Pith. sign in

REVIEW 2 major objections 5 minor 2 cited by

Stow: Robotic Packing of Items into Fabric Pods

T0 review · 2 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read A deployed robotic system can stow diverse items into packed warehouse pods at over 85 percent success and near-human speed.

desk verdict Real deployed robotic stowing at scale, with a headline density claim that the results never actually measure. read the letter →

arxiv 2505.04572 v1 pith:QZHTODYV submitted 2025-05-07 cs.RO

Nicolas Hudson , Josh Hooks , Rahul Warrier , Curt Salisbury , Ross Hartley , Kislay Kumar , Bhavana Chandrashekhar , Paul Birkmeyer
show 29 more authors
Bosch Tang Matt Frost Shantanu Thakar Tony Piaskowy Petter Nilsson Josh Petersen Neel Doshi Alan Slatter Ankit Bhatia Cassie Meeker Yuechuan Xue Dylan Cox Alex Kyriazis Bai Lou Nadeem Hasan Asif Rana Nikhil Chacko Ruinian Xu Siamak Faal Esi Seraj Mudit Agrawal Kevin Jamieson Alessio Bisagni Valerie Samzun Christine Fuller Alex Keklak Alex Frenkel Lillian Ratliff Aaron Parness
This is my paper · ORCID
classification cs.RO
keywords roboticstowingwarehouseautomationcompliantmanipulationfabricpodpackingbinlearneddepthperceptionrisk-awareplanningkinestheticfeedback
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

Stowing, placing individual items into densely packed fabric shelves for later retrieval, is a manual warehouse job done billions of times per year, and robots have historically failed at it because item diversity and contact-rich placement create too many defects. This paper claims that a production robotic system, deployed in a live e-commerce fulfillment center, performed over 500,000 stows with 85.86 percent success on the most recent 100,000 attempts, at 224 units per hour compared with 243 units per hour for human stowers on the same floor. The system divides the task into specialized hardware: a conveyor-paddle gripper with an extendable plank, a separate robot that opens the elastic bands, and a human inductor who quality-checks and feeds items, all coordinated by learned perception that sees through the translucent bands and by a risk-aware planner that chooses item-bin matches to maximize throughput. If the numbers hold, it is evidence that dense, contact-rich placement of diverse objects is automatable at commercial scale, and the reported failure modes point to the remaining bottlenecks: band interactions, deformable items, and damage during insertion.

What carries the argument

The load-bearing object is the multi-mask, a layered orthographic map of each bin compiled from learned depth and segmentation outputs that are trained to ignore the translucent elastic bands, giving the robot a frontal, perspective-corrected view of items and free space even under occlusion. Around that map, the system combines a space-estimation step that uses both perception and kinesthetic traces from previous stows, a set of canonical insertion behaviors (direct insert, stack, and several sweep variants) generated by convolving task-specific kernels with cost maps, and a match planner that scores each item-bin-behavior triple by expected units per hour using a success-risk model. The extendable plank and conveyor paddles are the hardware corollary: they let the robot create space without using the in-hand item as a pushing tool, which the paper argues keeps item damage low.

What would settle it

Remove the human inductor or feed the full unscreened warehouse item distribution to the same workcells, then run another 100,000 stow attempts and compare per-class success, amnesty, and damage rates; if they depart substantially from 85.86 percent success, 3.77 percent amnesty, and 0.24 percent damage, the central claim is falsified in its current scope.

Watch

Extended reading notes

Core claim

The central claim is that a compliant manipulation system can place items into densely packed fabric pods at the production standards of a real e-commerce warehouse. Over 100,000 stow attempts, 85.86 percent were successful, 9.31 percent were unproductive but recyclable, 3.77 percent resulted in amnesty (items falling to the floor), and 0.24 percent resulted in damage, while the robot stowed at 224 units per hour against 243 for human stowers on the same floor. The authors attribute this to a task decomposition: a band manipulator opens the elastic mesh; the stow end effector uses conveyor paddles to eject items without entering the bin and a thin plank to sweep and compress existing items; and the loop is driven by learned bin maps, kinesthetic feedback, and risk-based match planning. They also report that a learned risk model improves stow rate by about 7 percent over the deployed frequentist policy in a pod-level A/B test.

Load-bearing premise

The measured performance depends on a human inductor who quality-checks and filters every incoming item for robot eligibility, so the reported success and defect rates apply only to that filtered subset.

Editorial extensions

If this is right

  • Warehouse stowing can be automated without sacrificing throughput: the robot ran at 224 units per hour against 243 for humans, and robots can operate around the clock.
  • The dominant remaining failure modes are placement problems rather than grasping: band overlap caused 19 percent of amnesty, and deformable or thin items are hard to monitor kinesthetically.
  • A learned risk model that deliberately explores riskier item-bin matches raised stow rate by about 7 percent over a fixed heuristic, supporting continued investment in learned planning with exploration.
  • Damage, at 0.24 percent of attempts, concentrates in specific insertion interactions, including books damaged during insertion and lightweight boxes crushed by the fixed 80 N grip force.
  • Deploying robots to upper shelves could remove step-ladder use and raise overall human stow rates; the paper estimates a 4.5 percent lift if robots handle only the top rows of pods.

Reading between the lines

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

  • Because a human inductor quality-checks and filters every incoming item, the headline success and defect rates describe only the robot-eligible stream; automating singulation would widen the item distribution and would likely change all reported numbers.
  • The offline learned space-estimation model, with roughly 2.5 cm RMSE versus 4.0 cm for the deployed heuristics, suggests a testable upgrade: run it online and use visual tracking of deformable items during ejection to abort poor inserts before they become amnesty.
  • If the dataset described in the paper is released, it would let other groups train and benchmark placement-success predictors on real production outcomes and kinesthetic space traces rather than on simulation, which is where the authors argue stowing data is scarce.
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

2 major / 5 minor

Summary. The paper reports a production robotic stow system deployed in an Amazon fulfillment center. The system combines a conveyor-jaw end effector with an extendable plank, a band-opening manipulator, learned perception that predicts depth and segmentation through semi-transparent elastic bands, a library of force-controlled bin-manipulation behaviors, and a match planner that uses frequentist or learned risk models. Results from 100,000 annotated stow attempts give 85.86% success, 0.24% damage, 3.77% amnesty, and a rate of 224 UPH versus 243 UPH for human stowers on the same floor; an A/B test shows the learned risk model raises UPH by about 7% (p=0.008). The paper argues these design choices make dense, high-rate packing tractable and catalogs failure modes.

Significance. This is a meaningful systems-and-deployment contribution. The scale (100,000 human-annotated stows, more than 500,000 total), the randomized A/B comparison of task-planning algorithms, and the detailed failure-mode analysis are strengths that distinguish the paper from lab-demo manipulation work. The end-effector morphology and the separation of band manipulation, in-bin space creation, and item insertion are plausible and interesting design choices. The unpublished dataset, if released, would further increase the contribution. On the other hand, the headline 'human levels of packing density' claim is currently unmeasured, and the performance numbers are for a human-filtered item stream; these points limit the paper's conclusions as written.

major comments (2)
  1. [Abstract; Sections III, VIII, IX] The abstract's claim that the system 'achieves human levels of packing density and speed' is not supported for density. Section III identifies density as a key metric measured in volumetric occupancy or gross cubic utilization, but Section IX reports no such measurement for robot-stowed pods and no comparison with human-stowed pods. The average of 8 items per pod face (Section III) and the free-space RMSE values (Section IX-A) are not occupancy or utilization metrics, and the match planner in Section VIII optimizes expected UPH with density only implicitly encouraged. Please add a quantitative density comparison against human stowing, or revise the abstract and conclusion to claim speed and success parity without the density assertion.
  2. [Section V-A; Sections IX, X] The reported 85.86% success, 3.77% amnesty, and 224 UPH are measured on an item stream that has been pre-screened and singulated by a human inductor, who removes ineligible or damaged items (Section V-A). The paper discloses this, but the abstract and conclusion describe the robot as performing 'over 500,000 stows' without bounding the claim to this filtered input distribution. Since removing the human filter would widen the item distribution and likely change all headline metrics, the abstract or results should state the scope explicitly, and the introduction's 'designed to stow 80% of items' target should not be presented as achieved for the unfiltered stream.
minor comments (5)
  1. [Table III] The values in the 'Avg UPH ± 95%CI' column are printed as (313,302) and (336,316); clarify the convention of the interval and state how the confidence interval was constructed.
  2. [Conclusion vs. footnote 2] The conclusion states that a test dataset 'has been published and shared with the community', but footnote 2 says the dataset is planned for release after the paper is in review; align these statements.
  3. [Section X vs. Section IX] The conclusion says 'over 500,000 stows at greater than 85% success', while Section IX analyzes only the most recent 100,000 attempts; state explicitly that the 85% figure is for the analyzed batch.
  4. [Section IX-A] The kinesthetically informed free-space bias of 0.15 mm seems implausibly small relative to the perception-only bias of 36 mm; please verify the units.
  5. [Section VIII-B] The phrase 'training regiment' should be 'training regimen'.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the paper's central success, UPH, and A/B results are external empirical measurements, not reductions of fitted inputs.

full rationale

The paper's central claims are deployment measurements. Section IX reports 100,000 stow attempts with outcomes manually annotated and validated (Table II: 85.86% success, 3.77% amnesty, 0.24% damage), and Section IX-B compares 224 UPH robot vs 243 UPH human on the same floor. These are direct empirical observations, not predictions derived from fitted assumptions. The learned risk model (Section VIII-B) and learned space model (Section VI-B1) are trained on historical or kinesthetic data and then evaluated by A/B testing or offline RMSE; the evaluation labels are physical outcomes or held-out comparisons, so the improvements do not reduce by construction to the training targets. The self-citations that appear ([23] in Section II and [30] in Section VI-A4) are contextual pointers or an empirical baseline comparison on identical datasets; neither imports an unverified uniqueness claim nor forces the paper's conclusions. The abstract's 'human levels of packing density' assertion is not supported by any occupancy or utilization metric in Section IX, and the human-inductor pre-screening in Section V-A limits the scope of the reported rates. Both are correctness or measurement concerns rather than circularity: a missing or scoped measurement is not a derivation that reduces to its own inputs. No circular step can be exhibited from the paper's own equations, so the score is 0.

Assumptions & free parameters 2 free parameters · 3 assumptions · 0 invented entities

This is an empirical systems paper; the central claim rests on deployed hardware, a human-in-the-loop intake process, and learned models trained on real data. No new physical entities or mathematical objects are postulated.

free parameters (2)
  • Fixed EoAT clamp force = 80 N
    Set as a constant in Section VII-A; the paper attributes over 10% of damage cases to crushing lightweight boxes with this force, so it directly affects reported damage rates.
  • Frequentist planner safety margin = Not specified
    Used in Section VIII-A to prune item-bin matches; described as a 'heuristically set margin', it affects the balance of success rate and packing density but its value is not reported.
assumptions (3)
  • standard math The depth-to-3D conversion X=f*b/D from stereo disparity is valid for the bin crops.
    Invoked in Section VI-A4 to convert predicted disparity maps to depth; standard stereo geometry.
  • domain assumption The learned perception models trained on synthetic plus fine-tuned real images generalize to deployment bin states.
    The system relies on unseen band occlusion inference from Section VI-A; no quantitative generalization guarantee is given beyond the reported RMSE.
  • domain assumption A human inductor pre-screens every item for quality and robot eligibility before it reaches the robot.
    Stated in Section V-A; the success and defect rates are therefore conditional on the pre-filtered item distribution.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Stow: Robotic Packing of Items into Fabric Pods." pith.science (2026). https://pith.science/paper/QZHTODYV

@misc{pith2026250504572,
  author       = {Pith},
  title        = {Pith review of: Stow: Robotic Packing of Items into Fabric Pods},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QZHTODYV}},
  note         = {Machine review of arXiv:2505.04572}
}
read the original abstract

This paper presents a compliant manipulation system capable of placing items onto densely packed shelves. The wide diversity of items and strict business requirements for high producing rates and low defect generation have prohibited warehouse robotics from performing this task. Our innovations in hardware, perception, decision-making, motion planning, and control have enabled this system to perform over 500,000 stows in a large e-commerce fulfillment center. The system achieves human levels of packing density and speed while prioritizing work on overhead shelves to enhance the safety of humans working alongside the robots.

Figures

Figures reproduced from arXiv: 2505.04572 by the authors.

Figure 1
Figure 1. Storage Pod: Items are stored densely and hetero [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. In the Robotic Stow Process, a human checks inbound items for quality and feeds three robotic workcells. The workcells [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Top: Plan view of system with A) induct station, [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (7 more)
Figure 4
Figure 4. Figure 4: Orientation EoAT and Hand-off: A SCARA robot with [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: a) The robot EoAT uses parallel jaws with built-in conveyors. With this strategy, items are fed into the jaws by simple [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]
Figure 6
Figure 6. Figure 6: Synthetic images of pods along with their segmentation and depth maps that see through partially occluding translucent [PITH_FULL_IMAGE:figures/full_fig_p007_6.png]
Figure 7
Figure 7. Figure 7: System architecture for motion planning and control: [PITH_FULL_IMAGE:figures/full_fig_p009_7.png]
Figure 8
Figure 8. Figure 8: Affordance generation using cost map-based convolution. Individual multi-mask layers are transformed and fused into a [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]
Figure 9
Figure 9. Figure 9: Item-insertion cost map (left) generated from the bin [PITH_FULL_IMAGE:figures/full_fig_p011_9.png]
Figure 10
Figure 10. Figure 10: Stow outcomes over 100,000 attempts. TABLE II: Stow outcomes over 100,000 attempts. Bin Manipulation Outcome Count Percent Avg. Cycle Behavior Time (s) All Successful 85,859 85.86% – Unproductive 9,311 9.31% – Defect 4,830 4.83% – – Amnesty 3,767 3.77% – – Other 819 0…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 2 Pith papers

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

  1. Learning 3D Affordances for Blade Insertion in Cluttered Stowing

    cs.CV 2026-06 conditional novelty 6.5 of 10

    VulcanVoxel reconstructs blade occupancy with a 3D masked autoencoder, recovering multi-modal free-space affordances from unimodal warehouse stow data and raising top-5 coverage from 0.71 to 0.89.

  2. NeSyPack: A Neuro-Symbolic Framework for Bimanual Logistics Packing

    cs.RO 2025-06 conditional novelty 5.0 of 10

    NeSyPack, a hierarchical neuro-symbolic controller, achieved high packing success rates and won the WBCD competition at ICRA 2025.

Reference graph

Works this paper leans on

37 extracted references · 24 canonical work pages · cited by 2 Pith papers

  1. [1]

    Coordinating hundreds of cooperative, autonomous vehicles in warehouses,

    P. R. Wurman, R. D’Andrea, and M. Mountz, “Coordinating hundreds of cooperative, autonomous vehicles in warehouses,” AI magazine, vol. 29, no. 1, pp. 9–9, 2008

    2008

  2. [2]

    Integrating different levels of automation: Lessons from winning the amazon robotics challenge 2016,

    C. H. Corbato, M. Bharatheesha, J. van Egmond, J. Ju, and M. Wisse, “Integrating different levels of automation: Lessons from winning the amazon robotics challenge 2016,” IEEE Transactions on Industrial Informatics, vol. 14, no. 11, pp. 4916–4926, 2018

    work page 2016

  3. [3]

    Anal- ysis and observations from the first amazon picking challenge,

    N. Correll, K. E. Bekris, D. Berenson, O. Brock, A. Causo, K. Hauser, K. Okada, A. Rodriguez, J. M. Romano, and P. R. Wurman, “Anal- ysis and observations from the first amazon picking challenge,” IEEE Transactions on Automation Science and Engineering , vol. 15, no. 1, pp. 172–188, 2018

    work page 2018

  4. [4]

    Cartman: The low-cost cartesian manipulator that won the amazon robotics challenge,

    D. Morrison, A. Tow, M. McTaggart, R. Smith, N. Kelly-Boxall, S. Wade-McCue, J. Erskine, R. Grinover, A. Gurman, T. Hunn, D. Lee, A. Milan, T. Pham, G. Rallos, A. Razjigaev, T. Rowntree, K. Vijay, Z. Zhuang, C. Lehnert, I. Reid, P. Corke, and J. Leitner, “Cartman: The low-cost cartesian manipulator that won the amazon robotics challenge,” in 2018 IEEE Int...

    work page 2018

  5. [5]

    Push planning for object placement on cluttered table surfaces,

    A. Cosgun, T. Hermans, V . Emeli, and M. Stilman, “Push planning for object placement on cluttered table surfaces,” in 2011 IEEE/RSJ International Conference on Intelligent Robots and Systems , 2011, pp. 4627–4632

    work page 2011

  6. [6]

    Learning to place objects: Organizing a room,

    G. Basu, Y . Jiang, and A. Saxena, “Learning to place objects: Organizing a room,” in 2012 IEEE International Conference on Robotics and Automation, 2012, pp. 3545–3546

    work page 2012

  7. [7]

    Predicting Object Interactions with Behavior Primitives: An Application in Stowing Tasks

    H. Chen, Y . Niu, K. Hong, S. Liu, Y . Wang, Y . Li, and K. Driggs-Campbell, “Predicting object interactions with behavior primitives: An application in stowing tasks,” 2023. [Online]. Available: https://arxiv.org/abs/2309.16873 2We are planning to release the dataset to the public upon approval after the paper is in review 16

    work page Pith review arXiv 2023

  8. [8]

    Visual stability prediction and its application to manipulation,

    W. Li, A. Leonardis, and M. Fritz, “Visual stability prediction and its application to manipulation,” in AAAI Spring Symposia , 2017. [Online]. Available: http://aaai.org/ocs/index.php/SSS/SSS17/paper/view/15263

    work page 2017

Show all 37 references
  1. [9]

    ’good robot!’: Efficient reinforcement learning for multi-step visual tasks with sim to real transfer,

    A. Hundt, B. Killeen, N. Greene, H. Wu, H. Kwon, C. Paxton, and G. Hager, “’good robot!’: Efficient reinforcement learning for multi-step visual tasks with sim to real transfer,” IEEE Robotics and Automation Letters, vol. 5, no. 4, pp. 6724–6731, Oct. 2020

    2020

  2. [10]

    Beyond pick-and-place: Tackling robotic stacking of diverse shapes,

    A. X. Lee, C. Devin, Y . Zhou, T. Lampe, K. Bousmalis, J. T. Springenberg, A. Byravan, A. Abdolmaleki, N. Gileadi, D. Khosid, C. Fantacci, J. E. Chen, A. Raju, R. Jeong, M. Neunert, A. Laurens, S. Saliceti, F. Casarini, M. A. Riedmiller, R. Hadsell, and F. Nori, “Beyond pick-a...

    2021

  3. [11]

    Deep learning approaches to grasp synthesis: A review,

    R. Newbury, M. Gu, L. Chumbley, A. Mousavian, C. Eppner, J. Leitner, J. Bohg, A. Morales, T. Asfour, D. Kragic, D. Fox, and A. Cosgun, “Deep learning approaches to grasp synthesis: A review,” IEEE Trans- actions on Robotics , vol. 39, no. 5, pp. 3994–4015, 2023

    2023

  4. [12]

    K. M. Lynch and F. C. Park, Modern Robotics: Mechanics, Planning, and Control, 1st ed. USA: Cambridge University Press, 2017

    2017

  5. [13]

    Avoiding object damage in robotic manipulation,

    E. Aduh, F. Wang, D. Randle, K. Wang, P. Shah, C. Mitash, and M. Nambi, “Avoiding object damage in robotic manipulation,” in 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), 2024, pp. 9518–9525

    2024

  6. [14]

    Robopack: Learning tactile-informed dynamics models for dense packing,

    B. Ai, S. Tian, H. Shi, Y . Wang, C. Tan, Y . Li, and J. Wu, “Robopack: Learning tactile-informed dynamics models for dense packing,” Robotics: Science and Systems (RSS) , 2024. [Online]. Available: https://arxiv.org/abs/2407.01418

    2024 arXiv

  7. [15]

    Simulation-assisted learning for efficient bin-packing of deformable packages in a bimanual robotic cell,

    O. M. Manyar, H. Ye, M. Sagare, S. Mayya, F. Wang, and S. K. Gupta, “Simulation-assisted learning for efficient bin-packing of deformable packages in a bimanual robotic cell,” in 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , 2024, pp. 5211– 5218

    2024

  8. [16]

    Recent advances on two- dimensional bin packing problems,

    A. Lodi, S. Martello, and D. Vigo, “Recent advances on two- dimensional bin packing problems,” Discrete Applied Mathematics , vol. 123, no. 1, pp. 379–396, 2002. [Online]. Available: https: //www.sciencedirect.com/science/article/pii/S0166218X0100347X

    2002

  9. [17]

    The three-dimensional bin packing problem for deformable items,

    Q. Zuo, X. Liu, L. Xu, L. Xiao, C. Xu, J. Liu, and W. K. V . Chan, “The three-dimensional bin packing problem for deformable items,” in 2022 IEEE International Conference on Industrial Engineering and Engineering Management (IEEM) , 2022, pp. 0911–0918

    2022

  10. [18]

    Dense robotic packing of irregular and novel 3d objects,

    F. Wang and K. Hauser, “Dense robotic packing of irregular and novel 3d objects,” IEEE Transactions on Robotics , vol. 38, no. 2, pp. 1160– 1173, 2022

    2022

  11. [19]

    Towards robust product packing with a minimalistic end-effector,

    R. Shome, W. N. Tang, C. Song, C. Mitash, H. Kourtev, J. Yu, A. Boularias, and K. E. Bekris, “Towards robust product packing with a minimalistic end-effector,” in2019 International Conference on Robotics and Automation (ICRA) , 2019, pp. 9007–9013

    2019

  12. [20]

    Structdiffusion: Language-guided creation of physically-valid structures using unseen objects,

    W. Liu, Y . Du, T. Hermans, S. Chernova, and C. Paxton, “Structdiffusion: Language-guided creation of physically-valid structures using unseen objects,” Robotics: Science and Systems (RSS) , 2023. [Online]. Available: https://arxiv.org/abs/2211.04604

    2023 arXiv

  13. [21]

    Compositional Diffusion-Based Continuous Constraint Solvers,

    Z. Yang, J. Mao, Y . Du, J. Wu, J. B. Tenenbaum, T. Lozano-P ´erez, and L. P. Kaelbling, “Compositional Diffusion-Based Continuous Constraint Solvers,” in Conference on Robot Learning , 2023

    2023

  14. [22]

    Shelving, stacking, hanging: Relational pose diffusion for multi-modal rearrangement,

    A. Simeonov, A. Goyal, L. Manuelli, L. Yen-Chen, A. Sarmiento, A. Rodriguez, P. Agrawal, and D. Fox, “Shelving, stacking, hanging: Relational pose diffusion for multi-modal rearrangement,” 2023. [Online]. Available: https://arxiv.org/abs/2307.04751

    2023 arXiv

  15. [23]

    Pick planning strategies for large-scale package manipulation,

    S. Li, A. Keipour, K. Jamieson, N. Hudson, S. Szhao, C. Swan, and K. Bekris, “Pick planning strategies for large-scale package manipulation,” Robotics: Science and Systems (RSS) , 2023. [Online]. Available: https://roboticsconference.org/2023/program/papers/023/

    2023

  16. [24]

    Practical stereo matching via cascaded recurrent network with adaptive correlation,

    J. Li, P. Wang, P. Xiong, T. Cai, Z. Yan, L. Yang, J. Liu, H. Fan, and S. Liu, “Practical stereo matching via cascaded recurrent network with adaptive correlation,” in Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition , 2022, pp. 16 263–16 272

    2022

  17. [25]

    Rtmdet: An empirical study of designing real-time object detectors,

    C. Lyu, W. Zhang, H. Huang, Y . Zhou, Y . Wang, Y . Liu, S. Zhang, and K. Chen, “Rtmdet: An empirical study of designing real-time object detectors,” 2022. [Online]. Available: https://arxiv.org/abs/2212.07784

    2022 arXiv

  18. [26]

    Fully convolutional networks for semantic segmentation,

    J. Long, E. Shelhamer, and T. Darrell, “Fully convolutional networks for semantic segmentation,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition , 2015, pp. 3431–3440

    2015

  19. [27]

    Deep residual learning for image recognition,

    K. He, X. Zhang, S. Ren, and J. Sun, “Deep residual learning for image recognition,” in Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, 2016, pp. 770–778

    2016

  20. [28]

    Mask r-cnn,

    K. He, G. Gkioxari, P. Doll ´ar, and R. Girshick, “Mask r-cnn,” in 2017 IEEE International Conference on Computer Vision (ICCV) , 2017, pp. 2980–2988

    2017

  21. [29]

    Hartley and A

    R. Hartley and A. Zisserman, Multiple View Geometry in Computer Vision, 2nd ed. Cambridge University Press, 2004

    2004

  22. [30]

    Depth estimation through translucent surfaces,

    S. Dai, B. Lou, P. Nilsson, S. Thakar, C. Meeker, A. Gordon, X. Kong, J. Zhang, B. Knorlein, H. Liu, B. Chandrashekhar, and S. Karumanchi, “Depth estimation through translucent surfaces,”

  23. [31]

    Swin transformer v2: Scaling up capacity and resolution,

    Z. Liu, H. Hu, Y . Lin, Z. Yao, Z. Xie, Y . Wei, J. Ning, Y . Cao, Z. Zhang, L. Dong et al., “Swin transformer v2: Scaling up capacity and resolution,” in Proceedings of the IEEE/CVF conference on computer vision and pattern recognition , 2022, pp. 12 009–12 019

    2022

  24. [32]

    A tutorial on newton methods for constrained trajectory optimization and relations to slam, gaussian process smoothing, optimal control, and probabilistic inference,

    M. Toussaint, “A tutorial on newton methods for constrained trajectory optimization and relations to slam, gaussian process smoothing, optimal control, and probabilistic inference,” Geometric and numerical founda- tions of movements , pp. 361–392, 2017

    2017

  25. [33]

    A new approach to time-optimal path parameterization based on reachability analysis,

    H. Pham and Q.-C. Pham, “A new approach to time-optimal path parameterization based on reachability analysis,” IEEE Transactions on Robotics, vol. 34, no. 3, pp. 645–659, 2018

    2018

  26. [34]

    Motion planning with sequential convex optimization and convex collision checking,

    J. Schulman, Y . Duan, J. Ho, A. Lee, I. Awwal, H. Bradlow, J. Pan, S. Patil, K. Goldberg, and P. Abbeel, “Motion planning with sequential convex optimization and convex collision checking,” The International Journal of Robotics Research , vol. 33, no. 9, pp. 1251–1270, 2014

    2014

  27. [35]

    R. S. Sutton and A. G. Barto, Reinforcement learning: an introduction . Cambridge, MA: MIT Press, 1998

    1998

  28. [36]

    Deligrasp: Inferring object properties with llms for adaptive grasp policies,

    W. Xie, M. Valentini, J. Lavering, and N. Correll, “Deligrasp: Inferring object properties with llms for adaptive grasp policies,” in Conference on Robot Learning , 2024

    2024

  29. [2025]

    Available: https://www.amazon.science/publications/ depth-estimation-through-translucent-surfaces

    [Online]. Available: https://www.amazon.science/publications/ depth-estimation-through-translucent-surfaces

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.