Pith. sign in

REVIEW 5 major objections 5 minor 37 references

NeuralSVCD claims swept-volume collision detection that is 24x faster than sphere-based GPU methods and more than 100x faster than mesh-based GJK, while improving accuracy by 5-15% on unseen shapes.

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 ↗

A neural encoder-decoder with sphere-based broad-phase filtering performs swept-volume collision detection continuously along trajectories, beating baselines in accuracy and speed on manipulation benchmarks.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection NeuralSVCD is a genuinely new neural SVCD approach with strong empirical wins; the t† assumption is the right thing to poke at, and the planning comparison needs controls. the 5 major comments →

arxiv 2509.00499 v1 pith:226STIS3 submitted 2025-08-30 cs.RO cs.AIcs.LG

NeuralSVCD for Efficient Swept Volume Collision Detection

Dongwon Son , Hojin Jung , Beomjoon Kim This is my paper
classification cs.RO cs.AIcs.LG
keywords swept volume collision detectioncontinuous collision checkingneural collision predictiondistributed latent representationshape localitytemporal localityGPU-accelerated motion planningtrajectory optimization
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.

The reading

This paper tries to establish that continuous collision checking along an entire robot trajectory—known as swept-volume collision detection—can be made both faster and more accurate than current geometric methods by replacing global shape processing with a neural network trained on local shape patches and local trajectory segments. It proposes NeuralSVCD, which first uses bounding spheres to quickly locate the instant of closest approach for each pair of surface patches, then feeds a small neural decoder only the local latent descriptors and a locally linearized version of the motion around that instant. The reported results are 90% accuracy in 64 microseconds per query, peaking at 95.6% on in-domain shapes and 94.9% on unseen shapes, with speedups of 24x over sphere-based GPU methods and more than 100x over mesh-based GJK methods. If correct, this would make continuous collision checking practical inside GPU-accelerated motion planners and remove the tunneling errors of discrete waypoint sampling without the cost of exact swept-volume reconstruction.

Core claim

The paper's central claim is that collision events are local in two senses—local in surface geometry and local in time—and that an architecture built around these two localities can beat the accuracy-efficiency trade-off that dominates existing SVCD methods. Each object mesh is encoded into a distributed representation: N representative surface points, each carrying a latent vector for its local patch and a bounding sphere radius. During inference, a broad phase solves a sphere-swept-volume minimization to find candidate patch pairs and a pseudo-critical collision time t†, the instant of maximum sphere overlap. A narrow phase then linearizes the trajectory at t† with a first-order Taylor exp

What carries the argument

The carrying object is the distributed latent representation Z = {(p_i, z_i, r_i)}, where p_i is a representative surface point, z_i is a learned local latent vector, and r_i is a bounding-sphere radius derived from the distance to the nearest neighboring representative point. It enables a two-phase pipeline: a broad-phase sphere overlap optimization that returns candidate pairs and the pseudo-critical time t†, and a narrow-phase neural decoder fSVCD that checks collisions using only local latent vectors and the locally linearized trajectory at t†. The decoder's input preprocessing is constructed to be invariant to global SE(3) transformations and uniform scaling, which lets the same local m

Load-bearing premise

The broad phase finds a pseudo-critical collision time t† from coarse bounding spheres, and the whole pipeline assumes t† is close enough to the true critical time t* that the locally linearized trajectory segment fed to the decoder still contains the real collision event.

What would settle it

Evaluate the decoder densely in time along a set of trajectories and compare the time at which its collision probability peaks to the broad-phase t†. If, for thin or fast-rotating shapes, the peak probability occurs at a time far from t†—by more than a small fraction of the trajectory—the assumption that t† approximates t* fails and collisions can be missed.

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

If this is right

  • Swept-volume collision checking becomes a GPU-parallel tensor pipeline: sphere minimizations plus MLP evaluations, so per-query latency drops to tens of microseconds in the reported settings.
  • Motion planners can use continuous trajectory segments as collision cost rather than sparse waypoints, which should reduce tunneling failures in tight-tolerance tasks.
  • Unseen shapes generalize without retraining because the decoder sees local patches rather than global geometry; reported out-of-domain accuracy stays within 0.7% of in-domain peak accuracy.
  • The method's logits are proxy penetration distances, so the same architecture can serve as a differentiable collision cost in trajectory optimization, with MPPI and L-BFGS used to avoid local minima.
  • Compared with increasing sphere counts in sphere-based planners, NeuralSVCD claims comparable success with up to 6.2x less computation time across the three tested tasks.

Where Pith is reading between the lines

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

  • A direct stress test: query the decoder densely in time and compare its argmax time to t† from the sphere broad phase; if fast rotations or long thin objects create large gaps, the local linearization at t† may miss the true collision instant.
  • The same local-patch representation could in principle extend to deformable objects, but the equivariant rotation operator D(R) and rigid-twist input assume rigid-body motion, so that extension would need a new deformation-aware encoding.
  • Because training trajectories are restricted to linear paths in R6, the model's temporal locality has only been tested on straight-line segments; curved and multi-segment trajectories would be a natural out-of-distribution probe.
  • The accuracy gain over global-representation variants (95.6% vs 81% in-domain) suggests that distribution over local patches, not more data, is what buys generalization; a cheap experiment is to vary N while holding data fixed and observe the accuracy curve.
Share X Bluesky LinkedIn Reddit HN

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

5 major / 5 minor

Summary. NeuralSVCD proposes a neural encoder–decoder pipeline for swept-volume collision detection. Each mesh is represented by N sampled surface points with local latent codes and bounding-sphere radii. At inference, a broad phase finds candidate static–moving point pairs and the time t† that minimizes sphere distance along the trajectory; the narrow phase feeds a locally linearized trajectory segment to an MLP decoder, and max-pooling aggregates pair logits. Training uses binary cross-entropy against FCL ground-truth labels with a gradient regularization term. Experiments compare against convex-cell/GJK, sphere/mesh, implicit-function, and discrete/global variants on UR5, ARMADA, and mobile-UR5 tasks, reporting accuracy/runtime trade-offs and motion-planning success rates.

Significance. The paper targets a practical problem and has a clear, plausible architecture. The broad/narrow two-step design with temporal locality is a sensible way to make neural collision prediction continuous, and if the reported gains hold it would be a useful step over sphere- and hull-based SVCD. Strengths include public code and videos, held-out evaluation with in-domain/out-of-domain objects, and integration into a standard optimizer. However, the main accuracy claim rests on an unverified approximation about the pseudo-critical time; the motion-planning comparison varies optimizer hyperparameters across methods; and no uncertainty quantification is provided. These issues are fixable and do not invalidate the approach, but they must be addressed before the paper's claims are fully supported.

major comments (5)
  1. [§3.2, Eq. (1)] The continuous-accuracy claim in §4.1 (90% at 6.4e-5 s, peak 95.6%) is load-bearing on the assumption that t† ≈ t*. The broad phase minimizes the enclosing-sphere distance with Newton's method, but no initialization or multi-start is described; for spline trajectories with multiple local extrema the optimizer can return a time with no actual mesh contact. Moreover, even the global minimum of sphere separation is not generally the time of maximum mesh penetration for non-convex/thin geometry. The discrete-variant ablation does not isolate t† error. Please report (a) recall of the broad-phase candidate set against ground truth over a dense time grid, and (b) the distribution of |t† - argmax_t f_decoder(t)| on test cases. Without this, the speed/accuracy advantage over continuous GJK/sphere baselines is not established.
  2. [§3.1] The radius r_i = α min_j dist(p_i, p_j) is asserted to enclose the local geometry near p_i. For concave or elongated Voronoi cells the nearest-representative distance is not a bound on the extent of the local patch, so a truly colliding pair can be absent from the broad-phase candidate set and never reach the decoder; this is a false-negative mechanism distinct from t† error. Please provide a formal condition or a conservative radius computation, and measure broad-phase recall against FCL on the evaluation set.
  3. [§4.2, Tables 3–5] The claim that cuRobo-NeuralSVCD outperforms cuRobo-sphere with the 'same planner but different collision detectors' is weakened by the fact that optimizer hyperparameters differ across methods within each task: MPPI interpolation 4 vs 100, number of samples 200 vs 50, LBFGS interpolation 2 vs 100/50, control points, and activation distances all vary. Differences in success rate and planning time may be attributable to optimizer tuning rather than the collision detector. Please either keep optimizer hyperparameters fixed across methods, sweep them for all methods, or demonstrate that the conclusion is stable over a range of settings.
  4. [§4.1, Table 1] All accuracy, runtime, and planning results are single point estimates. No standard deviations, confidence intervals, or significance tests are reported; claims of 'consistently outperforms' and speed-ups (120x–590x, 24x) have no uncertainty attached. Please run multiple seeds (at least for network training and optimizer randomness) and report error bars or paired tests.
  5. [§6.1] The paper acknowledges that using logits as collision distance is an approximation that can lead optimizers to local minima, but it does not quantify the mismatch. Since the planner cost C_SVCD is built from these logits, the planner results depend on this surrogate. Please add an evaluation of logit quality against penetration depth (e.g., correlation or calibration on the SVCD test set) or compare against a distance-based version of the same collision cost.
minor comments (5)
  1. [Appendix A] There is a garbled duplicated sentence: 'The training objective is defined as ... However, binary classification loss alone results ... This is undesirable because The training objective is defined as ...' Please rewrite this paragraph.
  2. [Table 2] The notation U(min, max) with '⊂ R' or '⊂ Z' is confusing; if U denotes a uniform distribution, define it explicitly and distinguish the sampling domain from the sampled values.
  3. [Algorithm 1] The variable 'ccd output' should be 'SVCD output' for consistency, and the broadPhase function's output is not fully specified with respect to Eq. (1).
  4. [§3.2] Minor formatting: 'mesh mov and meshstatic' lacks subscript formatting; also the subscript on t† is sometimes omitted, making it hard to distinguish pair-specific times.
  5. [Figure 5] The axes are not labeled clearly, and the text alternates between 'success rate' and 'accuracy'. Please use consistent terminology and label axes.

Circularity Check

0 steps flagged

No significant circularity: predictions are trained and evaluated against external FCL ground truth, and the central claims rest on held-out empirical comparison rather than by-construction equivalence.

full rationale

The paper's core SVCD prediction is not defined in terms of its own output. Training uses ground-truth labels ySVCD computed with the external FCL library [24]: 'Ground-truth collision labels ySVCD are computed using the FCL library [24].' The reported accuracy numbers (90% at 6.4e-5 s, peaking at 95.6% in-domain and 94.9% out-of-domain) are then measured on held-out problem instances against the same external ground-truth standard, so the headline accuracy is an empirical result, not a forced consequence of the model's construction. The encoder architecture is taken from the authors' prior FER-VN-OccNet paper [33] and the local-crop collision idea from their prior collision-network paper [5], but these are published architectural and predecessor works with independent content; they are not invoked as a 'uniqueness theorem' forbidding alternatives, and the paper further benchmarks against sphere-based, mesh/GJK, implicit-function, and discrete- and global-representation ablated baselines. The broad-phase pseudo-critical time t† is introduced as an approximation: 'We assume that the pseudo-critical collision time t† sufficiently approximates the true critical collision time t*' (Section 3.2). That is a stated approximation assumption, not a self-referential definition; the same holds for the unproved sphere-enclosure radius choice ri = α min_j dist(pi,pj). These are correctness and robustness risks, not circularity. No fitted parameter is renamed as a prediction, and no equation reduces to its own input by construction. Therefore, under the hard rules, no circular step can be exhibited, and the honest finding is a score of 0.

Axiom & Free-Parameter Ledger

3 free parameters · 5 axioms · 0 invented entities

The central claim rests on a handful of modeling assumptions, most critically that the sphere-based broad-phase finds a good approximation to the true collision time and that neural logits act as reliable distance proxies. No entirely new physical or mathematical entities are introduced.

free parameters (3)
  • N (number of representative points per mesh)
    Number of local latent samples per object; not specified in the paper, but controls the granularity of shape locality and the candidate pair count.
  • alpha (sphere radius multiplier)
    Bounding sphere radius r_i = alpha * min neighbor distance, with alpha > 1. No value or sensitivity analysis is reported.
  • Trajectory interpolation density in neural decoder
    The local trajectory is linearized at t†; the interpolation number is a hyperparameter in Tables 3-5, set to 100 for NeuralSVCD but differently for baselines.
axioms (5)
  • domain assumption FCL collision labels ySVCD are correct for training and evaluation.
    Used as ground truth in Section 3.3 and Appendix B; all reported accuracy is measured against these labels.
  • ad hoc to paper Pseudo-critical time t† approximates true critical time t*.
    Stated in Section 3.2; load-bearing for the narrow-phase because the decoder only sees the local trajectory around t†.
  • ad hoc to paper Logits from fSVCD can serve as collision distance for optimization.
    Acknowledged in Section 6.1; the paper uses inverse sigmoid of probabilities as a surrogate for penetration depth in the planner.
  • ad hoc to paper Local linearization around t† captures the relevant swept volume segment.
    Section 3.2 relies on this to restrict decoder input to a first-order Taylor expansion of the trajectory.
  • domain assumption Encoder FER-VN-OccNet and latent rotation D(R) from [33] preserve rigid-body properties.
    Adopted from prior work, not re-derived; correctness is assumed for the latent transformation operation.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of NeuralSVCD for Efficient Swept Volume Collision Detection." pith.science (2026). https://pith.science/paper/226STIS3

@misc{pith2026250900499,
  author       = {Pith},
  title        = {Pith review of: NeuralSVCD for Efficient Swept Volume Collision Detection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/226STIS3}},
  note         = {Machine review of arXiv:2509.00499}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Robot manipulation in unstructured environments requires efficient and reliable Swept Volume Collision Detection (SVCD) for safe motion planning. Traditional discrete methods potentially miss collisions between these points, whereas SVCD continuously checks for collisions along the entire trajectory. Existing SVCD methods typically face a trade-off between efficiency and accuracy, limiting practical use. In this paper, we introduce NeuralSVCD, a novel neural encoder-decoder architecture tailored to overcome this trade-off. Our approach leverages shape locality and temporal locality through distributed geometric representations and temporal optimization. This enhances computational efficiency without sacrificing accuracy. Comprehensive experiments show that NeuralSVCD consistently outperforms existing state-of-the-art SVCD methods in terms of both collision detection accuracy and computational efficiency, demonstrating its robust applicability across diverse robotic manipulation scenarios. Code and videos are available at https://neuralsvcd.github.io/.

Figures

Figures reproduced from arXiv: 2509.00499 by Beomjoon Kim, Dongwon Son, Hojin Jung.

Figure 1
Figure 1. Figure 1: (a) Illustration of tunneling errors in discrete collision detection. Sampling only a finite [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: (a) SVCD problem definition. (b-d) Comparison of geometric representations used in [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: (Left - shape locality) Two different pairs of objects have completely different global shapes, but when we focus on the circled contact regions, they have very similar shapes. (Right - temporal locality) Two different object trajecto￾ries share the same collision moment, marked by rectangles. To overcome these limitations, we leverage two essential insights: shape locality and temporal locality. Shape loc… view at source ↗
Figure 4
Figure 4. Figure 4: Overview of SVCD pipeline. During encoding, (top), the canonical meshes (mesh [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Accuracy efficiency tradeoff graph of our SVCD and other baselines for (Left) in-domain [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: This figure illustrates the problem case for checking swept volume collision accu￾racy. The objective of each problem is to deter￾mine where a moving robot, starting from the ini￾tial configuration (marked in green) and moving to the goal configuration (marked in red), will col￾lide with a static object (marked in yellow). To quantify the accuracy of the SVCD, we per￾form a comparative experiment. We use 9… view at source ↗
Figure 7
Figure 7. Figure 7: Illustration of three robotic tasks solved by our proposed algorithm. (a) Dish insertion: A [PITH_FULL_IMAGE:figures/full_fig_p013_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Illustration of pre-processing for achieving invariance. Case 1 and 2 have the same relative [PITH_FULL_IMAGE:figures/full_fig_p014_8.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

37 extracted references · 34 canonical work pages · 4 internal anchors

  1. [1]

    C. Ericson. Real-Time Collision Detection. CRC Press, 2004

    work page 2004

  2. [2]

    P. G. Xavier. Fast swept-volume distance for robust collision detection. In Proceedings of International Conference on Robotics and Automation , volume 2, pages 1162–1169. IEEE, 1997

    work page 1997

  3. [3]

    Schulman, J

    J. Schulman, J. Ho, A. X. Lee, I. Awwal, H. Bradlow, and P. Abbeel. Finding locally opti- mal, collision-free trajectories with sequential convex optimization. In Robotics: science and systems, volume 9, pages 1–10. Berlin, Germany, 2013

    work page 2013

  4. [4]

    E. G. Gilbert, D. W. Johnson, and S. S. Keerthi. A fast procedure for computing the distance between complex objects in three-dimensional space. IEEE Journal on Robotics and Automa- tion, 4(2):193–203, 2002

    work page 2002

  5. [5]

    Local object crop collision network for efficient simulation of non-convex objects in GPU-based simulators

    D. Son and B. Kim. Local object crop collision network for efficient simulation of non-convex objects in gpu-based simulators. arXiv preprint arXiv:2304.09439, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  6. [6]

    Sundaralingam, S

    B. Sundaralingam, S. K. S. Hari, A. Fishman, C. Garrett, K. Van Wyk, V . Blukis, A. Millane, H. Oleynikova, A. Handa, F. Ramos, et al. Curobo: Parallelized collision-free robot motion 9 generation. In 2023 IEEE International Conference on Robotics and Automation (ICRA), pages 8112–8119. IEEE, 2023

    work page 2023

  7. [7]

    C. W. Ramsey, Z. Kingston, W. Thomason, and L. E. Kavraki. Collision-affording point trees: Simd-amenable nearest neighbors for fast collision checking. arXiv preprint arXiv:2406.02807, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  8. [8]

    S. Sui, L. Sentis, and A. Bylard. Hardware-accelerated ray tracing for discrete and continuous collision detection on gpus. arXiv preprint arXiv:2409.09918, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  9. [9]

    Downs, A

    L. Downs, A. Francis, N. Koenig, B. Kinman, R. Hickman, K. Reymann, T. B. McHugh, and V . Vanhoucke. Google scanned objects: A high-quality dataset of 3d scanned household items. In 2022 International Conference on Robotics and Automation (ICRA) , pages 2553–

    work page 2022

  10. [10]

    Deitke, R

    M. Deitke, R. Liu, M. Wallingford, H. Ngo, O. Michel, A. Kusupati, A. Fan, C. Laforte, V . V oleti, S. Y . Gadre, et al. Objaverse-xl: A universe of 10m+ 3d objects.Advances in Neural Information Processing Systems, 36:35799–35813, 2023

    work page 2023

  11. [11]

    J. Canny. Collision detection for moving polyhedra. IEEE Transactions on Pattern Analysis and Machine Intelligence, (2):200–209, 1986

    work page 1986

  12. [12]

    Kim and J

    B. Kim and J. Rossignac. Collision prediction for polyhedra under screw motions. In Proceed- ings of the eighth ACM symposium on Solid modeling and applications , pages 4–10, 2003

    work page 2003

  13. [13]

    Redon, A

    S. Redon, A. Kheddar, and S. Coquillart. An algebraic solution to the problem of collision detection for rigid polyhedral objects. In Proceedings 2000 ICRA. Millennium Conference. IEEE International Conference on Robotics and Automation. Symposia Proceedings (Cat. No. 00CH37065), volume 4, pages 3733–3738. IEEE, 2000

    work page 2000

  14. [14]

    M. Tang, Y . J. Kim, and D. Manocha. C2a: Controlled conservative advancement for con- tinuous collision detection of polygonal models. In 2009 IEEE International Conference on Robotics and Automation, pages 849–854. IEEE, 2009

    work page 2009

  15. [15]

    Peternell, H

    M. Peternell, H. Pottmann, T. Steiner, and H. Zhao. Swept volumes. Computer-Aided Design and Applications, 2(5):599–608, 2005

    work page 2005

  16. [16]

    Rossignac, J

    J. Rossignac, J. J. Kim, S. Song, K. Suh, and C. Joung. Boundary of the volume swept by a free-form solid in screw motion. Computer-Aided Design, 39(9):745–755, 2007

    work page 2007

  17. [17]

    Abrams and P

    S. Abrams and P. K. Allen. Computing swept volumes. The Journal of Visualization and Computer Animation, 11(2):69–82, 2000

    work page 2000

  18. [18]

    Y . J. Kim, G. Varadhan, M. C. Lin, and D. Manocha. Fast swept volume approximation of complex polyhedral models. In Proceedings of the eighth ACM symposium on Solid modeling and applications, pages 11–22, 2003

    work page 2003

  19. [19]

    Zhang, Y

    X. Zhang, Y . J. Kim, and D. Manocha. Reliable sweeps. In 2009 SIAM/ACM joint conference on geometric and physical modeling , pages 373–378, 2009

    work page 2009

  20. [20]

    Sell ´an, N

    S. Sell ´an, N. Aigerman, and A. Jacobson. Swept volumes via spacetime numerical continua- tion. ACM Transactions on Graphics (TOG), 40(4):1–11, 2021

    work page 2021

  21. [21]

    D. Son, H. Yang, and D. Lee. Sim-to-real transfer of bolting tasks with tight tolerance. In 2020 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages 9056–9063. IEEE, 2020

    work page 2020

  22. [22]

    Danielczuk, A

    M. Danielczuk, A. Mousavian, C. Eppner, and D. Fox. Object rearrangement using learned im- plicit collision functions. In 2021 IEEE International Conference on Robotics and Automation (ICRA), pages 6010–6017. IEEE, 2021. 10

    work page 2021

  23. [23]

    H. Wang, S. Sridhar, J. Huang, J. Valentin, S. Song, and L. J. Guibas. Normalized object coordinate space for category-level 6d object pose and size estimation. InThe IEEE Conference on Computer Vision and Pattern Recognition (CVPR) , June 2019

    work page 2019

  24. [24]

    J. Pan, S. Chitta, and D. Manocha. Fcl: A general purpose library for collision and proximity queries. In 2012 IEEE international conference on robotics and automation, pages 3859–3866. IEEE, 2012

    work page 2012

  25. [25]

    J. Kim, J. Kim, D. Lee, Y . Jang, and B. Kim. Design of a low-cost and lightweight 6 dof bimanual arm for dynamic and contact-rich manipulation. arXiv preprint arXiv:2502.16908 , 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  26. [26]

    M. Z. Irshad, S. Zakharov, R. Ambrus, T. Kollar, Z. Kira, and A. Gaidon. Shapo: Implicit representations for multi-object shape, appearance, and pose optimization. In European Con- ference on Computer Vision, pages 275–292. Springer, 2022

    work page 2022

  27. [27]

    Y . Hong, K. Zhang, J. Gu, S. Bi, Y . Zhou, D. Liu, F. Liu, K. Sunkavalli, T. Bui, and H. Tan. Lrm: Large reconstruction model for single image to 3d. arXiv preprint arXiv:2311.04400 , 2023

    Pith/arXiv arXiv 2023

  28. [28]

    Gropp, L

    A. Gropp, L. Yariv, N. Haim, M. Atzmon, and Y . Lipman. Implicit geometric regularization for learning shapes. In Proceedings of the 37th International Conference on Machine Learning , pages 3789–3799, 2020

    work page 2020

  29. [29]

    Williams, P

    G. Williams, P. Drews, B. Goldfain, J. M. Rehg, and E. A. Theodorou. Aggressive driving with model predictive path integral control. In 2016 IEEE international conference on robotics and automation (ICRA), pages 1433–1440. IEEE, 2016

    work page 2016

  30. [30]

    D. C. Liu and J. Nocedal. On the limited memory bfgs method for large scale optimization. Mathematical programming, 45(1):503–528, 1989

    1989

  31. [31]

    H. J. T. Suh, T. Pang, and R. Tedrake. Bundled gradients through contact via randomized smoothing. IEEE Robotics and Automation Letters , 7(2):4000–4007, 2022

    work page 2022

  32. [32]

    C. R. Qi, H. Su, K. Mo, and L. J. Guibas. Pointnet: Deep learning on point sets for 3d classification and segmentation. In Proceedings of the IEEE conference on computer vision and pattern recognition, pages 652–660, 2017

    2017

  33. [33]

    D. Son, J. Kim, S. Son, and B. Kim. An intuitive multi-frequency feature representation for SO(3)-equivariant networks. In The Twelfth International Conference on Learning Represen- tations, 2024. URL https://openreview.net/forum?id=5JWAOLBxwp

    work page 2024

  34. [34]

    O. Puny, M. Atzmon, E. J. Smith, I. Misra, A. Grover, H. Ben-Hamu, and Y . Lipman. Frame averaging for invariant and equivariant network design. InInternational Conference on Learn- ing Representations, 2022. URL https://openreview.net/forum?id=zIUyj55nXR

    work page 2022

  35. [35]

    J. Yang, C. Deng, J. Wu, R. Antonova, L. Guibas, and J. Bohg. Equivact: Sim (3)-equivariant visuomotor policies beyond rigid object manipulation. In 2024 IEEE international conference on robotics and automation (ICRA) , pages 9249–9255. IEEE, 2024. 11 A Implementation Details in training From given training datasetD, the training objective isP d∈D LSV CD(...

    work page 2024

  36. [36]

    Case 1 and 2 have the same relative transform between two objects, but their global transforms are different

    We also modify the calculating gradient during L-BFGS to be the bundled gradient [31], which 13 Figure 8: Illustration of pre-processing for achieving invariance. Case 1 and 2 have the same relative transform between two objects, but their global transforms are different. If we treat the two objects as a single composite rigid body, we can assign frames {...

    work page

  37. [37]

    Let the combined input be x = [ξ; z1; z2; p1; p2]. We define the preprocessing function as Φ(x) = 1 max(∥z1∥, ∥z2∥) (ϕ(x)−1 · x), with ϕ : RM → SE (3) such that ϕ(T · x) = T · ϕ(x) ∀T ∈ SE (3), where M is the dimension of x and ∥z∥ is Euclidean norm of z. It is straightforward to verify that Φ is invariant under any transformation in SE (3) and scale. We ...

    work page 2000

This paper was first reviewed by deepseek-v4-flash on August 5, 2026.