Pith. sign in

REVIEW 4 major objections 5 minor 36 references

SPLIT: SE(3)-diffusion via Local Geometry-based Score Prediction for 3D Scene-to-Pose-Set Matching Problems

T0 review · 4 major / 5 minor · reviewed 2026-08-12 · deepseek-v4-flash

Pith's one-line read SPLIT predicts a pose's success score from only local scene geometry, letting one diffusion model generate grasp, reorient, and hang poses.

desk verdict SPLIT has a genuinely useful architectural idea for SE(3) diffusion, but the locality claim is not cleanly isolated by the experiments and the evaluation needs tightening. read the letter →

arxiv 2411.10049 v1 pith:NKBZ3U23 submitted 2024-11-15 cs.RO

classification cs.RO
keywords SE(3)-diffusionscore-basedgenerativemodellocalgeometriccontextscene-to-pose-setmatchinggraspgenerationobjectreorientationspatiallocalitytaskconditioning
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

SPLIT argues that a broad class of robot pose-detection problems — deciding where to grasp, where to place, and which way to orient — can be expressed as one problem: matching a scene to the set of successful poses. The paper's core claim is that the success of a pose depends only on the local geometry around it, so the diffusion score that drives pose generation can be predicted from a local feature vector rather than the whole scene. On this basis, a single model with a task-conditioning vector generates multi-purpose poses for mug reorientation and hanging, reaching a 70% task success rate that is 90.3% of the ground-truth-perception rate (77.5%). The upshot is a perception module that replaces task-specific heuristics with a learned relationship between local geometry and task-relevant poses.

What carries the argument

The load-bearing identity is the score decomposition $s(H, X) = \nabla \log p(\mathrm{succ} \mid H_z)$, which follows from Bayes' rule plus the spatial-locality assumption $p(\mathrm{succ} \mid H, X) = p(\mathrm{succ} \mid H_z)$ and a uniform prior over poses. The mechanism that realizes it is a local-geometry encoder: a multi-scale convolutional network produces feature grids, a fixed seven-point kernel is transformed by the query pose to sample features near that pose, and the concatenated local features $H_z$ are fed to two MLPs — one predicting the sample-frame SE(3) score, one evaluating sample quality. The sample-frame parameterization, using the local exponential and logarithmic maps on SE(3), makes the score prediction independent of the absolute pose and lets a single encoder serve every pose sample in a scene.

What would settle it

Construct a task where success requires global information — for example, a mug whose handle is entirely outside the 8 cm local kernel, or a placement that must align with a feature on the far side of the scene — and test whether SPLIT's generated poses achieve chance-level success while a model with global context does not; the paper's own reported failure when the handle is completely invisible is a partial instance.

Watch

Extended reading notes

Core claim

The paper introduces 3D scene-to-pose-set matching and proves, under the spatial-locality assumption $p(\mathrm{succ} | H, X) = p(\mathrm{succ} | H_z)$, that the score function decomposes as $\nabla \log p(H \mid \mathrm{succ}, X) = \nabla \log p(\mathrm{succ} \mid H_z)$, so predicting the gradient of log-success from local context is sufficient for SE(3) pose generation. It parameterizes the score in the tangent space of the sample pose, which removes the dependence on the current pose value. The resulting model, SPLIT, encodes local geometry by transforming a small point kernel with the query pose and interpolating a multi-scale convolutional feature grid, then feeds the local vector to a noise-conditioned score network and a sample-evaluation network, both conditionable by a one-hot task vector. Experiments show it matches or exceeds prior grasp-generation baselines in packed and pile scenes, and one model produces the multiple pose types needed for mug reorientation and hanging with task success close to ground-truth perception.

Load-bearing premise

The assumption that a pose's success probability depends only on the local geometry immediately around that pose — if a successful pose needs information from outside that local neighborhood, such as a hidden handle or a distant constraint, the score decomposition used by SPLIT fails.

Editorial extensions

If this is right

  • A single trained model can output different types of poses (grasp, handle direction, upright placement) by switching a conditioning vector, removing the need for separate perception modules per task.
  • Because the score is predicted from local context and the sample frame, the geometry encoder is shared across all SE(3) samples, giving the architecture a natural translation-equivariance inductive bias.
  • The model can generate grasps directly from a single-view depth image without object segmentation or problem-specific heuristics.
  • The mug task result shows that predicted poses reach 90.3% of the ground-truth-perception success rate, leaving a small relative gap from perception alone.

Reading between the lines

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

  • If spatial locality holds for other manipulation subgoals, the same architecture could be extended to insertion, pushing, or tool use simply by adding task conditioning and collecting corresponding scene-pose datasets.
  • The learned score is literally the gradient of log success probability, so SPLIT's output could double as a cost function for motion optimization, connecting pose-sampling with trajectory planning.
  • A controlled sweep of the point-kernel radius (8 cm in the paper) would quantitatively test how far the locality assumption extends and where global context becomes necessary.
  • The model's ability to infer a hidden handle from priors suggests that training with more partial views or with a small amount of global context could mitigate its reported failure mode.
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 / 5 minor

Summary. This paper introduces a new problem formulation, 3D scene-to-pose-set matching, and proposes SPLIT, an SE(3)-diffusion model that generates pose sets from a scene. The method relies on a spatial locality assumption (Assumption 1, Sec. IV-A), under which the success probability of a pose depends only on the local geometric context, leading to the score decomposition in Eqs. (2)-(5). SPLIT encodes local context with a point kernel applied to multi-scale feature grids (Sec. IV-C) and uses sample-frame Langevin dynamics (Sec. IV-D). Experiments evaluate grasp generation on packed/pile scenes (Sec. VI-A) and a multi-purpose mug reorientation/hanging task (Sec. VI-B), reporting GSR/DR and a 70% task success rate versus 77.5% for ground-truth perception.

Significance. The problem formulation is a useful umbrella for grasp detection, object description, and pose-set matching, and the explicit locality-based score decomposition is a clean derivation that could inspire efficient pose-generation models. The paper's strengths are that the core derivation is transparent, the experiments use external task success rather than self-derived metrics, and the mug task demonstrates a single model serving multiple purposes. However, the central claim that score prediction uses only local geometry is not established by the current implementation, because the multi-scale feature encoder has access to global context. The hidden-handle failure in Sec. VI-B shows that the model relies on global priors in exactly the regime where Assumption 1 should be tested. If the locality claim is revised to 'approximately local with learned global priors,' the contribution becomes more modest but still useful.

major comments (4)
  1. [Sec. IV-C and Sec. V (Training Details)] The implemented geometry encoder is not restricted to local geometry. The 64x64x64 occupancy grid is processed by five downsampling CNN layers, and the feature grids used include coarse scales whose receptive fields span the scene. Since Hz is interpolated from all these scales, a pose query in empty space receives non-zero coarse-scale features, so the network can use global context. This breaks the premise of Eq. (5) in the experiments. Please either enforce locality (e.g., crop a local volume around H, or restrict to fine feature grids) or add an ablation that isolates the contribution of coarse scales and demonstrates that the local part of Hz is sufficient.
  2. [Sec. VI-B, Fig. 5(b)] The hidden-handle failure mode is evidence against Assumption 1. When the handle is completely invisible, the local geometry around the successful grasp/hang poses cannot determine the handle position; the model's occasional success comes from dataset priors about mug shape. The paper counts this as a strength, but it directly contradicts the claim that p(succ|H,X)=p(succ|Hz). Please discuss how this case is compatible with Assumption 1, or explicitly position Eq. (5) as an approximate decomposition with global prior terms.
  3. [Sec. VI-A and Sec. II-C] Diffusion-EDF [4] is described as the most relevant work but is not included in any comparison. Since the paper claims a general formulation and efficiency advantages over existing SE(3)-diffusion pose generation, a quantitative comparison on the mug task (or at least on a common benchmark) is necessary to support the claim. If a comparison is infeasible, please state the reason and provide a qualitative, but concrete, analysis of expected differences.
  4. [Sec. VI-B] The headline result of 70% versus 77.5% task success is based on 40 trials and reported without confidence intervals or statistical tests. At n=40, the standard error is about 7 percentage points, so the observed gap is not significant. Please report confidence intervals (e.g., Wilson intervals) or run more trials, and avoid the phrase '90.3% relative success' without this support.
minor comments (5)
  1. [Sec. I (Introduction)] The word 'featureugh' appears to be a typo for 'feature through' in the sentence 'based on the concatenated featureugh iterative score-based updates.'
  2. [Sec. IV-B] The notation for gradients is inconsistent: Eqs. (2)-(5) use ∇ without specifying whether it is a global or sample-frame gradient, while the subsequent derivation relies on this distinction. Please define the parametrization once and carry it through.
  3. [Sec. V and Sec. VI-B] The scene capture protocol is described as 'single-view depth camera' in Sec. V, but Sec. VI-B states that 'two depth images were captured from different angles.' Please clarify the protocol for the mug task.
  4. [Sec. IV-C] The point kernel consists of seven points at distance d, but the feature grids have different resolutions; it is unclear whether the kernel displacement d is in physical meters or grid cells, and whether d is tuned. Please specify.
  5. [Sec. VI-A, Table I] The SE(3)-Dif baseline omits the signed distance prediction loss, which is a modification of the original method; the effect of this modification on the comparison should be discussed.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: the score decomposition is an explicitly stated modeling assumption, and all results are validated against external task-success benchmarks.

full rationale

The paper's central derivation (Eqs. 2-5) is a direct consequence of Assumption 1, which states that success probability depends only on the local geometric context Hz. This is an explicitly labeled assumption, not a hidden result: the paper never claims to prove locality from data, and the score decomposition merely restates the assumption in the language of score functions. The validity of Assumption 1 is a modeling risk, and the paper's own failure case (invisible mug handle) and the multi-scale feature grids (which can carry global information) are relevant to whether the assumption holds, but they are not circularity. The method is evaluated against external, non-fitted benchmarks: grasping success is compared with VGN, GIGA, and SE(3)-Diff on the same scene-generation protocol, and mug reorientation/hanging is measured as physical task success relative to a ground-truth-perception baseline (77.5% vs. 70%). No fitted constants or self-derived benchmarks are used to validate the model, and there are no load-bearing self-citations. The only 'local geometry suffices' claim is an input assumption of the formulation, transparently stated as Assumption 1, so the derivation chain is self-contained relative to its assumptions.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The method's central claim rests on the spatial locality assumption and a handful of hand-set hyperparameters. No new entities are postulated.

free parameters (3)
  • Point kernel displacement d = 0.08 m
    Defines the radius of local geometric context used for score prediction; set to gripper width. If the true task-relevant context is larger, the locality assumption fails.
  • Noise schedule range = sigma from 0.02 to 0.1
    Hand-selected noise levels for the SE(3) diffusion; affects convergence and sample diversity.
  • Negative-sample time threshold t_th = 0.3
    Decides when noisy samples become negative examples for the evaluation network; chosen without sensitivity analysis.
assumptions (4)
  • domain assumption Spatial locality: p(succ|H,X) = p(succ|Hz) for a learned local feature Hz.
    Assumption 1 in Sec. IV-A. The entire score decomposition (Eq. 5) and the architecture depend on this. It is not proven; the paper provides empirical evidence only.
  • domain assumption Consistent local context yields consistent target poses.
    Assumption 2 in Sec. IV-A, needed so a single encoder can share features across the scene.
  • domain assumption Uniform prior over pose candidates p(H|X).
    Used to drop the p(H|X) term in Eq. (3). The paper treats pose candidates as drawn uniformly; in practice sampling is uniform in the workspace, but the target pose distribution is not uniform.
  • standard math Standard SE(3) Lie group calculus and automatic differentiation for gradients.
    Sec. III-B; background from [30].

how reviews work

0 comments
Cite this review

Pith. "Pith review of SPLIT: SE(3)-diffusion via Local Geometry-based Score Prediction for 3D Scene-to-Pose-Set Matching Problems." pith.science (2026). https://pith.science/paper/NKBZ3U23

@misc{pith2026241110049,
  author       = {Pith},
  title        = {Pith review of: SPLIT: SE(3)-diffusion via Local Geometry-based Score Prediction for 3D Scene-to-Pose-Set Matching Problems},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NKBZ3U23}},
  note         = {Machine review of arXiv:2411.10049}
}
read the original abstract

To enable versatile robot manipulation, robots must detect task-relevant poses for different purposes from raw scenes. Currently, many perception algorithms are designed for specific purposes, which limits the flexibility of the perception module. We present a general problem formulation called 3D scene-to-pose-set matching, which directly matches the corresponding poses from the scene without relying on task-specific heuristics. To address this, we introduce SPLIT, an SE(3)-diffusion model for generating pose samples from a scene. The model's efficiency comes from predicting scores based on local geometry with respect to the sample pose. Moreover, leveraging the conditioned generation capability of diffusion models, we demonstrate that SPLIT can generate the multi-purpose poses, required to complete both the mug reorientation and hanging manipulation within a single model.

Figures

Figures reproduced from arXiv: 2411.10049 by the authors.

Figure 1
Figure 1. Multiple pose descriptors and grasp detections are required to [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. A visual explanation of the spatial locality assumption of our [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The network architecture of SPLIT. To extract the local geometric [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Packed and pile scenes from [14], along with grasp generation [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 5
Figure 5. Figure 5: Two real-world examples of multi-purpose pose detection with the [PITH_FULL_IMAGE:figures/full_fig_p006_5.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references · 17 canonical work pages

  1. [4]

    Diffusion-edfs: Bi-equivariant denoising genera- tive modeling on se(3) for visual robotic manipulation,

    H. Ryu, J. Kim, J. Chang, H. S. Ahn, J. Seo, T. Kim, J. Choi, and R. Horowitz, “Diffusion-edfs: Bi-equivariant denoising genera- tive modeling on se(3) for visual robotic manipulation,” ArXiv, vol. abs/2309.02685, 2023

  2. [1]

    Bundlesdf: Neural 6-dof tracking and 3d reconstruction of unknown objects,

    B. Wen, J. Tremblay, V . Blukis, S. Tyree, T. Muller, A. Evans, D. Fox, J. Kautz, and S. Birchfield, “Bundlesdf: Neural 6-dof tracking and 3d reconstruction of unknown objects,” 2023 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pp. 606–617, 2023

  3. [2]

    Deep learning-based object pose estimation: A comprehensive survey,

    J. Liu, W. Sun, H. Yang, Z. Zeng, C. Liu, J. Zheng, X. Liu, H. Rahmani, N. Sebe, and A. Mian, “Deep learning-based object pose estimation: A comprehensive survey,” ArXiv, vol. abs/2405.07801, 2024

  4. [3]

    Neural descriptor fields: Se(3)- equivariant object representations for manipulation,

    A. Simeonov, Y . Du, A. Tagliasacchi, J. B. Tenenbaum, A. Ro- driguez, P. Agrawal, and V . Sitzmann, “Neural descriptor fields: Se(3)- equivariant object representations for manipulation,” 2022 Interna- tional Conference on Robotics and Automation (ICRA) , pp. 6394– 6400, 2021

  5. [5]

    Deep learning approaches to grasp synthesis: A review,

    R. Newbury, M. Gu, L. Chumbley, A. Mousavian, C. Eppner, J. Leit- ner, J. Bohg, A. Morales, T. Asfour, D. Kragic et al., “Deep learning approaches to grasp synthesis: A review,” IEEE Transactions on Robotics, 2023

  6. [6]

    Grasp learning: Models, methods, and performance,

    R. Platt, “Grasp learning: Models, methods, and performance,” Annual Review of Control, Robotics, and Autonomous Systems , vol. 6, pp. 363–389, 2023

  7. [7]

    Se(3)-diffusionfields: Learning smooth cost functions for joint grasp and motion opti- mization through diffusion,

    J. Urain, N. Funk, J. Peters, and G. Chalvatzaki, “Se(3)-diffusionfields: Learning smooth cost functions for joint grasp and motion opti- mization through diffusion,” 2023 IEEE International Conference on Robotics and Automation (ICRA) , pp. 5923–5930, 2022

  8. [8]

    Convolutional occupancy networks,

    S. Peng, M. Niemeyer, L. M. Mescheder, M. Pollefeys, and A. Geiger, “Convolutional occupancy networks,” ArXiv, vol. abs/2003.04618, 2020

Show all 36 references
  1. [9]

    Implicit functions in feature space for 3d shape reconstruction and completion,

    J. Chibane, T. Alldieck, and G. Pons-Moll, “Implicit functions in feature space for 3d shape reconstruction and completion,” in IEEE Conference on Computer Vision and Pattern Recognition (CVPR) . IEEE, jun 2020

  2. [10]

    Robotic grasping and contact: A review,

    A. Bicchi and V . Kumar, “Robotic grasping and contact: A review,” in Proceedings 2000 ICRA. Millennium conference. IEEE international conference on robotics and automation. Symposia proceedings (Cat. No. 00CH37065), vol. 1. IEEE, 2000, pp. 348–353

  3. [11]

    Data-driven grasp synthesis—a survey,

    J. Bohg, A. Morales, T. Asfour, and D. Kragic, “Data-driven grasp synthesis—a survey,” IEEE Transactions on robotics , vol. 30, no. 2, pp. 289–309, 2013

  4. [12]

    R. M. Murray, Z. Li, and S. S. Sastry, A mathematical introduction to robotic manipulation. CRC press, 2017

  5. [13]

    Graspit! a versatile simulator for robotic grasping,

    A. T. Miller and P. K. Allen, “Graspit! a versatile simulator for robotic grasping,” IEEE Robotics & Automation Magazine , vol. 11, pp. 110– 122, 2004

  6. [14]

    V olumetric grasping network: Real-time 6 dof grasp detection in clutter,

    M. Breyer, J. J. Chung, L. Ott, R. Siegwart, and J. Nieto, “V olumetric grasping network: Real-time 6 dof grasp detection in clutter,” in Conference on Robot Learning . PMLR, 2021, pp. 1602–1611

  7. [15]

    Contact- graspnet: Efficient 6-dof grasp generation in cluttered scenes,

    M. Sundermeyer, A. Mousavian, R. Triebel, and D. Fox, “Contact- graspnet: Efficient 6-dof grasp generation in cluttered scenes,” in 2021 IEEE International Conference on Robotics and Automation (ICRA) . IEEE, 2021, pp. 13 438–13 444

  8. [16]

    Rgb matters: Learning 7-dof grasp poses on monocular rgbd images,

    M. Gou, H.-S. Fang, Z. Zhu, S. Xu, C. Wang, and C. Lu, “Rgb matters: Learning 7-dof grasp poses on monocular rgbd images,” in 2021 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2021, pp. 13 459–13 466

  9. [17]

    Pointnet++ grasping: Learning an end-to-end spatial grasp generation algorithm from sparse point clouds,

    P. Ni, W. Zhang, X. Zhu, and Q. Cao, “Pointnet++ grasping: Learning an end-to-end spatial grasp generation algorithm from sparse point clouds,” in 2020 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2020, pp. 3619–3625

  10. [18]

    6-dof graspnet: Variational grasp generation for object manipulation,

    A. Mousavian, C. Eppner, and D. Fox, “6-dof graspnet: Variational grasp generation for object manipulation,” 2019 IEEE/CVF Interna- tional Conference on Computer Vision (ICCV) , pp. 2901–2910, 2019

  11. [19]

    kpam: Keypoint affordances for category-level robotic manipulation,

    L. Manuelli, W. Gao, P. R. Florence, and R. Tedrake, “kpam: Keypoint affordances for category-level robotic manipulation,” in International Symposium of Robotics Research , 2019

  12. [20]

    kpam 2.0: Feedback control for category- level robotic manipulation,

    W. Gao and R. Tedrake, “kpam 2.0: Feedback control for category- level robotic manipulation,” IEEE Robotics and Automation Letters , vol. 6, pp. 2962–2969, 2021

  13. [21]

    Dense object nets: Learn- ing dense visual object descriptors by and for robotic manipulation,

    P. R. Florence, L. Manuelli, and R. Tedrake, “Dense object nets: Learn- ing dense visual object descriptors by and for robotic manipulation,” ArXiv, vol. abs/1806.08756, 2018

  14. [22]

    Equivariant descriptor fields: Se(3)-equivariant energy-based models for end-to-end visual robotic manipulation learning,

    H. Ryu, J.-H. Lee, H. Lee, and J. Choi, “Equivariant descriptor fields: Se(3)-equivariant energy-based models for end-to-end visual robotic manipulation learning,” ArXiv, vol. abs/2206.08321, 2022

  15. [23]

    Denoising diffusion probabilistic models,

    J. Ho, A. Jain, and P. Abbeel, “Denoising diffusion probabilistic models,” ArXiv, vol. abs/2006.11239, 2020

  16. [24]

    Generative modeling by estimating gradients of the data distribution,

    Y . Song and S. Ermon, “Generative modeling by estimating gradients of the data distribution,” in Neural Information Processing Systems , 2019

  17. [25]

    Constrained 6-dof grasp generation on complex shapes for improved dual-arm manipulation,

    G. Singh, S. Kalwar, M. F. Karim, B. Sen, N. Govindan, S. Srid- har, and K. M. Krishna, “Constrained 6-dof grasp generation on complex shapes for improved dual-arm manipulation,” ArXiv, vol. abs/2404.04643, 2024

  18. [26]

    Score-based generative modeling through stochastic differential equations,

    Y . Song, J. N. Sohl-Dickstein, D. P. Kingma, A. Kumar, S. Ermon, and B. Poole, “Score-based generative modeling through stochastic differential equations,” ArXiv, vol. abs/2011.13456, 2020

  19. [27]

    Correlation functions and computer simulations,

    G. Parisi, “Correlation functions and computer simulations,” Nuclear Physics B, vol. 180, no. 3, pp. 378–384, 1981

  20. [28]

    Mcmc using hamiltonian dynamics,

    R. M. Neal, “Mcmc using hamiltonian dynamics,” arXiv preprint arXiv:1206.1901, 2012

  21. [29]

    A connection between score matching and denoising autoencoders,

    P. Vincent, “A connection between score matching and denoising autoencoders,” Neural Computation, vol. 23, pp. 1661–1674, 2011

  22. [30]

    A micro lie theory for state estimation in robotics,

    J. Sol `a, J. Deray, and D. Atchuthan, “A micro lie theory for state estimation in robotics,” ArXiv, vol. abs/1812.01537, 2018

  23. [31]

    Film: Visual reasoning with a general conditioning layer,

    E. Perez, F. Strub, H. de Vries, V . Dumoulin, and A. C. Courville, “Film: Visual reasoning with a general conditioning layer,” in AAAI Conference on Artificial Intelligence , 2017

  24. [32]

    Fourier features let networks learn high frequency functions in low dimensional domains,

    M. Tancik, P. P. Srinivasan, B. Mildenhall, S. Fridovich-Keil, N. Raghavan, U. Singhal, R. Ramamoorthi, J. T. Barron, and R. Ng, “Fourier features let networks learn high frequency functions in low dimensional domains,” ArXiv, vol. abs/2006.10739, 2020

  25. [33]

    A billion ways to grasp: An evaluation of grasp sampling schemes on a dense, physics-based grasp data set,

    C. Eppner, A. Mousavian, and D. Fox, “A billion ways to grasp: An evaluation of grasp sampling schemes on a dense, physics-based grasp data set,” in International Symposium of Robotics Research , 2019

  26. [34]

    Synergies between affordance and geometry: 6-dof grasp detection via implicit representations,

    Z. Jiang, Y . Zhu, M. Svetlik, K. Fang, and Y . Zhu, “Synergies between affordance and geometry: 6-dof grasp detection via implicit representations,” ArXiv, vol. abs/2104.01542, 2021

  27. [35]

    Shapenet: An information-rich 3d model repository,

    A. X. Chang, T. A. Funkhouser, L. J. Guibas, P. Hanrahan, Q.- X. Huang, Z. Li, S. Savarese, M. Savva, S. Song, H. Su, J. Xiao, L. Yi, and F. Yu, “Shapenet: An information-rich 3d model repository,” ArXiv, vol. abs/1512.03012, 2015

  28. [36]

    Dex-net 2.0: Deep learning to plan robust grasps with synthetic point clouds and analytic grasp metrics,

    J. Mahler, J. Liang, S. Niyaz, M. Laskey, R. Doan, X. Liu, J. A. Ojea, and K. Goldberg, “Dex-net 2.0: Deep learning to plan robust grasps with synthetic point clouds and analytic grasp metrics,” arXiv preprint arXiv:1703.09312, 2017

Pith tools

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