Pith. sign in

REVIEW 4 major objections 4 minor 1 cited by

SoGraB: A Visual Method for Soft Grasping Benchmarking and Evaluation

T0 review · 4 major / 4 minor · reviewed 2026-08-12 · deepseek-v4-flash

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

Pith's one-line read This paper introduces SoGraB, a visual benchmarking protocol that ranks soft grippers by the deformation they cause to grasped objects, measured as the Density-Aware Chamfer Distance between point clouds before and during grasping.

desk verdict SoGraB is a sensible, well-documented first step toward a standard soft-grasp benchmark, but the safety claim rests on an unvalidated proxy and the validation is mostly internal. read the letter →

arxiv 2411.19408 v1 pith:J3LORHHY submitted 2024-11-28 cs.RO

Benjamin G. Greenland , Josh Pinskier , Xing Wang , Daniel Nguyen , Ge Shi , Tirthankar Bandyopadhyay , Jen Jen Chung , David Howard This is my paper · ORCID
classification cs.RO
keywords softroboticsgraspbenchmarkingqualityobjectdeformationpointcloudcomparisonChamferdistanceFin-Raygrippernon-contactstressproxy
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

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

The reading

This paper proposes SoGraB, a benchmarking protocol that scores soft grippers by how much they deform the objects they grasp, measured by the Density-Aware Chamfer Distance between 3D point clouds captured before and during a grasp. The score combines grasp success, holding time, and deformation into a single scalar in [0,1]. The authors validate the protocol by ranking three Fin-Ray gripper designs (with 4, 6, and 8 ribs) plus a rigid gripper on 900 grasps across 45 objects with varied geometry and hardness. They find the protocol separates grippers cleanly in a middle stiffness range where soft gripping matters, and that very stiff or very soft objects do not distinguish grippers. If correct, SoGraB gives the field a standardized, instrumentation-free way to compare soft grippers and to guide design choices.

What carries the argument

The central object is the SoGraB score, a scalar in $[0,1]$ computed from grasp success, holding time, and the Density-Aware Chamfer Distance (DCD) between point clouds of the object before and during grasping. DCD is a bounded, density-aware variant of Chamfer distance that tolerates occlusion and density variation, making it suitable for comparing incomplete point clouds from a single depth camera pair. The score formula assigns $0$ to failed grasps, a time-weighted value in $[0,0.5]$ to drops, and $1 - d_{\mathrm{DCD}}/2$ to successful grasps, so deformation penalizes the score continuously. The protocol is completed by an iterative closest point alignment step that corrects for slippage and rotation between the two point clouds.

What would settle it

A direct test would instrument the same objects with force or stress sensing during the exact grasps SoGraB scores, or compare DCD values against visible damage thresholds for a soft material. If two grasps with equal DCD produce measurably different internal stress or material failure, or if a gripper that SoGraB ranks as safer is shown to damage an object more, the central claim that DCD tracks grasp safety is falsified.

Watch

Extended reading notes

Core claim

The central claim is that object deformation, captured visually as the density-aware Chamfer distance between pre- and post-grasp point clouds, is a usable non-contact proxy for the stress a gripper imposes on a soft object, and that combining it with success rate and holding time yields a benchmark that ranks grippers by grasp quality. The paper's experiments show that the protocol ranks the three Fin-Ray designs from softest to hardest on several soft objects, while scores converge on objects that are either very stiff (no deformation to distinguish grippers) or very soft relative to all grippers (all deform equally). This is taken as evidence that SoGraB identifies the stiffness range in which soft grasping is beneficial and that it can serve as a standard evaluator for future gripper designs.

Load-bearing premise

The protocol assumes that the density-aware Chamfer distance between pre- and post-grasp point clouds is a valid proxy for the stress or damage inflicted on the object, so that a smaller DCD always means a safer grasp; this premise is stated in Section III.A but is not checked against force-sensor readings, simulation stress fields, or material failure data.

Editorial extensions

If this is right

  • Any robotics lab with a robot arm, a 3D printer, and a depth camera can run the SoGraB protocol without modifying the gripper or object, enabling direct comparison of soft gripper designs across labs.
  • The published 900-grasp baseline dataset lets future gripper designs be scored against a fixed set of 45 objects and 4 reference grippers.
  • The protocol shows that soft grippers only outperform rigid ones in a middle stiffness range, so designers can use SoGraB to decide when a soft gripper is worth using for a given object.
  • Because the score is object-centric and continuous, it can rank not just grippers but also grasp configurations, grasp forces, and control policies for the same gripper-object pair.
  • The validation that scores converge for very soft and very stiff objects suggests SoGraB can be used to select evaluation objects that are actually informative for distinguishing gripper designs.

Reading between the lines

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

  • If DCD tracks stress as assumed, the same protocol could be extended to score grasp policies or to serve as an optimization objective for automated gripper design, since the score is differentiable with respect to the deformed point cloud.
  • A natural next validation is to compare SoGraB rankings with instrumented-object measurements on the same grasp trials; if rankings agree, the non-contact method becomes a calibration-free alternative to sensorized benchmarking.
  • SoGraB could be adapted to other deformation-sensitive tasks such as food handling or surgical manipulation, where grasp-induced damage is the primary failure mode rather than dropping the object.
  • The assumption that DCD is density-insensitive could be stress-tested by comparing DCD against a full mesh-based strain measurement on identical grasps; the protocol would need a fallback if occlusions bias the point clouds.
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

4 major / 4 minor

Summary. The paper proposes SoGraB, a benchmarking protocol for soft robotic grippers that scores grasping quality from grasp success, holding time, and object deformation. Deformation is quantified as the Density-Aware Chamfer Distance (DCD) between pre-grasp and post-grasp point clouds of the grasped object, with DCD used as a non-contact proxy for stress. The authors validate the protocol by ranking three Fin-Ray gripper designs and one rigid gripper on 15 objects (12 from EGAD plus 3 custom) at three Shore hardness levels, totaling 900 grasps. They report that the resulting rankings match the qualitative expectation that softer grippers deform soft objects less, and they argue that SoGraB can serve as a standardized benchmark for soft gripper comparison.

Significance. If the approach is valid, SoGraB would provide a practical, low-cost, object-centric benchmark for soft grasping that goes beyond success rate and retention force by capturing deformation. The 900-grasp dataset with associated point clouds is a useful community resource, and the protocol is explicitly designed for common lab hardware (robot arm, 3D printer, depth camera). However, the current manuscript does not establish the central claim that DCD-based deformation is a reliable proxy for grasp safety or stress, and the mathematical justification for the DCD normalization contains an error. The core idea is promising and the dataset is a strength, but the validation needs substantial strengthening before the benchmark can be accepted as proposed.

major comments (4)
  1. [Section III.A, Eq. (2)] The text states that DCD 'bounds the distances in the range [0, 1] by using the first order approximation of the Taylor Expansion (e^z ≈ 1 - ||x-y||2)'. This justification is mathematically incorrect. The expression inside Eq. (2) contains e^{-alpha ||x-ŷ||^2}, and the first-order Taylor approximation of e^{-z} around z=0 is 1 - z, which is not what appears in the formula. Moreover, the first-order approximation does not, by itself, bound the term in [0,1] for large distances; the bound actually follows from the fact that the exponential term lies in (0,1] together with the density normalization. Please correct the derivation and either cite the original DCD paper properly or provide a valid proof of the claimed bound.
  2. [Section IV / Eq. (2)] The DCD sensitivity parameter alpha in Eq. (2) is never specified anywhere in the manuscript. Because the numerical scores depend directly on alpha, this omission makes the reported results impossible to reproduce or to compare with future studies using the same protocol. Please report the exact value (or values) used for alpha, and ideally include a sensitivity analysis showing how the rankings change with alpha.
  3. [Section III.A and Section V] The validation of SoGraB is essentially circular. The paper motivates DCD as a 'non-contact stress proxy' (Section III.A), but then validates the protocol in Section V by showing that gripper rankings match the authors' prior expectation that softer grippers deform soft objects less. That is a consistency check on the metric's own construction, not an external validation. Since all 900 grasps were successful, the score reduces to 1 - dDCD/2, and the benchmark's only discriminating content is DCD. To support the claim that SoGraB ranks grippers by grasp safety, the authors should compare DCD scores against independent measurements of contact force, stress, or damage (e.g., force-torque sensing, instrumented objects, or material failure tests). Without such a comparison, the safety interpretation of the benchmark remains unsupported.
  4. [Section III.A, Eq. (1)] The score formula in Eq. (1) is introduced without justification for its specific functional form: the factor 1/2, the linear combination of DCD, and the t_dropped/t_cycle weighting all appear ad hoc. Because all recorded grasps in Section V were successful, the partially-successful branch is never exercised, and the score collapses to 1 - dDCD/2. While arbitrary weighting is not fatal in a benchmark, the paper should at least discuss the design rationale and the sensitivity of rankings to the chosen weights. As written, the claim that the three features (success, holding time, deformation) are jointly benchmarked is not supported by the experiments.
minor comments (4)
  1. [Section VI] There is a typo: '3D camara' should be '3D camera'. Also, the sentence 'Future users can contribute to the dataset by running the SoGraB protocol by: expanded the range of objects...' contains a grammatical error ('by expanded' should be 'by expanding').
  2. [Section III.B] The ICP alignment procedure would benefit from more detail: what ICP parameters were used (e.g., maximum iterations, convergence tolerance), and how was the initial alignment from robot kinematics obtained? For symmetric objects, the statement that 'centre of mass and principal axes were aligned' is vague; please specify how this was implemented and how failure cases were detected.
  3. [Section IV] The point cloud processing pipeline is not fully described. Please report the segmentation method used to isolate the object from the gripper and background, the typical number of points in the pre- and post-grasp clouds, and whether any downsampling or outlier removal was applied before computing DCD.
  4. [Equation (2)] The notation n_ŷ and n_x̂ is defined only in prose as 'the number of times a point is referenced as a nearest neighbour'. It would be clearer to define these variables directly in the equation and to state whether they are computed before or after the density normalization is applied.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: SoGraB's score is transparently defined from an external metric (DCD), and the reported validation is an empirical consistency check rather than a construction-level equivalence.

full rationale

SoGraB defines grasp quality via Equation (1), which combines grasp success, holding time, and object deformation. Deformation is measured by Density-Aware Chamfer Distance (DCD), an externally published metric [25], with no parameters fitted to the experimental outcomes. The paper does not derive a prediction from the metric and then claim that prediction as independent confirmation; rather, it states physical expectations (e.g., softer grippers should deform soft objects less) and then checks whether the metric reproduces those expectations in controlled experiments. That is a legitimate sensitivity/consistency check, not a circular reduction. The self-citations in the reference list (e.g., [1], [4], [5], [6], [10]) support background claims about soft gripper design or dataset generation and are not load-bearing for the central benchmarking methodology. The main weakness of the paper is that object deformation is asserted as a 'non-contact stress proxy' without independent validation against force, stress, or damage measurements; this is a validity or correctness concern, not a circularity concern, because the proxy assumption is an input to the method rather than a conclusion derived from the method. Similarly, the fact that all 900 grasps were successful, so scores reduce to 1 - dDCD/2, makes the success-rate term non-discriminating in this particular dataset, but this is an experimental limitation, not a definitional equivalence. No step in the paper's derivation chain reduces by construction to its own inputs, and no load-bearing argument depends on an unverified self-citation. Therefore, under the strict circularity criteria, the appropriate finding is no significant circularity.

Assumptions & free parameters 2 free parameters · 4 assumptions · 0 invented entities

The central claim depends on the unvalidated proxy of deformation for stress, a hand-chosen sensitivity parameter, and an approximate point-cloud comparison. None of these are externally benchmarked.

free parameters (2)
  • alpha (DCD sensitivity) = not specified
    The scalar alpha in Equation (2) controls how distances are penalized; no value or fitting procedure is given, and scores depend on it.
  • score formula weights = 1/2, t_dropped/2 t_cycle
    The constants in Equation (1) that set the ranges for successful (>=0.5) and partial (0-0.5) grasps are hand-chosen without justification.
assumptions (4)
  • domain assumption DCD is a suitable metric for comparing incomplete point clouds with occlusions
    The paper relies on the properties of DCD from Wu et al. [25] to measure deformation, without independent validation on this task.
  • ad hoc to paper Object deformation is a valid non-contact stress proxy
    Stated in Section III.A; no experiment connects DCD values to actual stress or damage.
  • ad hoc to paper The first-order Taylor approximation e^-z ~ 1 - z bounds dDCD in [0,1]
    Invoked in Section III.A after Eq. (2); this approximation is unbounded for large z, so the claimed bound is not guaranteed without constraints on alpha.
  • domain assumption ICP alignment approximates the true deformation transform for moderate deformations
    Assumed in Section III.B; fails for large deformations or symmetric objects, which are handled with an alternative method.

how reviews work

0 comments
Cite this review

Pith. "Pith review of SoGraB: A Visual Method for Soft Grasping Benchmarking and Evaluation." pith.science (2026). https://pith.science/paper/J3LORHHY

@misc{pith2026241119408,
  author       = {Pith},
  title        = {Pith review of: SoGraB: A Visual Method for Soft Grasping Benchmarking and Evaluation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/J3LORHHY}},
  note         = {Machine review of arXiv:2411.19408}
}
read the original abstract

Recent years have seen soft robotic grippers gain increasing attention due to their ability to robustly grasp soft and fragile objects. However, a commonly available standardised evaluation protocol has not yet been developed to assess the performance of varying soft robotic gripper designs. This work introduces a novel protocol, the Soft Grasping Benchmarking and Evaluation (SoGraB) method, to evaluate grasping quality, which quantifies object deformation by using the Density-Aware Chamfer Distance (DCD) between point clouds of soft objects before and after grasping. We validated our protocol in extensive experiments, which involved ranking three Fin-Ray gripper designs with a subset of the EGAD object dataset. The protocol appropriately ranked grippers based on object deformation information, validating the method's ability to select soft grippers for complex grasping tasks and benchmark them for comparison against future designs.

Figures

Figures reproduced from arXiv: 2411.19408 by the authors.

Figure 1
Figure 1. The SoGraB method: (Left) A soft, Shore 40A object grasped by [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Demonstration of ICP point cloud alignment (Shore 40A EGAD [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 4
Figure 4. Evaluation objects used in this study: B1-F5 are selected objects [PITH_FULL_IMAGE:figures/full_fig_p004_4.png] view at source ↗
Figures from the paper (3 more)
Figure 5
Figure 5. Figure 5: Heatmaps of the complete grasp evaluation dataset, showing the [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]
Figure 6
Figure 6. Figure 6: Point clouds of selected objects (all Shore 40A), comparing the [PITH_FULL_IMAGE:figures/full_fig_p005_6.png]
Figure 7
Figure 7. Figure 7: Score distributions for select objects, illustrating: Objects with unique rankings (B1 40A, C2 40A); a relatively rigid object (D1); an object with [PITH_FULL_IMAGE:figures/full_fig_p006_7.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

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

  1. SoftVTBench: A Safety-Aware Visuo-Tactile Benchmark for Physically Constrained Robotic Manipulation of Deformable Objects

    cs.RO 2026-07 conditional novelty 6.0 of 10

    Success-only metrics overstate deformable-manipulation performance; tactile sensing raises Safety Success (e.g. 21.4%→35.6% on Object-Soft) while Goal Success stays comparable.

Reference graph

Works this paper leans on

25 extracted references · 24 canonical work pages · cited by 1 Pith paper

  1. [1]

    Hybrid Soft Electrostatic Metamaterial Gripper for Multi- surface, Multi-object Adaptation,

    R. Kanno, P. H. Nguyen, J. Pinskier, D. Howard, S. Song, and M. Kovac, “Hybrid Soft Electrostatic Metamaterial Gripper for Multi- surface, Multi-object Adaptation,” 2024 IEEE 7th International Con- ference on Soft Robotics, RoboSoft 2024 , pp. 851–857, 2024

    work page 2024

  2. [2]

    Soft Robotics for Chemists,

    F. Ilievski, A. D. Mazzeo, R. F. Shepherd, X. Chen, and G. M. Whitesides, “Soft Robotics for Chemists,” Angewandte Chemie , vol. 123, no. 8, pp. 1930–1935, 2011

    work page 1930

  3. [3]

    SoRoForge: End-to-End Soft Actuator Design,

    L. Smith and R. MacCurdy, “SoRoForge: End-to-End Soft Actuator Design,” IEEE Transactions on Automation Science and Engineering , vol. 20, no. 3, pp. 1475–1486, 2023

    work page 2023

  4. [4]

    Towards Bespoke Soft Grip- pers through V oxel-Scale Metamaterial Topology Optimisation,

    J. Pinskier, J. Brett, and D. Howard, “Towards Bespoke Soft Grip- pers through V oxel-Scale Metamaterial Topology Optimisation,” in Robosoft 2024 . San Diego: IEEE, 2024, pp. 1–8

    work page 2024

  5. [5]

    Diversity-Based Topology Optimization of Soft Robotic Grippers,

    J. Pinskier, X. Wang, L. Liow, Y . Xie, P. Kumar, M. Langelaar, and D. Howard, “Diversity-Based Topology Optimization of Soft Robotic Grippers,” Advanced Intelligent Systems , vol. 6, no. 4, 2024

    work page 2024

  6. [6]

    Evolutionary Seeding of Diverse Structural Design Solutions via Topology Optimization,

    Y . Xie, J. Pinskier, X. Wang, and D. Howard, “Evolutionary Seeding of Diverse Structural Design Solutions via Topology Optimization,” ACM Transactions on Evolutionary Learning and Optimization , 2024

    work page 2024

  7. [7]

    The need for reproducible research in soft robotics,

    R. Baines, D. Shah, J. Marvel, J. Case, and A. Spielberg, “The need for reproducible research in soft robotics,” Nature Machine Intelligence , vol. 6, no. 7, pp. 740–741, 2024

    work page 2024

  8. [8]

    Hard questions for soft robotics,

    E. W. Hawkes, C. Majidi, and M. T. Tolley, “Hard questions for soft robotics,” Science Robotics , vol. 6, no. 53, p. eabg6049, 2021

    work page 2021

Show all 25 references
  1. [9]

    Predicting Grasp Success with a Soft Sensing Skin and Shape- Memory Actuated Gripper,

    J. Zimmer, T. Hellebrekers, T. Asfour, C. Majidi, and O. Kroemer, “Predicting Grasp Success with a Soft Sensing Skin and Shape- Memory Actuated Gripper,” IEEE International Conference on Intel- ligent Robots and Systems , pp. 7120–7127, 2019

    2019

  2. [10]

    A comprehensive dataset of grains for granular jamming in soft robotics: Grip strength and shock absorption,

    D. Howard, J. O’Connor, J. Letchford, T. Joseph, S. Lin, S. Baldwin, and G. Delaney, “A comprehensive dataset of grains for granular jamming in soft robotics: Grip strength and shock absorption,” in 2023 IEEE International Conference on Soft Robotics (RoboSoft) , 2023, pp. 1–8

    2023

  3. [11]

    Young and R

    W. Young and R. Budynas, Roark’s F ormulas for Stress and Strain , ser. MacGraw-Hill international edition. McGraw Hill LLC, 2001

    2001

  4. [12]

    Sensorized Reconfigurable Soft Robotic Gripper System for Automated Food Handling,

    J. H. Low, P. M. Khin, Q. Q. Han, H. Yao, Y . S. Teoh, Y . Zeng, S. Li, J. Liu, Z. Liu, P. Valdivia y Alvarado, I.-M. Chen, B. C. K. Tee, and C. H. Yeow, “Sensorized Reconfigurable Soft Robotic Gripper System for Automated Food Handling,” IEEE/ASME Transactions on Mechatronics...

    2021

  5. [13]

    Towards assessing com- pliant robotic grasping from first-object perspective via instrumented objects,

    M. Knopke, L. Zhu, P. Corke, and F. Zhang, “Towards assessing com- pliant robotic grasping from first-object perspective via instrumented objects,” IEEE Robotics and Automation Letters , vol. 9, no. 7, pp. 6320–6327, 2024

    2024

  6. [14]

    Soft Sensorized Physical Twin for Har- vesting Raspberries,

    K. Junge and J. Hughes, “Soft Sensorized Physical Twin for Har- vesting Raspberries,” 2022 IEEE 5th International Conference on Soft Robotics, RoboSoft 2022 , pp. 601–606, 2022

    2022

  7. [15]

    Benchmarking Protocols for Evaluating Grasp Strength, Grasp Cycle Time, Finger Strength, and Finger Repeatability of Robot End-Effectors,

    J. Falco, D. Hemphill, K. Kimble, E. Messina, A. Norton, R. Ro- pelato, and H. Yanco, “Benchmarking Protocols for Evaluating Grasp Strength, Grasp Cycle Time, Finger Strength, and Finger Repeatability of Robot End-Effectors,” IEEE Robotics and Automation Letters , vol. 5, no. ...

    2020

  8. [16]

    Guest Editorial Open Discussion of Robot Grasping Benchmarks, Protocols, and Metrics,

    J. Mahler, R. Platt, A. Rodriguez, M. Ciocarlie, A. Dollar, R. Detry, M. A. Roa, H. Yanco, A. Norton, J. Falco, K. v. Wyk, E. Messina, J. . Leitner, D. Morrison, M. Mason, O. Brock, L. Odhner, A. Kurenkov, M. Matl, and K. Goldberg, “Guest Editorial Open Discussion of Robot Gra...

    2018

  9. [17]

    Characterisation of Grasp Quality Metrics,

    C. Rubert, B. Le ´on, A. Morales, and J. Sancho-Bru, “Characterisation of Grasp Quality Metrics,” Journal of Intelligent and Robotic Systems: Theory and Applications , vol. 89, no. 3-4, pp. 319–342, 2018

    2018

  10. [18]

    Grasp quality measures: review and performance,

    M. A. Roa and R. Surez, “Grasp quality measures: review and performance,” Autonomous Robots , vol. 38, no. 1, pp. 65–88, Jan. 2015, number: 1

    2015

  11. [19]

    On Grasp Quality Mea- sures: Grasp Robustness and Contact Force Distribution in Underactu- ated and Compliant Robotic Hands,

    M. Pozzi, M. Malvezzi, and D. Prattichizzo, “On Grasp Quality Mea- sures: Grasp Robustness and Contact Force Distribution in Underactu- ated and Compliant Robotic Hands,” IEEE Robotics and Automation Letters, vol. 2, no. 1, pp. 329–336, Jan. 2017, number: 1 Conference Name: IE...

    2017

  12. [20]

    A Benchmarking Framework for Systematic Evaluation of Compliant Under-Actuated Soft End Effec- tors in an Industrial Context,

    P. Sotiropoulos, M. A. Roa, M. F. Martins, W. Fried, H. Mnyusiwalla, P. Triantafyllou, and G. Deacon, “A Benchmarking Framework for Systematic Evaluation of Compliant Under-Actuated Soft End Effec- tors in an Industrial Context,” in 2018 IEEE-RAS 18th International Conference ...

    2018

  13. [21]

    MultiGripperGrasp: A Dataset for Robotic Grasping from Parallel Jaw Grippers to Dexterous Hands,

    L. F. C. Murrilo, N. Khargonkar, B. Prabhakaran, and Y . Xi- ang, “MultiGripperGrasp: A Dataset for Robotic Grasping from Parallel Jaw Grippers to Dexterous Hands,” Mar. 2024, issue: arXiv:2403.09841 arXiv:2403.09841 [cs]

    2024 arXiv

  14. [22]

    The YCB object and Model set: Towards common bench- marks for manipulation research,

    B. Calli, A. Singh, A. Walsman, S. Srinivasa, P. Abbeel, and A. M. Dollar, “The YCB object and Model set: Towards common bench- marks for manipulation research,” in 2015 International Conference on Advanced Robotics (ICAR) , 2015, pp. 510–517

    2015

  15. [23]

    Dex-Net 1.0: A cloud-based network of 3D objects for robust grasp planning using a Multi-Armed Bandit model with correlated rewards,

    J. Mahler, F. T. Pokorny, B. Hou, M. Roderick, M. Laskey, M. Aubry, K. Kohlhoff, T. Krger, J. Kuffner, and K. Goldberg, “Dex-Net 1.0: A cloud-based network of 3D objects for robust grasp planning using a Multi-Armed Bandit model with correlated rewards,” in 2016 IEEE Internati...

    2016

  16. [24]

    EGAD! An Evolved Grasping Analysis Dataset for Diversity and Reproducibility in Robotic Ma- nipulation,

    D. Morrison, P. Corke, and J. Leitner, “EGAD! An Evolved Grasping Analysis Dataset for Diversity and Reproducibility in Robotic Ma- nipulation,” IEEE Robotics and Automation Letters , vol. 5, no. 3, pp. 4368–4375, 2020, number: 3

    2020

  17. [25]

    Chamfer Distance as a Comprehensive Metric for Point Cloud Completion,

    T. Wu, L. Pan, J. Zhang, T. W ANG, Z. Liu, and D. Lin, “Chamfer Distance as a Comprehensive Metric for Point Cloud Completion,” in Advances in Neural Information Processing Systems , vol. 34. Curran Associates, Inc., 2021, pp. 29 088–29 100

    2021

Pith tools

Reviewed August 12, 2026 · model on record in the stance chip above.