Pith. sign in

REVIEW 4 major objections 6 minor 49 references

This paper proposes that visual navigation policies should be conditioned on the agent's body geometry—length, width, height, and maximum traversable step—and paired with risk-perception and trajectory-correction modules; doing so yields sa

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · deepseek-v4-flash

2026-08-01 11:24 UTC pith:FJ2237G5

load-bearing objection A useful dataset plus a sensible embodiment-conditioned IL pipeline, but the safety gains rest more on the paper's own augmentation generator than on independent evidence. the 4 major comments →

arxiv 2607.19880 v2 pith:FJ2237G5 submitted 2026-07-22 cs.RO cs.CV

EA-Nav: Learning Safe Visual Navigation Policies with Embodiment Awareness

Jialu Zhang , Yong Du , Xianda Guo , Shunwang Sun , Xinqi Liu , Yue Sun , Guodong Lu , Wei Sui
show 1 more author
Jituo Li
This is my paper
classification cs.RO cs.CV
keywords visual navigationcross-embodimentimitation learningembodiment geometryspatial perceptiontrajectory correctionrisk-aware navigation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

Visual navigation policies trained purely on pixels face an ambiguity: the same camera view can mean "go straight through a gap" for a small robot and "turn away" for a large one. This paper tries to resolve that ambiguity by making body geometry—length, width, height, and maximum traversable step height—an explicit conditioning input, and by adding two decoupled modules: one that perceives the nearest-obstacle distance along the predicted trajectory and one that corrects the trajectory with a small yaw rotation when that distance is too small. To train those modules without abundant real near-collision data, the authors generate high-risk trajectories offline by inflating obstacle maps with the robot's dimensions and checking collision for a pool of scaled and rotated variants. They report that the full system improves success rate and lowers collision rate across embodiment sizes in simulation and on physical robots, with the largest gains in cluttered scenes. The authors explicitly scope the contribution to size-related geometry, leaving turning radius and motion constraints as future work.

Core claim

The paper's central claim is that embodiment geometry—body length, width, height, and maximum traversable step height—should be a condition on which the navigation policy is trained, not a property it must guess from pixels. The claim is established through a two-stage imitation-learning pipeline: an internet-video pretraining stage that injects the embodiment vector as conditional tokens alongside visual, depth, and goal tokens, and a fine-tuning stage in which a spatial-perception head regresses per-waypoint minimum obstacle distance and a risk-aware correction head picks a global yaw offset from discretized bins when the distance falls below a threshold. The same predicted trajectory is s

What carries the argument

The load-bearing object is the embodiment vector m = [L_b, W_b, H_b, P_max], a four-dimensional geometric description that is injected as a conditional token so the policy can produce different actions from the same image sequence. Around it, the architecture is a decoupled prediction–perception–correction pipeline: a shared input encoder with unidirectional cross-attention keeps the context feature clean; an action head emits velocity commands; a spatial-perception head converts the trajectory into features and regresses the nearest-obstacle distance for each waypoint; and, when risk exceeds a threshold, a correction head classifies nineteen yaw-offset bins covering -45 to +45 degrees and s

Load-bearing premise

The collision and correction labels that supervise the risk modules are computed offline from monocular depth estimates and a fitted ground plane; if those pseudo-geometric labels are inaccurate in cluttered scenes, or if the future observations used to build the point cloud are unavailable at deployment, the reported safety gains may come from the augmentation generator rather than from genuine geometry.

What would settle it

Deploy the final policy on a robot with an enlarged embodiment in a cluttered corridor while feeding the spatial-perception module either the paper's monocular-depth-derived occupancy grid or a LiDAR-derived ground-truth grid; if collision rates do not worsen with the pseudo-labels, the safety benefit is not attributable to the geometric labels the augmentation pipeline depends on.

Watch this falsifier — get emailed when new claim-graph text bears on it.

If this is right

  • One imitation-learned policy can serve robots of different sizes if body geometry is supplied as an input; in qualitative runs the small body threads gaps while the large body detours.
  • In simulated cluttered scenes, adding the perception and correction modules lifts success rate over the base model by roughly a third, with larger gains where obstacles are denser.
  • Training with augmented high-risk trajectories raises recognition of collision samples roughly fivefold over training without augmentation, with near-zero additional false alarms.
  • On physical robots, the embodiment-conditioned policy sustains nonzero success when the body is enlarged, while vision-only point-goal and image-goal baselines fall to zero success or full collision.
  • The correction head keeps mean deflection between roughly 12 and 18 degrees, allowing recovery from risky trajectories with small, controllable turns even as the feasible angular region shrinks for large bodies.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • Because the risk modules are decoupled from the action head, their supervision could in principle be used to audit any existing waypoint-based navigation policy for embodiment-specific collision risk before deployment—a use the paper does not pursue.
  • The reliance on offline, future-looking geometry suggests a testable boundary: if the augmentation labels were recomputed from a live sensor (e.g., LiDAR) instead of monocular depth, the reported safety margins would likely change; measuring that shift would separate the conditioning effect from the label-source effect.
  • The paper's own limitation points to a concrete extension: adding turning radius and acceleration bounds to the embodiment vector should let the correction head respect non-holonomic constraints, which may matter more for vehicles and legged robots than size alone.
  • The pretraining-stage analysis implies that a visual-only model averages over behavioral modes; if so, balancing the internet-video dataset across embodiment categories may matter more than raw hours for cross-embodiment generalization.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. The paper proposes EA-Nav, an imitation-learning navigation framework that conditions the policy on embodiment geometry m=[L_b,W_b,H_b,P_max] (Eq. 1) and uses a decoupled action-prediction, spatial-perception, and risk-aware-correction architecture (Eqs. 5--6). Pretraining uses a newly collected ~1,000-hour cross-embodiment internet-video dataset with pseudo-depth labels; fine-tuning augments real navigation data with a risk-trajectory augmentation pipeline (Sec. 3.3) that builds an embodiment-inflated occupancy grid and synthesizes collision trajectories and feasible correction bins. The paper reports spatial-perception metrics on i2Nav, correction metrics on augmented trajectories, simulation results on three scenes, and real-world results with TurtleBot and Unitree Go2, claiming improved success and collision rates over point-goal and image-goal baselines.

Significance. If the claims are fully supported, the paper makes a useful contribution: it introduces a large heterogeneous multi-embodiment pretraining dataset, a practical IL-based alternative to RL for cross-embodiment navigation, and a modular way to inject embodiment geometry into perception and correction. The decoupled training design and the explicit handling of scarce high-risk samples are sensible and potentially reusable. I particularly credit the paper for reporting the limitations in Sec. 5 and Sec. A.6, and for providing results in both simulation and real-world settings. However, the current evidence for the safety-critical submodules is weaker than the narrative suggests: the perception and correction evaluations are based on labels produced by the same augmentation pipeline used for training, and the real-world results are based on only five trials per condition. The central navigation claim is not fully circular, because the simulation comparison in Table 4 is independent of the augmentation generator, but the magnitude of the claimed safety gains is not yet established.

major comments (4)
  1. [Sec. 4.4 / Sec. 3.3, Table 3] The trajectory-correction evaluation is circular. Sec. 4.4 states: "we use the risk augmentation module to generate collision trajectories and their corresponding correction targets." This is the same module that produces the training supervision for Risk-Aware Correction in Sec. 3.3. Consequently, the CSR and IoU values in Table 3 measure how well the network reproduces the generator's feasible-bin labels, not how well it avoids real obstacles. Because the generator relies on Depth-Anything-3 depth and a RANSAC ground-plane fit, any systematic bias in monocular depth, ground-plane estimation, or cluttered scenes is shared by the training and evaluation labels. In addition, the point-cloud fusion in Sec. 3.3 uses short-horizon future observations that are not available at deployment, making the evaluation task easier than the deployment task. I request an evaluation on independently obta
  2. [Sec. 4.3, Table 2] The spatial-perception metrics are computed against the same pseudo-geometric targets used to train the module. The augmented condition improves TPR by roughly 5x, but the targets are produced by the same Depth-Anything-3 + RANSAC pipeline described in Sec. 3.3, not by an independent geometric sensor or simulator. If the depth estimates or ground-plane fits are biased, the same bias is present in both training labels and evaluation labels, so high TPR/FAR/MAE values do not establish accurate risk perception in real cluttered scenes. Please add an independent ground-truth comparison, for example simulator depth/occupancy or LiDAR-derived grids, and report perception metrics separately for near/far and cluttered scenarios.
  3. [Sec. 4.5, Table 5] The real-world comparison is based on only five trials per method and embodiment size, with no variance or confidence intervals. On five trials, a single success/failure changes SR or CR by 0.2, so the differences between 0.8 and 0.6 in Table 5 are well within sampling noise. The statement that the method "maintains a high SR and a low CR" is not statistically supported at this sample size. I also could not find a description of how the Body, Body+, and Body++ embodiment sizes were physically or virtually realized on TurtleBot and Unitree Go2. Please report exact binomial confidence intervals (or more trials) and specify the embodiment-parameter changes used for each condition.
  4. [Tables 4--5, Sec. 4.5] The baseline comparison mixes point-goal navigation methods (iPlanner, NavDP) with image-goal methods (NoMaD*, ExAug*). Because goal representations and task definitions differ, the reported SR/CR differences may reflect the task formulation rather than navigation quality or embodiment awareness. For example, image-goal methods can fail from goal-image ambiguity rather than lack of embodiment conditioning. Please compare within the same task setting, or clearly separate and interpret the two groups. In addition, the claimed "average performance by approximately 31%" refers to the internal comparison between Ours(w/o corr) and Ours(w/ corr) in Table 4, not to improvements over baselines; this should be stated explicitly to avoid overstatement.
minor comments (6)
  1. [Sec. 3.1, Eqs. (3), (5), (6)] There are several formatting/notation errors: Eq. (3) has an improperly displayed norm, Eq. (5) shows an inconsistent assignment of \hat d, and Eq. (6) is missing parentheses in the printed text. These should be corrected.
  2. [Sec. 4.1] The hyperparameters \delta_s (0.5 m) and \tau (correction feasibility threshold) are fixed without sensitivity analysis. Since they directly determine what is counted as a high-risk sample and what correction is selected, please report the results for a small range of threshold values.
  3. [Table 4] No number of episodes, seeds, or variance measures are reported. Adding mean and std over multiple seeds (or at least episode counts) would make the simulation comparison more reliable.
  4. [Sec. 4.5 / Ref. [35]] The text refers to the simulation environment as InternUtopia, but reference [35] is titled Grutopia. Please reconcile the naming.
  5. [Sec. 4.2 / Fig. 5] The t-SNE and loss-curve analysis is suggestive but qualitative. Consider reporting a numeric measure of action ambiguity, such as the variance of action predictions or a conditional-entropy estimate, to support the claim that embodiment tokens reduce ambiguity.
  6. [Sec. 5 / Appendix A.6] The stated limitation is honest but incomplete: it omits the evaluation limitations discussed above (circular generation of perception/correction labels, small real-world sample size). Adding these caveats would make the limitation section more accurate.

Circularity Check

2 steps flagged

Correction/perception metrics are evaluated on labels produced by the same risk-augmentation pipeline that generated their training supervision; the headline navigation claim retains independent support from Tables 4-5.

specific steps
  1. self definitional [Sec. 4.4 / Table 3; Sec. 3.3 (collision candidates and feasible bins)]
    "Since collision samples are scarce in real-world data, we use the risk augmentation module to generate collision trajectories and their corresponding correction targets. ... Colliding candidates are treated as risk trajectories, while collision-free candidates define feasible correction bins. These bins provide multi-hot supervision over the discretized correction space, enabling large-scale supervision from limited fine-tuning data."

    The evaluation in Table 3 is built from the same Risk Trajectory Augmentation module that supplies the training supervision for Risk-Aware Correction (Eq. 6). The 'feasible correction bins' used as ground truth in the evaluation are, by construction, the outputs of the augmentation module's inflated-occupancy collision check. CSR and IoU therefore measure how well the model reproduces the augmentation generator, not whether the corrected trajectory is safe against independent geometry. This is an in-distribution evaluation of a module trained on the same labels, so the reported correction capability is partially circular.

  2. fitted input called prediction [Sec. 4.3 / Table 2; Sec. 3.3-3.4 (point cloud, ground plane, occupancy grid)]
    "Given depth observations, trajectories, and camera intrinsics over the next H timesteps, we fuse them into a robot-centric local point cloud map. ... We then estimate the ground plane and identify obstacles under embodiment constraints. ... obstacle regions are inflated to construct an embodiment-aware occupancy grid for subsequent collision checking. ... short image sequences are fused into local point clouds, from which the ground plane and obstacle regions are extracted to construct geometric supervision."

    The perception metrics in Table 2 (TPR/FAR/MAE) are defined against minimum obstacle distances on i2Nav, but the paper does not report an independent LiDAR or simulator ground truth. The same Depth-Anything-3 depth estimations, RANSAC ground-plane fit, and inflated occupancy grid that generate the augmented high-risk training samples are used to construct 'geometric supervision' in Sec. 3.4. Thus the spatial-perception 'predictions' are fitted to a pseudo-label generator and then evaluated against the same generator; the w/ Aug vs w/o Aug comparison quantifies agreement with that generator rather than true geometric safety.

full rationale

The central derivation chain is: pretrain a BC policy on Internet video with embodiment-geometry tokens (Eqs. 1-3), then fine-tune with a decoupled action/perception/correction architecture (Eqs. 5-6) supervised by Risk Trajectory Augmentation (Sec. 3.3). The headline cross-embodiment navigation claim is supported by Table 4, which uses NavDP/InternUtopia simulation assets, and by Table 5, which uses real TurtleBot and Unitree Go2 trials; those environments are external to the augmentation pipeline, so the main claim is not circular. The circularity is component-level: Table 3's correction targets are explicitly generated by the same risk-augmentation module that produces the multi-hot correction labels, and Table 2's perception targets come from the same pseudo-geometric pipeline (Depth-Anything-3 + RANSAC + occupancy inflation) used to create augmented training data. These tables therefore measure in-distribution fit to the label generator, not independent safety. The real-world evidence in Table 5 is too small (five trials per condition) to independently confirm the safety gain, but small sample size is a statistical limitation rather than a circularity. Self-citations are not load-bearing: ExAug is prior work by other authors, and external models (Depth-Anything-3, DINOv3, ConvNeXt) are independent tools. Score 4 reflects one clear construction-level circularity in the correction evaluation, with the central navigation result still carrying independent content; not a 6+ because the headline claim does not reduce to the augmentation generator.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 0 invented entities

The method's novel machinery rests on pseudo-label quality and augmentation-label correctness; no new physical entities are introduced.

free parameters (3)
  • Risk threshold δ_s = 0.5 m
    Used to label high-risk trajectories and as evaluation threshold for TPR/FAR (Sec. 4.1); chosen by hand, no sensitivity analysis.
  • Correction feasibility threshold τ = not reported
    Used to select feasible correction bins at inference (Sec. 3.2); value not specified in the paper.
  • Augmentation scale sampling α = sampled around safety coefficient s_d
    Controls how much trajectories are scaled toward obstacles in augmentation (Eq. 8); distribution not fully specified.
axioms (3)
  • domain assumption Depth-Anything-3 produces enough geometric accuracy for pseudo-labeling internet video.
    Sec. A.3: manual curation and scale-consistency argument; paper admits reconstruction errors for non-rigid motion.
  • domain assumption RANSAC ground-plane and inflated occupancy grid correctly reflect traversability.
    Sec. 3.3: collision checking for augmentation and labels; errors propagate to all risk modules.
  • ad hoc to paper Scale consistency, not metric accuracy, is sufficient for pretraining.
    Sec. A.3: authors argue factor-of-two scale deviation still yields useful samples; this is an unproven domain assumption tailored to the method.

pith-pipeline@v1.3.0-alltime-deepseek · 15218 in / 11672 out tokens · 120172 ms · 2026-08-01T11:24:38.612129+00:00 · methodology

0 comments
read the original abstract

Cross-embodiment navigation is a key challenge in embodied intelligence. Due to differences in embodiment, the same visual observation may imply different actions for different agents, making prediction ambiguous when relying solely on vision. Existing studies mainly rely on reinforcement learning, which requires large-scale interaction and careful reward design, making it difficult to support scalable pretraining and real-world adaptation. In contrast, imitation-learning-based approaches remain limited. To address these challenges, we propose an imitation-learning-based embodiment-aware navigation framework with a modular multi-stage design. In pretraining, we construct a cross-embodiment navigation dataset from Internet videos and introduce embodiment geometry as conditional tokens to reduce action ambiguity under the same observation. In fine-tuning, we design a multimodal information injection mechanism based on a decoupled architecture. Specifically, we design a trajectory augmentation strategy to generate high-risk samples, which are used to train spatial perception and risk-aware correction separately, thereby explicitly incorporating embodiment geometry for safe navigation. Experimental results show that the proposed method effectively improves navigation performance across different embodiment settings, demonstrating the effectiveness of incorporating embodiment geometry into embodied navigation.

Figures

Figures reproduced from arXiv: 2607.19880 by Guodong Lu, Jialu Zhang, Jituo Li, Shunwang Sun, Wei Sui, Xianda Guo, Xinqi Liu, Yong Du, Yue Sun.

Figure 1
Figure 1. Figure 1: Overview of our contributions. (a) We collect large-scale first-person videos from the Internet across diverse embod [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Overview of the EA-Nav framework. The Input Encoder jointly encodes visual observations, the goal, and embodiment [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Visualization of environment modeling. The red planar region represents the estimated ground obtained via plane [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Augmentation pipeline. Light blue: original trajec [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Analysis of embodiment geometry in pre-training. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Qualitative analysis of spatial perception perfor [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Embodiment-Aware Navigation Strategies Across Different Robots. Smaller robots tend to navigate through narrow [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Real-world experimental setup. Body denotes the [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Effect of scale deviation on navigation training [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Layout of the real-world evaluation environment. [PITH_FULL_IMAGE:figures/full_fig_p011_10.png] view at source ↗
Figure 11
Figure 11. Figure 11: Simulation assets used for evaluation, including [PITH_FULL_IMAGE:figures/full_fig_p011_11.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

49 extracted references · 7 linked inside Pith

  1. [1]

    Jong Wook Bae, Jungho Kim, Junyong Yun, Changwon Kang, Jeongseon Choi, Chanhyeok Kim, Junho Lee, Jungwook Choi, and Jun Won Choi. 2023. Sit dataset: socially interactive pedestrian trajectory dataset for social navigation robots. Advances in neural information processing systems36 (2023), 24552–24563

    2023

  2. [2]

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

    2025

  3. [3]

    Wenzhe Cai, Jiaqi Peng, Yuqiang Yang, Yujian Zhang, Meng Wei, Hanqing Wang, Yilun Chen, Tai Wang, and Jiangmiao Pang. 2025. Navdp: Learning sim-to-real navigation diffusion policy with privileged information guidance.arXiv preprint arXiv:2505.08712(2025)

    arXiv 2025

  4. [4]

    Jie Chen, Yuxin Cai, Yizhuo Wang, Ruofei Bai, Yuhong Cao, Jun Li, Yau Wei Yun, and Guillaume Sartoretti. 2026. ImagiNav: Scalable Embodied Navigation via Gen- erative Visual Prediction and Inverse Dynamics.arXiv preprint arXiv:2603.13833 (2026)

    arXiv 2026

  5. [5]

    Ziyi Chen, Yingnan Guo, Zedong Chu, Minghua Luo, Yanfen Shen, Mingchao Sun, Junjun Hu, Shichao Xie, Kuan Yang, Pei Shi, et al. 2025. Socialnav: Training human-inspired foundation model for socially-aware embodied navigation.arXiv preprint arXiv:2511.21135(2025)

    arXiv 2025

  6. [6]

    An-Chieh Cheng, Yandong Ji, Zhaojing Yang, Zaitian Gongye, Xueyan Zou, Jan Kautz, Erdem Bıyık, Hongxu Yin, Sifei Liu, and Xiaolong Wang. 2024. Nav- ila: Legged robot vision-language-action model for navigation.arXiv preprint arXiv:2412.04453(2024)

    Pith/arXiv arXiv 2024

  7. [7]

    Cheng Chi, Zhenjia Xu, Siyuan Feng, Eric Cousineau, Yilun Du, Benjamin Burch- fiel, Russ Tedrake, and Shuran Song. 2025. Diffusion policy: Visuomotor policy learning via action diffusion.The International Journal of Robotics Research44, 10-11 (2025), 1684–1704

    2025

  8. [8]

    Ainaz Eftekhar, Rose Hendrix, Luca Weihs, Jiafei Duan, Ege Caglar, Jordi Salvador, Alvaro Herrasti, Winson Han, Eli VanderBil, Aniruddha Kembhavi, et al. 2024. The one ring: a robotic indoor navigation generalist.arXiv preprint arXiv:2412.14401 (2024)

    Pith/arXiv arXiv 2024

  9. [9]

    Noriaki Hirose, Catherine Glossop, Dhruv Shah, and Sergey Levine. 2025. Om- niVLA: An omni-modal vision-language-action model for robot navigation.arXiv preprint arXiv:2509.19480(2025)

    arXiv 2025

  10. [10]

    Noriaki Hirose, Catherine Glossop, Ajay Sridhar, Oier Mees, and Sergey Levine

  11. [11]

    Noriaki Hirose, Lydia Ignatova, Kyle Stachowicz, Catherine Glossop, Sergey Levine, and Dhruv Shah. 2025. Learning to drive anywhere with model-based reannotation.IEEE Robotics and Automation Letters11, 2 (2025), 1242–1249

    2025

  12. [12]

    Noriaki Hirose, Dhruv Shah, Ajay Sridhar, and Sergey Levine. 2023. ExAug: Robot-conditioned navigation policies via geometric experience augmentation. In2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 4077–4084

    2023

  13. [13]

    Noriaki Hirose, Fei Xia, Roberto Martín-Martín, Amir Sadeghian, and Silvio Savarese. 2019. Deep visual mpc-policy learning for navigation.IEEE Robotics and Automation Letters4, 4 (2019), 3184–3191

    2019

  14. [14]

    Jiahui Huang, Qunjie Zhou, Hesam Rabeti, Aleksandr Korovko, Huan Ling, Xu- anchi Ren, Tianchang Shen, Jun Gao, Dmitry Slepichev, Chen-Hsuan Lin, et al

  15. [15]

    Xing Hui Jing, Xin Xiong, Fu Hao Li, Tao Zhang, and Long Zeng. 2024. A two-stage reinforcement learning approach for robot navigation in long-range indoor dense crowd environments. In2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 5489–5496

    2024

  16. [16]

    Vipe: Video pose engine for 3d geometric perception.arXiv preprint arXiv:2508.10934(2025)

    Pith/arXiv arXiv 2025

  17. [17]

    Zhengqi Li, Richard Tucker, Forrester Cole, Qianqian Wang, Linyi Jin, Vickie Ye, Angjoo Kanazawa, Aleksander Holynski, and Noah Snavely. 2025. Megasam: Accurate, fast and robust structure and motion from casual dynamic videos. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition. 10486–10496

    2025

  18. [18]

    Haresh Karnan, Anirudh Nair, Xuesu Xiao, Garrett Warnell, Sören Pirk, Alexander Toshev, Justin Hart, Joydeep Biswas, and Peter Stone. 2022. Socially compliant navigation dataset (scand): A large-scale dataset of demonstrations for social navigation.IEEE Robotics and Automation Letters7, 4 (2022), 11807–11814

    2022

  19. [19]

    Haotong Lin, Sili Chen, Junhao Liew, Donny Y Chen, Zhenyu Li, Guang Shi, Jiashi Feng, and Bingyi Kang. 2025. Depth anything 3: Recovering the visual space from any views.arXiv preprint arXiv:2511.10647(2025)

    Pith/arXiv arXiv 2025

  20. [20]

    Jing Liang, Dibyendu Das, Daeun Song, Md Nahid Hasan Shuvo, Mohammad Durrani, Karthik Taranath, Ivan Penskiy, Dinesh Manocha, and Xuesu Xiao. 2025. Gnd: Global navigation dataset with multi-modal perception and multi-category traversability in outdoor campus environments. In2025 IEEE International Con- ference on Robotics and Automation (ICRA). IEEE, 2383–2390

    2025

  21. [21]

    Abhijeet Nayak, Débora NP Oliveira, Samiran Gode, Cordelia Schmid, and Wol- fram Burgard. 2025. MetricNet: Recovering Metric Scale in Generative Navigation Policies.arXiv preprint arXiv:2509.13965(2025)

    arXiv 2025

  22. [22]

    Xinhao Liu, Jintong Li, Yicheng Jiang, Niranjan Sujay, Zhicheng Yang, Juexiao Zhang, John Abanes, Jing Zhang, and Chen Feng. 2025. Citywalker: Learning embodied urban navigation from web-scale videos. InProceedings of the Computer Vision and Pattern Recognition Conference. 6875–6885

    2025

  23. [23]

    Jesús Ortega-Peimbert, Finn Lukas Busch, Timon Homberger, Quantao Yang, and Olov Andersson. 2025. DIV-Nav: Open-Vocabulary Spatial Relationships for Multi-Object Navigation.arXiv preprint arXiv:2510.16518(2025)

    arXiv 2025

  24. [24]

    Dujun Nie, Xianda Guo, Yiqun Duan, Ruijun Zhang, and Long Chen. 2025. Wm- nav: Integrating vision-language models into world models for object goal navi- gation. In2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2392–2399

    2025

  25. [25]

    Dhruv Shah, Benjamin Eysenbach, Nicholas Rhinehart, and Sergey Levine. 2022. Rapid Exploration for Open-World Navigation with Latent Goal Models. InCon- ference on Robot Learning. PMLR, 674–684

    2022

  26. [26]

    Manthan Patel, Fan Yang, Yuheng Qiu, Cesar Cadena, Sebastian Scherer, Marco Hutter, and Wenshan Wang. 2025. Tartanground: A large-scale dataset for ground robot perception and navigation. In2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 20524–20531

    2025

  27. [27]

    Dhruv Shah, Ajay Sridhar, Nitish Dashora, Kyle Stachowicz, Kevin Black, Noriaki Hirose, and Sergey Levine. 2023. ViNT: A Foundation Model for Visual Navigation. InConference on Robot Learning. PMLR, 711–733

    2023

  28. [28]

    Dhruv Shah, Ajay Sridhar, Arjun Bhorkar, Noriaki Hirose, and Sergey Levine. 2023. Gnm: A general navigation model to drive any robot. In2023 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 7226–7233

    2023

  29. [29]

    Oriane Siméoni, Huy V Vo, Maximilian Seitzer, Federico Baldassarre, Maxime Oquab, Cijo Jose, Vasil Khalidov, Marc Szafraniec, Seungeun Yi, Michaël Rama- monjisoa, et al. 2025. Dinov3.arXiv preprint arXiv:2508.10104(2025)

    Pith/arXiv arXiv 2025

  30. [30]

    Hardik Shah, Erica Tevere, Deegan Atha, Marcel Kaufmann, Shehryar Khattak, Manthan Patel, Marco Hutter, Jonas Frey, and Patrick Spieler. 2026. WildOS: Open-Vocabulary Object Search in the Wild.arXiv preprint arXiv:2602.19308 (2026)

    arXiv 2026

  31. [31]

    Hailiang Tang, Tisheng Zhang, Liqiang Wang, Xin Ding, Man Yuan, Zhiyu Xiang, Jujin Chen, Yuhan Bian, Shuangyan Liu, Yuqing Wang, et al. 2025. I2Nav-robot: A large-scale indoor-outdoor robot dataset for multi-sensor fusion navigation and mapping.arXiv preprint arXiv:2508.11485(2025)

    Pith/arXiv arXiv 2025

  32. [32]

    Ajay Sridhar, Dhruv Shah, Catherine Glossop, and Sergey Levine. 2024. Nomad: Goal masked diffusion policies for navigation and exploration. In2024 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 63–70

    2024

  33. [33]

    Samuel Triest, Matthew Sivaprakasam, Sean J Wang, Wenshan Wang, Aaron M Johnson, and Sebastian Scherer. 2022. Tartandrive: A large-scale dataset for learning off-road dynamics models. In2022 International Conference on Robotics and Automation (ICRA). IEEE, 2546–2552

    2022

  34. [34]

    Yi-Fan Tang, Cong Tai, Fang-Xing Chen, Wan-Ting Zhang, Tao Zhang, Xue-Ping Liu, Yong-Jin Liu, and Long Zeng. 2024. Mobile robot oriented large-scale indoor dataset for dynamic scene understanding. In2024 IEEE international conference on robotics and automation (ICRA). IEEE, 613–620

    2024

  35. [35]

    Hanqing Wang, Jiahe Chen, Wensi Huang, Qingwei Ben, Tai Wang, Boyu Mi, Tao Huang, Siheng Zhao, Yilun Chen, Sizhe Yang, et al. 2024. Grutopia: Dream general robots in a city at scale.arXiv preprint arXiv:2407.10943(2024)

    Pith/arXiv arXiv 2024

  36. [36]

    Laurens Van der Maaten and Geoffrey Hinton. 2008. Visualizing data using t-SNE. Journal of machine learning research9, 11 (2008)

    2008

  37. [37]

    Jianyuan Wang, Minghao Chen, Nikita Karaev, Andrea Vedaldi, Christian Rup- precht, and David Novotny. 2025. Vggt: Visual geometry grounded transformer. InProceedings of the Computer Vision and Pattern Recognition Conference. 5294– 5306

    2025

  38. [38]

    Haitong Wang, Aaron Hao Tan, Angus Fung, and Goldie Nejat. 2025. X-Nav: Learning End-to-End Cross-Embodiment Navigation for Mobile Robots.IEEE Robotics and Automation Letters11, 1 (2025), 698–705

    2025

  39. [39]

    Shaoan Wang, Jiazhao Zhang, Minghan Li, Jiahang Liu, Anqi Li, Kui Wu, Fangwei Zhong, Junzhi Yu, Zhizheng Zhang, and He Wang. 2025. TrackVLA: Embodied Visual Tracking in the Wild. InConference on Robot Learning. PMLR, 4139–4164

    2025

  40. [40]

    Liuyi Wang, Xinyuan Xia, Hui Zhao, Hanqing Wang, Tai Wang, Yilun Chen, Chengju Liu, Qijun Chen, and Jiangmiao Pang. 2025. Rethinking the embodied gap in vision-and-language navigation: A holistic study of physical and visual disparities. InProceedings of the IEEE/CVF International Conference on Computer Vision. 9455–9465

    2025

  41. [41]

    Ruipu Wu, Yige Zhang, Jinyu Chen, Linjiang Huang, Shifeng Zhang, Xu Zhou, Liang Wang, and Si Liu. 2025. AeroDuo: Aerial Duo for UAV-based Vision and Language Navigation. InProceedings of the 33rd ACM International Conference on Multimedia. 2576–2585

    2025

  42. [42]

    Sanghyun Woo, Shoubhik Debnath, Ronghang Hu, Xinlei Chen, Zhuang Liu, In So Kweon, and Saining Xie. 2023. Convnext v2: Co-designing and scaling convnets with masked autoencoders. InProceedings of the IEEE/CVF conference on computer vision and pattern recognition. 16133–16142

    2023

  43. [43]

    Fan Yang, Chen Wang, Cesar Cadena, and Marco Hutter. 2023. iPlanner: Im- perative Path Planning.Proceedings of Robotics: Science and System XIX(2023), 064

    2023

  44. [44]

    Haoyu Xi, Mingao Tan, Xinming Zhang, Siwei Cheng, Shanze Wang, Yin Gu, Xiaoyu Shen, and Wei Zhang. 2026. CeRLP: A Cross-embodiment Robot Local Planning Framework for Visual Navigation.arXiv preprint arXiv:2603.19602 (2026). MM’26, November, 2026, Rio de Janeiro, Brazil Zhang et al

    arXiv 2026

  45. [45]

    Jiazhao Zhang, Anqi Li, Yunpeng Qi, Minghan Li, Jiahang Liu, Shaoan Wang, Haoran Liu, Gengze Zhou, Yuze Wu, Xingxing Li, et al. 2025. Embodied navigation foundation model.arXiv preprint arXiv:2509.12129(2025)

    arXiv 2025

  46. [46]

    Kai Yang, Tianlin Zhang, Zhengbo Wang, Zedong Chu, Xiaolong Wu, Yang Cai, and Mu Xu. 2025. CE-Nav: Flow-Guided Reinforcement Refinement for Cross-Embodiment Local Navigation.arXiv preprint arXiv:2509.23203(2025)

    arXiv 2025

  47. [47]

    Haokun Zhu, Zongtai Li, Zihan Liu, Kevin Guo, Zhengzhi Lin, Yuxin Cai, Guofei Chen, Chen Lv, Wenshan Wang, Jean Oh, et al. 2026. SysNav: Multi-Level Sys- tematic Cooperation Enables Real-World, Cross-Embodiment Object Navigation. arXiv preprint arXiv:2603.06914(2026). EA-Nav: Learning Safe Visual Navigation Policies with Embodiment Awareness MM’26, Novemb...

    arXiv 2026

  48. [48]

    Zijie Zheng, Zeshun Li, Yunpeng Wang, Qinghongbing Xie, and Long Zeng. 2025. DVS: Dynamic Virtual-Real Simulation Platform for Mobile Robotic Tasks. In RSS

    2025

  49. [2025]

    InConference on Robot Learning

    LeLaN: Learning A Language-Conditioned Navigation Policy from In-the- Wild Video. InConference on Robot Learning. PMLR, 666–688