Tien Lewis Chiang, Zhuo Xu, Zipeng Fu, Mithun George Jacob, Tingnan Zhang, Tsang-Wei Ed- ward Lee, Wenhao Yu, Connor Schenck, David Rendleman, Dhruv Shah, et al. Mobility vla: Multimodal instruction navigation with long-context vlms and topological graphs. arXiv preprint a...

    2024 arXiv

  3. [11]

    End-to- end imitation learning with safety guarantees using control barrier functions

    Ryan K Cosner, Yisong Yue, and Aaron D Ames. End-to- end imitation learning with safety guarantees using control barrier functions. In 2022 IEEE 61st Conference on Decision and Control (CDC), pages 5316–5322. IEEE, 2022. 1

    2022

  4. [12]

    Superpoint: Self-supervised interest point detection and description

    Daniel DeTone, Tomasz Malisiewicz, and Andrew Rabi- novich. Superpoint: Self-supervised interest point detection and description. In Proceedings of the IEEE conference on computer vision and pattern recognition workshops , pages 224–236, 2018. 4

    2018

  5. [13]

    A brief survey of sim2real methods for robot learning

    Konstantinos Dimitropoulos, Ioannis Hatzilygeroudis, and Konstantinos Chatzilygeroudis. A brief survey of sim2real methods for robot learning. In International Conference on Robotics in Alpe-Adria Danube Region , pages 133–140. Springer, 2022. 2

    2022

  6. [14]

    Certi- fiable robustness to adversarial state uncertainty in deep rein- forcement learning

    Michael Everett, Bj ¨orn L¨utjens, and Jonathan P How. Certi- fiable robustness to adversarial state uncertainty in deep rein- forcement learning. IEEE Transactions on Neural Networks and Learning Systems, 33(9):4184–4198, 2021. 1

    2021

  7. [15]

    Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography.Communications of the ACM, 24(6):381–395, 1981

    Martin A Fischler and Robert C Bolles. Random sample consensus: a paradigm for model fitting with applications to image analysis and automated cartography.Communications of the ACM, 24(6):381–395, 1981. 4

    1981

  8. [16]

    Wayfast: Navigation with predictive traversability in the field

    Mateus V Gasparino, Arun N Sivakumar, Yixiao Liu, An- dres EB Velasquez, Vitor AH Higuti, John Rogers, Huy Tran, and Girish Chowdhary. Wayfast: Navigation with predictive traversability in the field. IEEE Robotics and Automation Letters, 7(4):10651–10658, 2022. 2

    2022

  9. [17]

    Wayfaster: a self-supervised traversability prediction for increased navigation awareness

    Mateus V Gasparino, Arun N Sivakumar, and Girish Chowd- hary. Wayfaster: a self-supervised traversability prediction for increased navigation awareness. In 2024 IEEE Inter- national Conference on Robotics and Automation (ICRA) , pages 8486–8492. IEEE, 2024. 2

    2024

  10. [18]

    Flownav: Learning efficient navigation policies via condi- tional flow matching

    Samiran Gode, Abhijeet Nayak, and Wolfram Burgard. Flownav: Learning efficient navigation policies via condi- tional flow matching. In 2nd CoRL Workshop on Learning Effective Abstractions for Planning, 2024. 1, 2, 6

    2024

  11. [19]

    Sacson: Scalable autonomous control for social nav- igation

    Noriaki Hirose, Dhruv Shah, Ajay Sridhar, and Sergey Levine. Sacson: Scalable autonomous control for social nav- igation. IEEE Robotics and Automation Letters, 9(1):49–56,

  12. [20]

    Lelan: Learning a language-conditioned navigation policy from in-the-wild video

    Noriaki Hirose, Catherine Glossop, Ajay Sridhar, Oier Mees, and Sergey Levine. Lelan: Learning a language-conditioned navigation policy from in-the-wild video. In Conference on Robot Learning, pages 666–688. PMLR, 2025. 2

    2025

  13. [21]

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

    Sanghun Jung, JoonHo Lee, Xiangyun Meng, Byron Boots, and Alexander Lambert. V-strong: Visual self-supervised traversability learning for off-road navigation. In 2024 IEEE International Conference on Robotics and Automation (ICRA), pages 1766–1773. IEEE, 2024. 2

    2024

  14. [22]

    Socially compliant navigation dataset (scand): A large-scale dataset of demonstrations for social navigation

    Haresh Karnan, Anirudh Nair, Xuesu Xiao, Garrett War- nell, S ¨oren Pirk, Alexander Toshev, Justin Hart, Joydeep Biswas, and Peter Stone. Socially compliant navigation dataset (scand): A large-scale dataset of demonstrations for social navigation. IEEE Robotics and Automation ...

    2022

  15. [23]

    Learning semantic traversability with egocentric video and automated annotation strategy

    Yunho Kim, Jeong Hyun Lee, Choongin Lee, Juhyeok Mun, Donghoon Youm, Jeongsoo Park, and Jemin Hwangbo. Learning semantic traversability with egocentric video and automated annotation strategy. IEEE Robotics and Automa- tion Letters, 2024. 2, 3

    2024

  16. [24]

    Segment any- thing

    Alexander Kirillov, Eric Mintun, Nikhila Ravi, Hanzi Mao, Chloe Rolland, Laura Gustafson, Tete Xiao, Spencer White- head, Alexander C Berg, Wan-Yen Lo, et al. Segment any- thing. In Proceedings of the IEEE/CVF international confer- ence on computer vision, pages 4015–4026, 2023. 2

    2023

  17. [25]

    Vision-based navigation using deep reinforcement learning

    Jon ´aˇs Kulh ´anek, Erik Derner, Tim De Bruin, and Robert Babuˇska. Vision-based navigation using deep reinforcement learning. In 2019 european conference on mobile robots (ECMR), pages 1–8. IEEE, 2019. 2

    2019

  18. [26]

    Interpretable deep learning: Interpretation, interpretability, trustworthi- ness, and beyond

    Xuhong Li, Haoyi Xiong, Xingjian Li, Xuanyu Wu, Xiao Zhang, Ji Liu, Jiang Bian, and Dejing Dou. Interpretable deep learning: Interpretation, interpretability, trustworthi- ness, and beyond. Knowledge and Information Systems , 64 (12):3197–3234, 2022. 1

    2022

  19. [27]

    Citywalker: Learning embodied urban navigation from web-scale videos

    Xinhao Liu, Jintong Li, Yicheng Jiang, Niranjan Sujay, Zhicheng Yang, Juexiao Zhang, John Abanes, Jing Zhang, and Chen Feng. Citywalker: Learning embodied urban navigation from web-scale videos. In Proceedings of the Computer Vision and Pattern Recognition Conference, pages 68...

    2025

  20. [28]

    Robot operating system 2: Design, architecture, and uses in the wild

    Steven Macenski, Tully Foote, Brian Gerkey, Chris Lalancette, and William Woodall. Robot operating system 2: Design, architecture, and uses in the wild. Science Robotics, 7(66):eabm6074, 2022. 5

    2022

  21. [29]

    Performance of optical flow techniques for indoor navigation with a mobile robot

    Chris McCarthy and Nick Bames. Performance of optical flow techniques for indoor navigation with a mobile robot. In IEEE International Conference on Robotics and Automation,

  22. [30]

    Toward human-like social robot navigation: A large-scale, multi-modal, social human navigation dataset

    Duc M Nguyen, Mohammad Nazeri, Amirreza Payandeh, Aniket Datar, and Xuesu Xiao. Toward human-like social robot navigation: A large-scale, multi-modal, social human navigation dataset. In 2023 IEEE/RSJ International Confer- ence on Intelligent Robots and Systems (IROS), pages 7442–

    2023

  23. [31]

    Visual navigation of mo- bile robot using optical flow and visual potential field

    Naoya Ohnishi and Atsushi Imiya. Visual navigation of mo- bile robot using optical flow and visual potential field. In International Workshop on Robot Vision , pages 412–426. Springer, 2008. 2

    2008

  24. [32]

    Dinov2: Learning robust visual features without supervision

    Maxime Oquab, Timoth ´ee Darcet, Th ´eo Moutakanni, Huy V o, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, et al. Dinov2: Learning robust visual features without supervision. Transactions on Machine Learning Research Jou...

    2024

  25. [33]

    Vision-based perception for autonomous vehicles in obsta- cle avoidance scenarios, 2025

    Van-Hoang-Anh Phan, Chi-Tam Nguyen, Doan-Trung Au, Thanh-Danh Phan, Minh-Thien Duong, and My-Ha Le. Vision-based perception for autonomous vehicles in obsta- cle avoidance scenarios, 2025. 2

    2025

  26. [34]

    Viplanner: Visual semantic imperative learn- ing for local navigation

    Pascal Roth, Julian Nubert, Fan Yang, Mayank Mittal, and Marco Hutter. Viplanner: Visual semantic imperative learn- ing for local navigation. In 2024 IEEE International Confer- ence on Robotics and Automation (ICRA), pages 5243–5249. IEEE, 2024. 2

    2024

  27. [35]

    Superglue: Learning feature matching with graph neural networks

    Paul-Edouard Sarlin, Daniel DeTone, Tomasz Malisiewicz, and Andrew Rabinovich. Superglue: Learning feature matching with graph neural networks. In Proceedings of the IEEE/CVF conference on computer vision and pattern recognition, pages 4938–4947, 2020. 4

    2020

  28. [36]

    Semi-parametric topological memory for navigation

    Nikolay Savinov, Alexey Dosovitskiy, and Vladlen Koltun. Semi-parametric topological memory for navigation. In In- ternational Conference on Learning Representations , 2018. 2

    2018

  29. [37]

    Viking: Vision-based kilometer-scale navigation with geographic hints

    Dhruv Shah and Sergey Levine. Viking: Vision-based kilometer-scale navigation with geographic hints. arXiv preprint arXiv:2202.11271, 2022. 2

    2022 arXiv

  30. [38]

    Rapid exploration for open- world navigation with latent goal models

    Dhruv Shah, Benjamin Eysenbach, Gregory Kahn, Nicholas Rhinehart, and Sergey Levine. Rapid exploration for open- world navigation with latent goal models. arXiv preprint arXiv:2104.05859, 2021. 1, 2

    2021 arXiv

  31. [39]

    Ving: Learning open- world navigation with visual goals

    Dhruv Shah, Benjamin Eysenbach, Gregory Kahn, Nicholas Rhinehart, and Sergey Levine. Ving: Learning open- world navigation with visual goals. In 2021 IEEE Inter- national Conference on Robotics and Automation (ICRA) , pages 13215–13222. IEEE, 2021. 2

    2021

  32. [40]

    Gnm: A general navigation model to drive any robot

    Dhruv Shah, Ajay Sridhar, Arjun Bhorkar, Noriaki Hirose, and Sergey Levine. Gnm: A general navigation model to drive any robot. arXiv preprint arXiv:2210.03370, 2022. 1, 2

    2022 arXiv

  33. [41]

    Vint: A foundation model for visual navigation

    Dhruv Shah, Ajay Sridhar, Nitish Dashora, Kyle Stachow- icz, Kevin Black, Noriaki Hirose, and Sergey Levine. Vint: A foundation model for visual navigation. In 7th Annual Conference on Robot Learning, 2023. 1, 2, 3, 6, 7

    2023

  34. [42]

    Nomad: Goal masked diffusion policies for nav- igation and exploration

    Ajay Sridhar, Dhruv Shah, Catherine Glossop, and Sergey Levine. Nomad: Goal masked diffusion policies for nav- igation and exploration. In 2024 IEEE International Con- ference on Robotics and Automation (ICRA) , pages 63–70. IEEE, 2024. 1, 2, 6, 7

    2024

  35. [43]

    Topological structural analysis of dig- itized binary images by border following

    Satoshi Suzuki et al. Topological structural analysis of dig- itized binary images by border following. Computer vision, graphics, and image processing, 30(1):32–46, 1985. 5

    1985

  36. [44]

    Deep patch visual odometry

    Zachary Teed and Jia Deng. Deep patch visual odometry. In European Conference on Computer Vision, pages 460–477. Springer, 2022. 3

    2022

  37. [45]

    Anyloc: A foun- dation model for long-term visual place recognition

    Qi Wang, Zixin Cao, Yifan Yu, Zhichao Wang, Chang Fu, Xin Yang, Hang Zhou, and Andreas Geiger. Anyloc: A foun- dation model for long-term visual place recognition. arXiv preprint arXiv:2307.16849, 2023. 1, 3

    2023 arXiv

  38. [46]

    Information theoretic mpc for model-based re- inforcement learning

    Grady Williams, Nolan Wagener, Brian Goldfain, Paul Drews, James M Rehg, Byron Boots, and Evangelos A Theodorou. Information theoretic mpc for model-based re- inforcement learning. In 2017 IEEE international confer- ence on robotics and automation (ICRA), pages 1714–1721. IEEE...

    2017

  39. [47]

    Segformer: Simple and efficient design for semantic segmentation with transform- ers

    Enze Xie, Wenhai Wang, Zhiding Yu, Anima Anandkumar, Jose M Alvarez, and Ping Luo. Segformer: Simple and efficient design for semantic segmentation with transform- ers. Advances in neural information processing systems, 34: 12077–12090, 2021. 3

    2021

  40. [48]

    A survey on visual navigation for artificial agents with deep reinforce- ment learning

    Fanyu Zeng, Chen Wang, and Shuzhi Sam Ge. A survey on visual navigation for artificial agents with deep reinforce- ment learning. IEEE Access, 8:135426–135442, 2020. 2

    2020

  41. [49]

    Target-driven vi- sual navigation in indoor scenes using deep reinforcement learning

    Yuke Zhu, Roozbeh Mottaghi, Eric Kolve, Joseph J Lim, Ab- hinav Gupta, Li Fei-Fei, and Ali Farhadi. Target-driven vi- sual navigation in indoor scenes using deep reinforcement learning. In 2017 IEEE international conference on robotics and automation (ICRA), pages 3357–3364. I...

    2017

  42. [2004]

    Proceedings. ICRA’04. 2004, pages 5093–5098. IEEE,

    2004

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.