Pith. sign in

REVIEW 3 major objections 6 minor 72 references

This paper claims that a small set of ruled surfaces—vertical extrusions of smooth curves that hug a shape's outer contour lines without colliding—can rough-machine material effectively with a straight hot wire.

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 ↗

Vertical extrusion of collision-free splines fitted to chosen outer contours generates efficient ruled surfaces for hot-wire rough machining with fewer cuts and lower error than manual design.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection A practical and novel hot-wire rough-machining pipeline with real fabrication evidence, but the collision-free guarantee is not actually conservative and the comparison numbers need cleaning up. the 3 major comments →

arxiv 2509.00840 v1 pith:XILIUJOJ submitted 2025-08-31 cs.CG

Outer Contour-driven Ruled Surface Generation for Linear Hot-wire Rough Machining

Zheng Zhang , Kang Wu , Yi-Fei Li , Xu Liu , Xiang Wang , Ligang Liu , Xiao-Ming Fu This is my paper
classification cs.CG MSC 65D1768U05
keywords ruled surfaceslinear hot-wire cuttingrough machiningcollision-free path planningnext-best-viewB-spline curve fittingdigital fabrication
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

Rough machining is the fast first pass that carves away bulk material before fine finishing; in linear hot-wire cutting, each pass of a straight wire sweeps out a ruled surface. This paper claims that a small set of such surfaces is enough if every surface is built from the target's outline: take the outer contour line seen from a well-chosen viewpoint, fit a smooth curve around it without touching the shape, and extrude the curve along the viewing direction. The proposed pipeline does this automatically, alternating a genetic search for the viewpoint that maximizes the projected area of material that would be removed, a constrained B-spline fit to the silhouette, and extrusion followed by material removal. On 35 test shapes, including 10 models actually cut on a hot-wire robot, the method needs only 5–15 cuts per model, keeps average errors mostly below one percent of the bounding-box diagonal, and reports lower error than manual toolpath design with the same number of cuts. If correct, it turns the path-planning part of rough machining into a push-button step that needs no manual cutting paths.

Core claim

This paper claims that an object's outer contour lines — its silhouettes from chosen viewpoints — are the right primitives for hot-wire rough-machining paths. A smooth curve hugging a silhouette without touching it, extruded along the view direction, makes a long, collision-free ruled surface that removes a large band of material in one pass. The authors build an automatic loop: genetic viewpoint search, constrained B-spline fit to the silhouette, extrusion, and material removal. On 35 shapes, including 10 physical fabrications, they report 5–15 cuts, errors mostly below 10⁻² of the box diagonal, and better accuracy than manual design or the earlier cover-to-fit method.

What carries the argument

The load-bearing object is the extruded-silhouette ruled surface: a B-spline curve fitted to the outer contour polygon g at a chosen camera, extruded along the view direction. A contour line wraps the shape and is long, so its vertical extrusion is a large, collision-free swept surface that strips material in one pass. Viewpoint selection maximizes the squared difference of rendered area images of remaining material and target. Fitting enforces the hard constraint by subdividing the control net 8 times with continuous collision detection to cap each control point's step size, while a squared-distance objective, shielding constant ϵ, mask δi, and control-point insertion reduce the error.

Load-bearing premise

The collision-free guarantee assumes that checking a control net subdivided eight times catches every collision the true B-spline curve could have with the target shape; if the curve can graze the shape while its subdivided control polygon clears it, the hard safety constraint is not actually guaranteed.

What would settle it

Take a silhouette with a deep narrow notch, fit the B-spline as the paper prescribes, then test the true curve (dense sampling or exact curve–polygon intersection) against the silhouette boundary. If any fitted curve intersects the shape while the eight-times-subdivided control net passes the continuous collision check, the 'guaranteed' collision-free claim fails; a physical version would show a wire cut that enters the target material.

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

If this is right

  • Straight-wire rough machining becomes push-button: given a mesh and a material box, the algorithm plans a small set of collision-free cutting surfaces with no manual path input, removing the main bottleneck of earlier guaranteed-collision-free methods.
  • A practical cut budget holds on complex shapes: in the paper's 22-model benchmark the pipeline stops within 15 cuts and keeps average errors mostly below one percent of the bounding-box diagonal, so fine machining starts from a close envelope.
  • At equal cut counts the automatic result has lower average error than a human designer's toolpaths, and compared with the earlier cover-to-fit pipeline it uses roughly a third of the cuts with about 70% lower error on the reported cases.
  • The method tolerates poor meshes: sparse, noisy, non-watertight, and non-manifold inputs all yield small errors, so raw scanned geometry is usable without cleanup.
  • Results transfer to hardware: 10 models cut on a six-axis hot-wire robot match the simulated cutting surfaces, supporting feasibility beyond simulation.

Where Pith is reading between the lines

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

  • In silhouette terms, the greedy loop approximates the shape's visual hull by a few prismatic extrusions; connecting to visual-hull theory could give a priori bounds on the number of cuts a given target requires.
  • The viewpoint fitness uses projected area as a stand-in for removed volume; a volume-weighted or opacity-accumulating render should correlate better with true material removal and might reduce the number of iterations, a testable modification the paper does not explore.
  • Because the paper reports that connecting cuts and returning the tool to standby dominate total fabrication time and that cut order is currently user-chosen, an obvious extension is to optimize cut sequencing over the generated surfaces; a fabrication-time benchmark would settle how much this saves.
  • The outer-contour primitive is not tied to straight wires: combining straight-wire passes for convex silhouettes with curved-wire passes for concavities could extend the method to the concave and non-zero-genus shapes the paper identifies as out of reach.
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

3 major / 6 minor

Summary. The paper proposes a pipeline for generating ruled surfaces for linear hot-wire rough machining. Given a triangular mesh M and an enclosing material box B, the method iteratively (1) selects a viewpoint with a genetic algorithm that maximizes projected-area difference between the remaining material and M, (2) extracts the outer contour of M from that viewpoint, (3) fits a collision-free B-spline curve to that contour in 2D using an adaptive squared-distance minimization with explicit collision checks, and (4) extrudes the curve along the view direction to form a ruled surface, then removes the material outside that surface. The process repeats until a volume-difference threshold or iteration cap. The experiments include ablations, stress tests, comparisons against a sampling baseline, differentiable rendering, a practitioner, and the authors' prior cover-to-fit method [6], plus 10 physical fabrications, reporting fewer cuts and lower approximation error than the baselines.

Significance. The application is timely and the empirical study is substantial: 35 models, 10 physical fabrications, stress tests on imperfect meshes, and ablations for the main algorithmic components. The central geometric observation — that silhouette-aligned extruded curves can remove material effectively in rough machining — is plausible and supported by physical examples. If the technical concerns below are addressed, this would be a solid contribution to digital fabrication and geometry processing. In particular, the comparison against practitioners and the physical fabrication validation are valuable strengths. However, the paper's hard collision-free guarantee is not established as written, and several quantitative comparison claims are inconsistent with the reported raw numbers. These points need to be fixed before the central claims can be accepted.

major comments (3)
  1. [Section 3.3.1, 'Optimization'] The hard cutting-validity constraint (Section 3.1) is enforced only through the step-size cap computed as follows: 'we first subdivide the control net of c k times (k = 8), then perform the continuous collision detection algorithm [68] between the subdivided control net and g to compute the maximum allowable step size tmax.' This check is not conservative for a B-spline curve. A B-spline lies in the convex hull of its control points, not on the subdivided control polygon; after k=8 knot insertions the polygon approximates the curve but does not bound it. A curve segment can intersect g while the tested polyline remains collision-free, so tmax may permit a collision. The same proxy is used in the 3D attempt (Section 4.1). To support the claim that the output surfaces 'do not collide with the input shape,' the line search must use a conservative test, e.g., checking the convex hulls of the
  2. [Section 4.1, 'Comparisons with the differentiable rendering'] The text states 'Our approximation error and viewpoint selection’s running time are both smaller.' The bird row in Fig. 13 reports differentiable rendering (1.04×10−3, 5, 7.5) vs. ours (7.92×10−3, 5, 0.6); the first quantity is the average error, so our error is about 7.6× larger. Either the tuple order or the reported value is wrong. Please correct this and re-evaluate the comparison claim.
  3. [Section 4.1, 'Comparisons with other methods'] The summary says 'our approximation error decreases by about 70%, and we use roughly 270% fewer cuts; meanwhile, our runtime is reduced by approximately 414%.' From the Fig. 14 tuples, the error reductions are about 35%, 48%, and 39%; the cut reductions are 74%, 67%, and 76%; the runtime reductions are 71%, 80%, and 85%. None of these matches '270% fewer cuts' or '414%' (a reduction cannot exceed 100%). These quantitative claims need to be recomputed and restated, e.g., as speedup factors.
minor comments (6)
  1. [Abstract] The sentence 'and present an adaptive algorithm' is missing a subject; consider 'and we present an adaptive algorithm.'
  2. [Section 3.2.3] 'Where ϕ and θ are the elevation and azimuth' should begin with a lowercase 'where' after the equation.
  3. [Section 3.3.1 vs Eq. (9)] The symbol k is used both for the control-net subdivision depth (k=8) and for the angle scaling factor in Eq. (9) (k=5). Rename one to avoid confusion.
  4. [Section 3.3.1] The phrase 'influence interval' of a control point is used but never defined. Please define it or replace it with a precise description (e.g., the support interval of the corresponding B-spline basis function).
  5. [Table 1] The table formatting is broken in several rows: entries run together (e.g., '22.3Fig. 9 plane 8 6 .28×10−3 ...') and numbers are hard to parse. Please fix line breaks and spacing.
  6. [Section 4.1, 'Optimizing ruled surfaces'] The negative result that the 3D optimization leaves the surface unchanged is interesting, but the phrase 'has a very high probability to trigger collision' is informal. Since the 3D step is abandoned, please state the reason more precisely, e.g., 'no descent step satisfied the conservative collision test.'

Circularity Check

0 steps flagged

No significant circularity: principal outputs are externally benchmarked; collision-test issue is a soundness gap, not a circular reduction.

full rationale

Walking the claimed derivation chain: (1) The outer-contour observation is an asserted design principle, not derived from the paper's own outputs; its effectiveness is checked against manual practitioner cuts (Fig. 11) and 10 physical fabrications (Figs. 18-19), i.e., external benchmarks. (2) The viewpoint objective E^ours_material = ||I^a_Bi - I^a_M||^2 (Eq. 2) is an optimization criterion for selecting camera directions; the reported quality metric d_avg (Eq. 8) is a bidirectional geometric distance to the input, not the same fitted quantity, so no metric is renamed as a prediction. (3) The B-spline c is fit to the outer contour polygon g, and the final ruled surface is its extrusion; no equation in Section 3.3 makes the output equal to the input by construction. Enlarging the initial circle until enclosing/collision-free constraints are met is legitimate initialization, not fitting a parameter then predicting it. (4) The comparison with [6] uses the same group's prior work as a baseline, but that prior work is not cited as a premise for correctness or uniqueness; it is an empirical competitor, and the central claims also hold against practitioners. (5) The 'limitations' passages (concave/non-zero genus shapes, abandoned 3D optimization, cutting path optimality) are honesty caveats and do not reveal circularity. The one genuine weakness is a correctness/soundness concern: the collision-free guarantee relies on subdividing the control net as a proxy for the B-spline curve (Section 3.3.1), and no proof is given that this test is conservative. That is a potential violation of the hard 'cutting validity' constraint, but it is not a case where a prediction reduces to an input by definition or a self-citation chain. Per the rubric, such correctness risk is outside the circularity score.

Axiom & Free-Parameter Ledger

7 free parameters · 5 axioms · 0 invented entities

The method rests on hand-set hyperparameters and on modeling approximations: the area proxy for volume, the subdivided-control-net proxy for collision, and the unspecified silhouette extraction. None of these are fitted to external data, but several are load-bearing for the paper's hard guarantees.

free parameters (7)
  • termination volume threshold alpha = 0.025
    Stops the greedy iteration when remaining material volume difference vs input shape is below alpha; chosen by hand (Section 4).
  • max iterations Niter = 15
    Caps number of cuts; directly limits 'small set of ruled surfaces' claim (Section 4).
  • collision subdivision depth k = 8
    Depth of control-net subdivision in step-size search; load-bearing for the approximated collision-free guarantee (Section 3.3.1).
  • init d decay schedule = 1% dbb, factor 0.8/iteration
    Convergence control in 2D fitting; prevents parts of the curve from approaching g too early (Section 3.3.1).
  • GA sample/population sizes = Nsample=5000, Npop=30
    Discretization of viewpoint search; determines quality of selected viewpoint (Section 3.2.4).
  • 2D fitting weights = w=1, eta0=0, eta1=1
    Balance approximation vs smoothness in B-spline fitting (Section 3.3.1 Eq. 4).
  • convex hull simplification threshold beta = 2% dbb
    Reduces initial control points; affects starting curve and optimization path (Section 3.3.1).
axioms (5)
  • domain assumption Projected area difference E_ours (Eq. 2) approximates material removal volume E_material (Eq. 1) well enough for viewpoint selection.
    The whole NBV fitness is (2); the paper acknowledges it is an approximation but does not bound its error (Section 3.2.1).
  • ad hoc to paper Collision-freedom of the k=8 subdivided control net implies collision-freedom of the B-spline curve c.
    Used to compute max allowable step size in line search (Section 3.3.1); not stated or proven, and false in general since a B-spline need not lie inside the control polygon.
  • domain assumption The outer contour line g is obtained accurately from the input mesh and viewpoint.
    ObtainOuterContour(M, Ck) in Alg. 1 is never specified; relies on Pytorch3D rendering settings (Section 3.2.3).
  • domain assumption Vertical extrusion of a curve that approximates the outer contour yields a ruled surface that removes material effectively without collision.
    This is the paper's central observation; it is validated empirically but not derived (Abstract, Section 3.3.2).
  • standard math Standard B-spline properties and Newton/CCD algorithms behave as cited.
    Background from [51], [52], [67], [68] is taken as given.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Outer Contour-driven Ruled Surface Generation for Linear Hot-wire Rough Machining." pith.science (2026). https://pith.science/paper/XILIUJOJ

@misc{pith2026250900840,
  author       = {Pith},
  title        = {Pith review of: Outer Contour-driven Ruled Surface Generation for Linear Hot-wire Rough Machining},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/XILIUJOJ}},
  note         = {Machine review of arXiv:2509.00840}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

We propose a novel method to generate a small set of ruled surfaces that do not collide with the input shape for linear hot-wire rough machining. Central to our technique is a new observation: the ruled surfaces constructed by vertical extrusion from planar smooth curves, which approach the input shape's outer contour lines while having no collisions, are capable of removing materials effectively during rough machining. Accordingly, we develop an iterative algorithm that alternates in each iteration between computing a viewpoint to determine an outer contour line and optimizing a smooth curve to approximate that contour line under the collision-free constraint. Specifically, a view selection approach based on genetic algorithm is used to optimize the viewpoint for removing materials as much as possible, and present an adaptive algorithm to find the constrained curves. The feasibility and practicability of our method are demonstrated through 10 physical examples. Compared with manual designs, our method obtains lower errors with the same number of cuts.

Figures

Figures reproduced from arXiv: 2509.00840 by Kang Wu, Ligang Liu, Xiang Wang, Xiao-Ming Fu, Xu Liu, Yi-Fei Li, Zheng Zhang.

Figure 1
Figure 1. Figure 1: We generate a small set of ruled surfaces to guide the cutting of material [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: (a) A viewpoint of the cat model. (b) A smooth curve (red) that [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Initializing control points. Given a viewpoint (a), we place initial control [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: (a) Input shape. (b) The convex hull H. The vertices are colored blue, and the edges are colored orange. (c) The simplified convex hull H′ . However, a large number of initial control points results in a low optimization speed. Thus, after obtaining the initial control points, we reduce the number of {p h i } by simplifying H. If the number of {p h i } in the initialization step is larger than Nc, we run t… view at source ↗
Figure 7
Figure 7. Figure 7: (b) and (c)). Since δi removes mismatched correspondences between c and g, while the diminishing constant ϵ controls the (a) (b) (c) (d) [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Comparison with using a barrier function. (a) Input shape. (b) [PITH_FULL_IMAGE:figures/full_fig_p007_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Stress test. From left to right, the number of cuts for each model is 8, [PITH_FULL_IMAGE:figures/full_fig_p008_9.png] view at source ↗
Figure 11
Figure 11. Figure 11: Comparisons with practitioners. The second and third columns show [PITH_FULL_IMAGE:figures/full_fig_p008_11.png] view at source ↗
Figure 10
Figure 10. Figure 10: Genetic algorithm vs the sampling method. We evaluate our method [PITH_FULL_IMAGE:figures/full_fig_p008_10.png] view at source ↗
Figure 13
Figure 13. Figure 13: Comparisons with the differentiable rendering. The second and third columns show the differentiable rendering results and our results, respectively. The text below each result indicates the average error, number of cuts, and viewpoint optimization time consumption (minutes). and the time to obtain the optimal viewpoint. Our approximation error and viewpoint selection’s running time are both smaller. Compa… view at source ↗
Figure 15
Figure 15. Figure 15: The result of 3D approximation. (a) Input shape. (b) The B-spline [PITH_FULL_IMAGE:figures/full_fig_p010_15.png] view at source ↗
Figure 16
Figure 16. Figure 16: Twelve examples. For each example, the light blue model is the input, and the white is the result. [PITH_FULL_IMAGE:figures/full_fig_p011_16.png] view at source ↗
Figure 17
Figure 17. Figure 17: Robotic arm product model: KUKA® Quantec KR120 R2700 extra HA Robot (left). The fabrication process of the duck model (right) [PITH_FULL_IMAGE:figures/full_fig_p011_17.png] view at source ↗
Figure 18
Figure 18. Figure 18: All fabrication results. into multiple zero-genus components for individual cutting may partially overcome these issues, which is an intriguing direction for future research. Optimal cutting path. During machining, the end-effector track￾ing error does not remain constant in the cutting path. Actually, when the cutting tool is near a singularity where the Jacobian’s condition number is high, the control e… view at source ↗
Figure 22
Figure 22. Figure 22: Left: an example of cutting a concave shape. Right: an example of [PITH_FULL_IMAGE:figures/full_fig_p012_22.png] view at source ↗
Figure 19
Figure 19. Figure 19: Physical fabrication of ten models. For each example, we show the [PITH_FULL_IMAGE:figures/full_fig_p012_19.png] view at source ↗
Figure 20
Figure 20. Figure 20: (a) The material is placed on a cylindrical workbench during fabrication. [PITH_FULL_IMAGE:figures/full_fig_p012_20.png] view at source ↗
Figure 21
Figure 21. Figure 21: Simulation error distribution. We use the one-sided distance from the [PITH_FULL_IMAGE:figures/full_fig_p012_21.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

72 extracted references · 69 canonical work pages

  1. [1]

    Yan, C.-H

    C. Yan, C.-H. Lee, X. Li, Y . Zhang, K. Min, A multi-objective tool-axis optimization algorithm based on covariant field functional, Journal of Manu- facturing Systems 48 (2018) 30–37

    work page 2018

  2. [2]

    D. Liu, M. Luo, G. Urbikain Pelayo, D. O. Trejo, D. Zhang, Position-oriented process monitoring in milling of thin-walled parts, Journal of Manufacturing Systems 60 (2021) 360–372

    work page 2021

  3. [3]

    C. Li, G. Zhao, F. Meng, S. Yu, B. Yao, H. Liu, Multi-objective optimization of machining parameters in complete peripheral milling process with vari- able curvature workpieces, Journal of Manufacturing Processes 117 (2024) 95–110

    work page 2024

  4. [4]

    Z. Chen, Y . Zhang, G. Zhang, W. Li, Modeling and reducing workpiece corner error due to wire deflection in wedm rough corner-cutting, Journal of Manufacturing Processes 36 (2018) 557–564

    work page 2018

  5. [5]

    Z. Xu, H. Huang, C. Cui, X. Liao, M. Wu, Z. Xue, Complex-shaped metal parts high efficiency sawing with diamond wire, International Journal of Mechanical Sciences 250 (2023) 108306

    work page 2023

  6. [6]

    Zhang, Y

    Z. Zhang, Y . Li, K. Wu, H. Zhao, X. Liu, X. Wang, L. Liu, X. Fu, Carving shapes with ruled surfaces for rough machining, Computers & Graphics (2025) 104386

    work page 2025

  7. [7]

    van Sosin, M

    B. van Sosin, M. Bartoˇn, G. Elber, Accessibility for line-cutting in freeform surfaces, Comput. Aided Des. 114 (2019) 202–214

    work page 2019

  8. [8]

    Duenser, R

    S. Duenser, R. Poranne, B. Thomaszewski, S. Coros, Robocut: Hot-wire cutting with robot-controlled flexible rods, ACM Trans. Graph. 39 (4) (2020) 98–1

    work page 2020

  9. [9]

    R.-S. Lin, Y . Koren, Ruled surface machining on five-axis cnc machine tools, Journal of Manufacturing Processes 2 (1) (2000) 25–35. 12

    work page 2000

  10. [10]

    Koc, Y .-S

    B. Koc, Y .-S. Lee, Adaptive ruled layers approximation of stl models and multiaxis machining applications for rapid prototyping, Journal of Manufac- turing Systems 21 (3) (2002) 153–166

    work page 2002

  11. [11]

    Sprott, B

    K. Sprott, B. Ravani, Cylindrical milling of ruled surfaces, The International Journal of Advanced Manufacturing Technology 38 (2008) 649–656

    work page 2008

  12. [12]

    Gong, L.-X

    H. Gong, L.-X. Cao, J. Liu, Improved positioning of cylindrical cutter for flank milling ruled surfaces, Comput. Aided Des. 37 (12) (2005) 1205–1213

    work page 2005

  13. [13]

    C. Li, S. Bedi, S. Mann, Flank milling of a ruled surface with conical tools—an optimization approach, The International Journal of Advanced Manufacturing Technology 29 (2006) 1115–1124

    work page 2006

  14. [14]

    H. Hua, T. Jia, Wire cut of double-sided minimal surfaces, The Visual Computer 34 (6) (2018) 985–995

    work page 2018

  15. [15]

    Chu, H.-Y

    C.-H. Chu, H.-Y . Chen, C.-H. Chang, Continuity-preserving tool path gener- ation for minimizing machining errors in five-axis cnc flank milling of ruled surfaces, Journal of Manufacturing Systems 55 (2020) 171–178

    work page 2020

  16. [16]

    Y . Zhou, T. Xing, Y . Song, Y . Li, X. Zhu, G. Li, S. Ding, Digital-twin-driven geometric optimization of centrifugal impeller with free-form blades for five-axis flank milling, Journal of Manufacturing Systems 58 (2021) 22–35

    work page 2021

  17. [17]

    B. Yamauchi, A frontier-based approach for autonomous exploration, in: Proceedings 1997 IEEE International Symposium on Computational Intelli- gence in Robotics and Automation CIRA’97.’Towards New Computational Principles for Robotics and Automation’, IEEE, 1997, pp. 146–151

    work page 1997

  18. [18]

    J. I. Vasquez-Gomez, L. E. Sucar, R. Murrieta-Cid, J.-C. Herrera-Lozada, Tree-based search of the next best view/state for three-dimensional object reconstruction, International Journal of Advanced Robotic Systems 15 (1) (2018) 1729881418754575

    work page 2018

  19. [19]

    J. I. Vasquez-Gomez, L. E. Sucar, R. Murrieta-Cid, View planning for 3d object reconstruction with a mobile manipulator robot, in: 2014 IEEE/RSJ International Conference on Intelligent Robots and Systems, IEEE, 2014, pp. 4227–4233

    work page 2014

  20. [20]

    J. I. Vasquez-Gomez, L. E. Sucar, R. Murrieta-Cid, View /state planning for three-dimensional object reconstruction under uncertainty, Autonomous Robots 41 (2017) 89–109

    work page 2017

  21. [21]

    J. I. V ´asquez-G´omez, E. L ¨opez-Damian, L. E. Sucar, View planning for 3d object reconstruction, in: 2009 IEEE/RSJ International Conference on Intelligent Robots and Systems, IEEE, 2009, pp. 4015–4020

    work page 2009

  22. [22]

    J. I. Vasquez-Gomez, L. E. Sucar, R. Murrieta-Cid, E. Lopez-Damian, V olu- metric next-best-view planning for 3d object reconstruction with positioning error, International Journal of Advanced Robotic Systems 11 (10) (2014) 159

    work page 2014

  23. [23]

    Potthast, G

    C. Potthast, G. S. Sukhatme, A probabilistic framework for next best view estimation in a cluttered environment, Journal of Visual Communication and Image Representation 25 (1) (2014) 148–164

    work page 2014

  24. [24]

    Delmerico, S

    J. Delmerico, S. Isler, R. Sabzevari, D. Scaramuzza, A comparison of vol- umetric information gain metrics for active 3d object reconstruction, Au- tonomous Robots 42 (2) (2018) 197–208

    work page 2018

  25. [25]

    S. Wu, W. Sun, P. Long, H. Huang, D. Cohen-Or, M. Gong, O. Deussen, B. Chen, Quality-driven poisson-guided autoscanning, ACM Transactions on Graphics (TOG) 33 (6) (2014) 1–12

    work page 2014

  26. [26]

    Maver, R

    J. Maver, R. Bajcsy, Occlusions as a guide for planning the next view, IEEE transactions on pattern analysis and machine intelligence 15 (5) (1993) 417–433

    work page 1993

  27. [27]

    Yuan, A mechanism of automatic 3d object modeling, IEEE Transactions on pattern analysis and machine intelligence 17 (3) (1995) 307–311

    X. Yuan, A mechanism of automatic 3d object modeling, IEEE Transactions on pattern analysis and machine intelligence 17 (3) (1995) 307–311

    work page 1995

  28. [28]

    Pito, A solution to the next best view problem for automated surface acquisition, IEEE Transactions on pattern analysis and machine intelligence 21 (10) (1999) 1016–1030

    R. Pito, A solution to the next best view problem for automated surface acquisition, IEEE Transactions on pattern analysis and machine intelligence 21 (10) (1999) 1016–1030

    work page 1999

  29. [29]

    S. Chen, Y . Li, Vision sensor planning for 3-d model acquisition, IEEE Transactions on Systems, Man, and Cybernetics, Part B (Cybernetics) 35 (5) (2005) 894–904

    work page 2005

  30. [30]

    X. Zhou, B. He, Y . Li, A novel view planning method for automatic re- construction of unknown 3-d objects based on the limit visual surface, in: 2009 Fifth International Conference on Image and Graphics, IEEE, 2009, pp. 301–306

    work page 2009

  31. [31]

    Kriegel, T

    S. Kriegel, T. Bodenm¨uller, M. Suppa, G. Hirzinger, A surface-based next- best-view approach for automated 3d model completion of unknown objects, in: 2011 IEEE International Conference on Robotics and Automation, IEEE, 2011, pp. 4869–4874

    work page 2011

  32. [32]

    Mendoza, J

    M. Mendoza, J. I. Vasquez-Gomez, H. Taud, L. E. Sucar, C. Reta, Super- vised learning of the next-best-view for 3d object reconstruction, Pattern Recognition Letters 133 (2020) 224–231

    work page 2020

  33. [33]

    Z. Wu, S. Song, A. Khosla, F. Yu, L. Zhang, X. Tang, J. Xiao, 3d shapenets: A deep representation for volumetric shapes, in: Proceedings of the IEEE conference on computer vision and pattern recognition, 2015, pp. 1912– 1920

    work page 2015

  34. [34]

    R. Zeng, W. Zhao, Y .-J. Liu, Pc-nbv: A point cloud based deep network for e fficient next best view planning, in: 2020 IEEE /RSJ International Conference on Intelligent Robots and Systems (IROS), IEEE, 2020, pp. 7050–7057

    work page 2020

  35. [35]

    Z. Liu, Z. Cao, J. Li, P. Guan, J. Yu, Dg-nbv: A cognitive framework for direct generation of next best view in continuous view space, IEEE Transactions on Cognitive and Developmental Systems (2024)

    work page 2024

  36. [36]

    X. Xiao, Q. Fang, W. Peng, Y . Wang, L. Hong, J. Ye, Next best view planning via deep reinforcement leaning for 3d reconstruction of turbine blades, in: 2024 36th Chinese Control and Decision Conference (CCDC), IEEE, 2024, pp. 4022–4027

    work page 2024

  37. [37]

    L. Li, X. Zhang, Boundary exploration of next best view policy in 3d robotic scanning, arXiv preprint arXiv:2412.10444 (2024)

    Pith/arXiv arXiv 2024

  38. [38]

    L. Jin, X. Chen, J. R¨uckin, M. Popovi´c, Neu-nbv: Next best view planning using uncertainty estimation in image-based neural rendering, in: 2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), IEEE, 2023, pp. 11305–11312

    work page 2023

  39. [39]

    S. Lee, L. Chen, J. Wang, A. Liniger, S. Kumar, F. Yu, Uncertainty guided policy for active robotic 3d reconstruction using neural radiance fields, IEEE Robotics and Automation Letters 7 (4) (2022) 12070–12077

    work page 2022

  40. [40]

    Y . Ran, J. Zeng, S. He, J. Chen, L. Li, Y . Chen, G. Lee, Q. Ye, Neurar: Neural uncertainty for autonomous 3d reconstruction with implicit neural representations, IEEE Robotics and Automation Letters 8 (2) (2023) 1125– 1132

    work page 2023

  41. [41]

    X. Pan, Z. Lai, S. Song, G. Huang, Activenerf: Learning where to see with uncertainty estimation, in: European Conference on Computer Vision, Springer, 2022, pp. 230–246

    work page 2022

  42. [42]

    D. Yan, J. Liu, F. Quan, H. Chen, M. Fu, Active implicit object reconstruction using uncertainty-guided next-best-view optimization, IEEE Robotics and Automation Letters (2023)

    work page 2023

  43. [43]

    Park, Choosing nodes and knots in closed b-spline curve interpolation to point data, Computer-Aided Design 33 (13) (2001) 967–974

    H. Park, Choosing nodes and knots in closed b-spline curve interpolation to point data, Computer-Aided Design 33 (13) (2001) 967–974

    work page 2001

  44. [44]

    T. I. Vassilev, Fair interpolation and approximation of b-splines by energy minimization and points insertion, Computer-Aided Design 28 (9) (1996) 753–760

    work page 1996

  45. [45]

    X. Wang, F. F. Cheng, B. A. Barsky, Energy and b-spline interproximation, Computer-Aided Design 29 (7) (1997) 485–496

    work page 1997

  46. [46]

    Dierckx, Curve and surface fitting with splines, Oxford University Press, 1995

    P. Dierckx, Curve and surface fitting with splines, Oxford University Press, 1995

    work page 1995

  47. [47]

    Zheng, P

    W. Zheng, P. Bo, Y . Liu, W. Wang, Fast b-spline curve fitting by l-bfgs, Computer Aided Geometric Design 29 (7) (2012) 448–462

    work page 2012

  48. [48]

    Javidrad, An accelerated simulated annealing method for b-spline curve fitting to strip-shaped scattered points, International Journal of CAD/CAM 12 (1) (2012) 9–19

    F. Javidrad, An accelerated simulated annealing method for b-spline curve fitting to strip-shaped scattered points, International Journal of CAD/CAM 12 (1) (2012) 9–19

    work page 2012

  49. [49]

    Lin, L.-Y

    F. Lin, L.-Y . Shen, C.-M. Yuan, Z. Mi, Certified space curve fitting and trajectory planning for cnc machining with cubic b-splines, Computer-Aided Design 106 (2019) 13–29

    work page 2019

  50. [50]

    I. L. Jover, T. Debarre, S. Aziznejad, M. Unser, Coupled splines for sparse curve fitting, IEEE Transactions on Image Processing 31 (2022) 4707–4718

    work page 2022

  51. [51]

    W. Wang, H. Pottmann, Y . Liu, Fitting b-spline curves to point clouds by curvature-based squared distance minimization, ACM Transactions on Graphics (ToG) 25 (2) (2006) 214–238

    work page 2006

  52. [52]

    Pottmann, S

    H. Pottmann, S. Leopoldseder, M. Hofer, Approximation with active b-spline curves and surfaces, in: 10th Pacific Conference on Computer Graphics and Applications, 2002. Proceedings., IEEE, 2002, pp. 8–25

    work page 2002

  53. [53]

    Y . Liu, W. Wang, A revisit to least squares orthogonal distance fitting of parametric curves and surfaces, in: Advances in Geometric Modeling and Processing: 5th International Conference, GMP 2008, Hangzhou, China, April 23-25, 2008. Proceedings 5, Springer, 2008, pp. 384–397

    work page 2008

  54. [54]

    Blake, M

    A. Blake, M. Isard, Active contours: the application of techniques from graphics, vision, control theory and statistics to visual tracking of shapes in motion, Springer Science & Business Media, 2012

    work page 2012

  55. [55]

    Ebrahimi, G

    A. Ebrahimi, G. B. Loghmani, B-spline curve fitting by diagonal approxima- tion bfgs methods, Iranian Journal of Science and Technology, Transactions A: Science 43 (2019) 947–958

    work page 2019

  56. [56]

    Bergstr¨om, I

    P. Bergstr¨om, I. S ¨oderkvist, Fitting nurbs using separable least squares techniques, International Journal of Mathematical Modelling and Numerical Optimisation 3 (4) (2012) 319–334

    work page 2012

  57. [57]

    H. Xie, H. Qin, Automatic knot determination of nurbs for interactive geo- metric design, in: Proceedings International Conference on Shape Modeling and Applications, IEEE, 2001, pp. 267–276

    work page 2001

  58. [58]

    Irshad, S

    M. Irshad, S. Khalid, M. Z. Hussain, M. Sarfraz, Outline capturing using rational functions with the help of genetic algorithm, Applied mathematics 13 and computation 274 (2016) 661–678

    work page 2016

  59. [59]

    G´alvez, A

    A. G´alvez, A. Iglesias, Firefly algorithm for explicit b-spline curve fitting to data points, Mathematical Problems in Engineering 2013 (1) (2013) 528215

    work page 2013

  60. [60]

    A. Y . Hasegawa, C. Tormena, R. S. Parpinelli, B´ezier curve parametrization using a multiobjective evolutionary algorithm., International Journal of Computer Science & Applications 11 (2) (2014)

    work page 2014

  61. [61]

    Zhang, Z

    J. Zhang, Z. Lu, M. Li, Active contour-based method for finger-vein image segmentation, IEEE Transactions on Instrumentation and Measurement 69 (11) (2020) 8656–8665

    work page 2020

  62. [62]

    Monemian, H

    M. Monemian, H. Rabbani, Analysis of a novel segmentation algorithm for optical coherence tomography images based on pixels intensity correlations, IEEE Transactions on Instrumentation and Measurement 70 (2020) 1–12

    work page 2020

  63. [63]

    Sasmal, M

    P. Sasmal, M. K. Bhuyan, S. Gupta, Y . Iwahori, Detection of polyps in colonoscopic videos using saliency map-based modified particle filter, IEEE Transactions on Instrumentation and Measurement 70 (2021) 1–9

    work page 2021

  64. [64]

    C. Yang, L. Wu, Y . Chen, G. Wang, G. Weng, An active contour model based on retinex and pre-fitting reflectance for fast image segmentation, Symmetry 14 (11) (2022) 2343

    work page 2022

  65. [65]

    M. Kass, A. Witkin, D. Terzopoulos, Snakes: Active contour models, Inter- national journal of computer vision 1 (4) (1988) 321–331

    work page 1988

  66. [66]

    J. D. Bayer, J. Beaumont, A. Krol, Laplace–dirichlet energy field specifi- cation for deformable models. an fem approach to active contour fitting, Annals of biomedical engineering 33 (2005) 1175–1186

    work page 2005

  67. [67]

    N. Ravi, J. Reizenstein, D. Novotny, T. Gordon, W.-Y . Lo, J. John- son, G. Gkioxari, Accelerating 3d deep learning with pytorch3d, arXiv:2007.08501 (2020)

    Pith/arXiv arXiv 2007

  68. [68]

    Redon, A

    S. Redon, A. Kheddar, S. Coquillart, Fast continuous collision detection between rigid bodies, Computer Graphics Forum 21 (3) (2002) 279–287

    work page 2002

  69. [69]

    McNeel, Associates, Rhinoceros 3d (2024)

    R. McNeel, Associates, Rhinoceros 3d (2024). URL https://www.rhino3d.com/

    work page 2024

  70. [70]

    S. Liu, T. Li, W. Chen, H. Li, Soft rasterizer: A differentiable renderer for image-based 3d reasoning, in: Proceedings of the IEEE/CVF international conference on computer vision, 2019, pp. 7708–7717

    work page 2019

  71. [71]

    D. P. Kingma, J. Ba, Adam: A method for stochastic optimization, CoRR abs/1412.6980 (2014). URL https://api.semanticscholar.org/CorpusID:6628106

    Pith/arXiv arXiv 2014

  72. [72]

    Lynch, Modern Robotics, Cambridge University Press, 2017

    K. Lynch, Modern Robotics, Cambridge University Press, 2017. 14

    work page 2017

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