Coverage Path Planning using Path Primitive Sampling and Primitive Coverage Graph for Visual Inspection

Di Deng; Kenji Shimada; Wei Jing; Yong Liu; Zhe Xiao

REVIEW 3 major objections 5 minor 31 references

Coverage Path Planning using Path Primitive Sampling and Primitive Coverage Graph for Visual Inspection

T0 review · 3 major / 5 minor · reviewed 2026-08-14 · deepseek-v4-flash

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

Pith's one-line read Directly sampling flight segments, not viewpoints, shortens UAV inspection paths by 18–29%.

desk verdict A genuinely new primitive-based coverage formulation with real practical promise, but the 18-29% savings claim is undercut by missing achieved-coverage numbers and no guarantee that the greedy search can reach the 99% target. read the letter →

arxiv 1908.02901 v1 pith:ZCJL7HTT submitted 2019-08-08 cs.RO

Wei Jing , Di Deng , Zhe Xiao , Yong Liu , Kenji Shimada This is my paper · ORCID

classification cs.RO

keywords coveragepathplanningprimitivesamplinggraphUAVvisualinspectionvoxeldilationandsubtractiongreedyneighborhoodsearchvisibilityestimationcontinuousvideocapture

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

The paper seeks to show that a UAV inspecting a 3D structure can be planned by sampling and searching over continuous flight segments—path primitives—rather than by first choosing discrete viewpoints and then routing between them. It builds a Primitive Coverage Graph whose edges carry flight distance and precomputed visibility, then grows a single branchless path greedily until a required surface-coverage ratio is reached. In simulation on two building-like structures, the resulting paths reach 99% target coverage while cutting flight distance by 18.4% and 29.2% relative to viewpoint-plus-TSP planning and a greedy viewpoint baseline. The same approach is also demonstrated in a field test with a reconstructed 3D model. A sympathetic reader would take the central claim to be that continuous path-primitive search is a viable, more efficient alternative to discrete viewpoint planning for video-stream inspection.

What carries the argument

The central object is the Primitive Coverage Graph (PCG), a graph whose nodes are sampled via-points and whose edges are path primitives. Each edge stores the flight distance and an m-dimensional binary visibility vector indicating which surface patches of the target are visible from that segment. The load-bearing mechanism is the Greedy Neighborhood Search: starting from an initial edge, it repeatedly extends the current path by the neighboring primitive with the highest incremental coverage per unit flight distance until the desired coverage ratio is reached. Supporting mechanisms are voxel dilation and subtraction to define the sampling region and a sampling-based visibility estimator that combines the visibility of several viewpoints along each primitive into one edge attribute.

What would settle it

On either of the two test structures, run the proposed planning pipeline with the reported parameters, fly the resulting path in a camera-equipped simulator, and reconstruct the target as a voxel model; if the achieved surface coverage falls below 99% for a feasible sampling budget, or if the flown path length does not beat the 507.7 m and 587.5 m VPP-TSP baselines, the central claim fails.

Watch

Extended reading notes

Core claim

The central claim is that the coverage path planning problem for UAV visual inspection is better solved by making the path itself the sampling unit. The authors argue that sampling via-points in a shell around the target, obtained by voxel dilation and subtraction, connecting nearby via-points into collision-free path primitives, and encoding these primitives as edges in a Primitive Coverage Graph allows a greedy search to assemble a connected, unbranched inspection path that directly satisfies coverage constraints while minimizing flight distance. They report that on two target structures this yields 425.6 m and 466.2 m paths versus 507.7 m and 587.5 m for the viewpoint-plus-TSP baseline and 531.0 m and 687.1 m for the greedy viewpoint baseline, corresponding to 18.4% and 29.2% reductions in required inspection time at constant speed. The paper claims this advantage follows because the path primitive is the natural unit for a camera that records video continuously, and because visibility is precomputed per path segment rather than per viewpoint.

Load-bearing premise

The load-bearing premise is that the randomly sampled via-points and path primitives happen to form a connected, branchless path that can reach the desired 99% coverage; the paper gives no guarantee that this happens for a chosen sampling density.

Editorial extensions

If this is right

Inspection paths can be generated as continuous collision-free flight segments rather than as an abstract set of waypoints, so the planned trajectory matches video-stream capture directly.
The 18.4% and 29.2% flight-time reductions imply that, for constant-speed UAVs, comparable coverage can be obtained with proportionally shorter endurance requirements.
Because visibility is precomputed per path primitive, the graph search itself only combines coverage bitsets and distances, keeping the search cheap after the offline graph construction.
The modular design means other local planners, sampling strategies, visibility models, or search algorithms can be plugged in without changing the PCG encoding.
The field-test reconstruction from 87 sampled video frames suggests that the planned paths provide enough visual overlap and viewpoint continuity for structure-from-motion reconstruction.

Reading between the lines

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

Treating edges rather than nodes as the coverage unit suggests a general principle for streaming sensors: when the camera records continuously, the segment, not the viewpoint, is the atomic measurement, and the same reframing may apply to manipulator painting or polishing tasks.
The greedy edge-addition rule resembles submodular coverage maximization, so an exchange or local-search step after Greedy Neighborhood Search might recover some of the optimality gap; the paper does not claim optimality.
A testable extension is to re-run the comparison with visibility computed only from discrete viewpoints rather than from path primitives; if the gap disappears, the advantage may come from the visibility model rather than from path-primitive search itself.
Because the framework is offline and model-based, adapting it to changing environments would require incremental PCG updates, which the paper leaves open.

Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Desk Editor's Note

read the letter

Colleague,

Quick take: this paper is worth a serious look. The core idea—sample path primitives instead of viewpoints and search a graph of those primitives—is a real reformulation of the coverage planning problem, not just a tweak. It builds on the authors' earlier viewpoint work, but the search space and the PCG encoding are genuinely new. That alone makes it interesting for anyone working on UAV inspection.

The algorithms are coherent. The voxel dilation/subtraction sampling region is sensible. The PCG edge-based visibility encoding is a natural way to extend binary visibility from points to short paths. The greedy search is simple and plausible. The reported path-length reductions of 18.4% and 29.2% over two viewpoint-plus-TSP baselines are plausible if the comparison is fair.

But here is the soft spot: the claim that coverage is satisfied is not actually evidenced. The paper never reports the achieved coverage ratio for the proposed paths or for the baselines in the Drake-Octomap simulation. Table II gives only mean path lengths, with no standard deviation across the 10 trials. The visualization in Fig. 10 is qualitative. If the proposed paths cover 95% while the baselines cover 99%, the flight-time comparison is meaningless. That is a load-bearing gap, not a minor omission.

The algorithmic side has a related gap. Algorithm 3 loops while delta <= delta_d, but nothing guarantees that a randomly sampled PCG contains a connected, branchless path that reaches delta_d. FindTraj can get stuck at a via-point with no coverage-improving neighbor. No connectivity, completeness, or termination analysis is given. This is a real theoretical hole, though in practice with dense sampling it may rarely matter. Still, it should be flagged.

The comparison set is also narrower than the Related Work promises. The baselines are two-stage viewpoint methods; the continuous-path planners cited in the introduction (e.g., Bircher et al., Papadopoulos et al.) are not benchmarked. That limits the strength of the claimed advantage.

On the positive side, there is no fitting-to-data in the derivation. The parameters are user-chosen, the greedy selection is not tuned to match measured distances, and the field test with real UAV footage and SFM reconstruction is a genuine attempt at external validation. The self-citation to the authors' own visibility model is not a problem; it is their prior work and they reuse it consistently.

The bottom line: the contribution is solid and the idea deserves attention, but the paper as submitted overclaims its evidence. If the authors add achieved coverage numbers, error bars, a comparison to at least one continuous-path planner, and some discussion of the search's completeness, I would be comfortable accepting it. As is, it is a conditional reject or major revision.

For peer review: yes, this deserves referee time. The idea is novel enough and the experiments, while incomplete, are on the right track. I would send it out.

Referee Report

3 major / 5 minor

Summary. The paper proposes an offline coverage path planning (CPP) method for UAV visual inspection that operates directly on path primitives rather than on discrete viewpoints. The method voxelizes the target structure, generates via-points in a dilated-and-subtracted sampling region, connects via-point pairs into path primitives, precomputes visibility per primitive, encodes the result in a Primitive Coverage Graph (PCG), and searches the graph with a Greedy Neighborhood Search (GNS) to find a connected, branchless path meeting a prescribed coverage ratio. Simulation with Drake and Octomap and one real-world DJI field test are presented, and the reported flight-path lengths are 18.4% and 29.2% shorter on average than two viewpoint-based baselines.

Significance. The path-primitive formulation is a genuine and sensible departure from the prevailing viewpoint-selection-plus-TSP pipeline, and the modular design, the edge-encoded visibility representation, and the inclusion of a real drone field test are clear strengths. The reported savings are suggestive, but the central evidence is incomplete: the achieved coverage ratio is never quantified, no variance is reported for the path lengths, and the greedy search has no termination or feasibility guarantee. Because the central claim is that the method satisfies coverage requirements while reducing flight time, these gaps are load-bearing. If the authors supply quantitative coverage verification, statistical support, and a convergence or repair mechanism for the search, the contribution would be solid and of interest to the robotics and inspection-planning communities.

major comments (3)

[Section IV-B, Table II] The central quantitative claim—18.4% and 29.2% reductions in flight time—is not tied to a demonstrated coverage level. Table II reports only mean path lengths, and Section IV-B states that coverage is validated only through the reconstructed voxel visualizations in Fig. 10. For an apples-to-apples comparison, the paper must report the achieved coverage ratio (e.g., fraction of surface patches visible or fraction of target voxels reconstructed) for the proposed method and for both baselines in the Drake-Octomap simulation, ideally per trial and with standard deviations. Without this, the savings could reflect incomplete coverage rather than improved efficiency.
[Section III-E, Algorithm 3] The Greedy Neighborhood Search has no termination or feasibility guarantee. The while loop (δ≤δd) assumes that the randomly sampled PCG contains a connected, branchless path whose union reaches δd, but Algorithms 1–2 provide no completeness or connectivity guarantee for a given sampling density. In particular, FindTraj (Eq. 4) may reach a via-point whose neighbor set contains no edge with positive incremental coverage, or no unvisited neighbor, before δd is reached; the algorithm then cannot proceed. The paper should either prove termination/completeness under explicit sampling assumptions, add an expansion/repair mechanism, or report an empirical study over sampling densities showing how often and at what cost δd is reached.
[Section IV-B, Table II] The results are averages over 10 trials reported without variance or statistical testing, so the 18.4% and 29.2% improvements cannot be distinguished from run-to-run noise. Report standard deviations (or confidence intervals) and, ideally, a paired test across the same target models.

minor comments (5)

[Algorithm 3, line 2] Line 2 computes δ←FindCoverage(G,e) using e before it is initialized in line 3; reorder the initialization or remove this line.
[Equations (1)–(2) and Algorithm 1] The symbol dmax is used both for the maximum range of the local potential field in Eq. (2) and for the maximum distance for path-primitive sampling in Algorithm 1; use distinct symbols to avoid ambiguity.
[Section IV-B, Fig. 10] The reconstructed voxel models are shown only visually; a quantitative comparison with the ground-truth voxel model (e.g., precision/recall or IoU) would strengthen the coverage validation.
[Section IV-C] For the field test, the paper mentions that 87 pictures were sampled from the video and used for SFM reconstruction, but no coverage metric is reported for the real-world result; a quantitative completeness measure would help.
[Throughout] There are typographical and formatting issues, including 'UA Vs' in the Related Work, 'papaer' in the Field Test section, and inconsistent spacing around references; these should be corrected.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the planning derivation is self-contained and the flight-time reductions are experimental outputs, not fitted or definitionally forced quantities.

full rationale

The paper's planning chain is self-contained: Algorithm 1 samples via-points and path primitives, Algorithm 2 constructs the Primitive Coverage Graph with edge lengths and binary visibility vectors, and Algorithm 3 greedily searches that graph. The reported path lengths in Table II are outputs of this search, not parameters fitted to reproduce the claimed 18.4% and 29.2% reductions. The coverage requirement enters as a user-chosen input (delta_d = 99.0% in Table I), and Algorithm 3's loop condition enforces it within the planner's own visibility model; this is a planning constraint, not a circular prediction. The visibility model is stated as an assumption drawn from prior work and similar external work, and the Drake/Octomap simulation and SFM field reconstruction are presented as independent validation attempts. The paper does cite the authors' own prior work for the viewing-direction computation, visibility model, and the VPP-TSP baseline, but these self-citations are component choices and benchmark comparisons rather than load-bearing derivations of the main claim. The possible absence of a termination guarantee for Algorithm 3 and the lack of reported achieved coverage ratios are correctness and evidence concerns, not circularity. No equation or quantity in the paper reduces by construction to an input or to a self-citation.

Assumptions & free parameters 7 free parameters · 5 assumptions · 0 invented entities

All quantities are user-set planning or sensor parameters, not fitted to make the algorithm succeed. The axioms are standard modeling assumptions for offline model-based inspection planning; the greedy heuristic assumption is the least supported and is empirical.

free parameters (7)

Maximum distance for path primitive sampling (d_max) = not reported
Controls which via-point pairs become path primitives; not listed in Table I, directly affects graph connectivity.
Max viewing range (d_vis) = 50 m simulation, 10 m field test
Used in voxel dilation and the visibility model; user-set for the experiments.
Safety distance (d_safe) = 2 m simulation, 0.5 m field test
Defines the inner boundary of the feasible sampling region.
Desired coverage ratio (delta_d) = 99.0%
Termination threshold for the greedy search; chosen by the user.
Number of via-points (n_vp) = not reported; 1,765 and 1,410 primitives generated
Sampling count is user-set but not stated; determines graph size and coverage completeness.
Visibility sampling density per path primitive = not reported
Number of viewpoints sampled along each primitive for visibility estimation is not specified, so coverage estimates could vary.
Voxel resolution = not reported
Resolution of the voxel model in via-point sampling affects the sampling region and is not given.

assumptions (5)

domain assumption The target 3D model is available and static during planning and execution.
Stated in the Introduction; justifies offline planning.
domain assumption Visibility from a path can be approximated by binary visibility from finite viewpoints sampled along that path.
Section III-C describes a sampling-based visibility estimate with no error analysis.
domain assumption Straight-line segments with collision checking are executable by the UAV.
Section III-B uses straight lines as the local planner; no dynamics validation for the segments.
domain assumption Surface coverage can be measured as the fraction of triangular patches seen by at least one selected path primitive.
The visibility vector and FindCoverage in the algorithms assume patch-level binary coverage.
ad hoc to paper The greedy incremental coverage-per-distance heuristic returns an acceptable inspection path.
GNS is a heuristic with no optimality or completeness argument; validated only empirically.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Coverage Path Planning using Path Primitive Sampling and Primitive Coverage Graph for Visual Inspection." pith.science (2026). https://pith.science/paper/ZCJL7HTT

@misc{pith2026190802901,
  author       = {Pith},
  title        = {Pith review of: Coverage Path Planning using Path Primitive Sampling and Primitive Coverage Graph for Visual Inspection},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/ZCJL7HTT}},
  note         = {Machine review of arXiv:1908.02901}
}

read the original abstract

Planning the path to gather the surface information of the target objects is crucial to improve the efficiency of and reduce the overall cost, for visual inspection applications with Unmanned Aerial Vehicles (UAVs). Coverage Path Planning (CPP) problem is often formulated for these inspection applications because of the coverage requirement. Traditionally, researchers usually plan and optimize the viewpoints to capture the surface information first, and then optimize the path to visit the selected viewpoints. In this paper, we propose a novel planning method to directly sample and plan the inspection path for a camera-equipped UAV to acquire visual and geometric information of the target structures as a video stream setting in complex 3D environment. The proposed planning method first generates via-points and path primitives around the target object by using sampling methods based on voxel dilation and subtraction. A novel Primitive Coverage Graph (PCG) is then proposed to encode the topological information, flying distances, and visibility information, with the sampled via-points and path primitives. Finally graph search is performed to find the resultant path in the PCG to complete the inspection task with the coverage requirements. The effectiveness of the proposed method is demonstrated through simulation and field tests in this paper.

Figures

Figures reproduced from arXiv: 1908.02901 by the authors.

**Figure 2.** The Block Diagram of the Proposed Coverage Planning Framework [PITH_FULL_IMAGE:figures/full_fig_p002_2.png] view at source ↗

**Figure 3.** Voxelizing the polygonal model of the target structure [PITH_FULL_IMAGE:figures/full_fig_p003_3.png] view at source ↗

Figures from the paper (8 more)

**Figure 4.** Figure 4: Generating sampling region for via-points: the orig [PITH_FULL_IMAGE:figures/full_fig_p003_4.png]

**Figure 6.** Figure 6: Visibility estimation of path primitive, green areas are [PITH_FULL_IMAGE:figures/full_fig_p004_6.png]

**Figure 5.** Figure 5: Simplified 2D illustration of the path primitives and [PITH_FULL_IMAGE:figures/full_fig_p004_5.png]

**Figure 7.** Figure 7: Illustration of sampled path primitives and via-points [PITH_FULL_IMAGE:figures/full_fig_p004_7.png]

**Figure 8.** Figure 8: Visualization of planned UAV inspection path around [PITH_FULL_IMAGE:figures/full_fig_p006_8.png]

**Figure 9.** Figure 9: Simulation Flowchart with Drake and Octomap [PITH_FULL_IMAGE:figures/full_fig_p006_9.png]

**Figure 12.** Figure 12: Visualization of 3D reconstruction results of the field [PITH_FULL_IMAGE:figures/full_fig_p007_12.png]

**Figure 11.** Figure 11: Visualization of the Planned Path for the Experiment [PITH_FULL_IMAGE:figures/full_fig_p007_11.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

31 extracted references · 29 canonical work pages

[1]

Sampling-based view planning for 3d visual coverage task with unmanned aerial vehicle,

W. Jing, J. Polden, W. Lin, and K. Shimada, “Sampling-based view planning for 3d visual coverage task with unmanned aerial vehicle,” in IEEE/RSJ International Conference on Intelligent Robots and Systems. IEEE, 2016, pp. 1808–1815

work page 2016
[2]

Structural inspection path planning via iterative viewpoint resampling with application to aerial robotics,

A. Bircher, K. Alexis, M. Burri, P. Oettershagen, S. Omari, T. Mantel, and R. Siegwart, “Structural inspection path planning via iterative viewpoint resampling with application to aerial robotics,” in IEEE International Conference on Robotics and Automation . IEEE, 2015, pp. 6423–6430

work page 2015
[3]

Heterogeneous vehicles routing for water canal damage assessment,

D. Deng, T. Pang, P. Palli, F. Shu, and K. Shimada, “Heterogeneous vehicles routing for water canal damage assessment,” in IEEE/RSJ International Conference on Intelligent Robots and Systems . IEEE, 2018, pp. 2375–2382

work page 2018
[4]

A survey on coverage path planning for robotics,

E. Galceran and M. Carreras, “A survey on coverage path planning for robotics,” Robotics and Autonomous Systems , vol. 61, no. 12, pp. 1258–1276, 2013

work page 2013
[5]

View planning for 3d shape reconstruction of buildings with unmanned aerial vehicles,

W. Jing, J. Polden, P. Y . Tao, W. Lin, and K. Shimada, “View planning for 3d shape reconstruction of buildings with unmanned aerial vehicles,” in International Conference on Control, Automation, Robotics and Vision . IEEE, 2016, pp. 1–6

work page 2016
[6]

Three-dimensional coverage planning for an underwater inspection robot,

B. Englot and F. S. Hover, “Three-dimensional coverage planning for an underwater inspection robot,”The International Journal of Robotics Research, vol. 32, no. 9-10, pp. 1048–1073, 2013

work page 2013
[7]

Model-based view planning,

W. R. Scott, “Model-based view planning,” Machine Vision and Applications, vol. 20, no. 1, pp. 47–69, 2009

work page 2009
[8]

View planning for automated 3d object reconstruction inspection,

W. Scott, G. Roth, and J.-F. Rivest, “View planning for automated 3d object reconstruction inspection,” ACM Computing Surveys , vol. 35, no. 1, 2003

work page 2003

Show all 31 references

[9]

Active vision in robotic systems: A survey of recent developments,

S. Chen, Y . Li, and N. M. Kwok, “Active vision in robotic systems: A survey of recent developments,” The International Journal of Robotics Research, vol. 30, no. 11, pp. 1343–1377, 2011

2011
[10]

Robotic task sequenc- ing problem: A survey,

S. Alatartsev, S. Stellmacher, and F. Ortmeier, “Robotic task sequenc- ing problem: A survey,” Journal of Intelligent & Robotic Systems , vol. 80, no. 2, pp. 279–298, 2015

2015
[11]

Coverage planning for robotic vision applications in complex 3d environment,

W. Jing, “Coverage planning for robotic vision applications in complex 3d environment,” Ph.D. dissertation, Carnegie Mellon University, 2017

2017
[12]

Planning for complete sensor coverage in inspection,

G. H. Tarbox and S. N. Gottschlich, “Planning for complete sensor coverage in inspection,” Computer Vision and Image Understanding , vol. 61, no. 1, pp. 84–111, 1995

1995
[13]

View plan- ning for 3d object reconstruction with a mobile manipulator robot,

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

2014
[14]

Online inspection path planning for autonomous 3d modeling using a micro-aerial vehicle,

S. Song and S. Jo, “Online inspection path planning for autonomous 3d modeling using a micro-aerial vehicle,” in IEEE International Conference on Robotics and Automation . IEEE, 2017, pp. 6217– 6224

2017
[15]

A reinforcement learning approach to the view planning problem

M. D. Kaba, M. G. Uzunbas, and S.-N. Lim, “A reinforcement learning approach to the view planning problem.” in Conference on Computer Vision and Pattern Recognition (CVPR) , 2017, pp. 5094–5102

2017
[16]

Automatic sensor placement for model-based robot vision,

S. Chen and Y . Li, “Automatic sensor placement for model-based robot vision,” IEEE Transactions on Systems, Man, and Cybernetics, Part B: Cybernetics, vol. 34, no. 1, pp. 393–408, 2004

2004
[17]

Sampling-based coverage motion planning for industrial inspection application with redundant robotic system,

W. Jing, J. Polden, C. F. Goh, M. Rajaraman, W. Lin, and K. Shimada, “Sampling-based coverage motion planning for industrial inspection application with redundant robotic system,” in IEEE/RSJ International Conference on Intelligent Robots and Systems. IEEE, 2017, pp. 5211– 5218

2017
[18]

View planning problem with combined view and traveling cost,

P. Wang, R. Krishnamurti, and K. Gupta, “View planning problem with combined view and traveling cost,” in IEEE International Conference on Robotics and Automation . IEEE, 2007, pp. 711–716

2007
[19]

A survey on inspecting structures using robotic systems,

R. Almadhoun, T. Taha, L. Seneviratne, J. Dias, and G. Cai, “A survey on inspecting structures using robotic systems,” International Journal of Advanced Robotic Systems , vol. 13, no. 6, p. 1729881416663664, 2016

2016
[20]

Coverage for robotics–a survey of recent results,

H. Choset, “Coverage for robotics–a survey of recent results,” Annals of mathematics and artiﬁcial intelligence , vol. 31, no. 1-4, pp. 113– 126, 2001

2001
[21]

A survey of sensor planning in computer vision,

K. A. Tarabanis, P. K. Allen, and R. Y . Tsai, “A survey of sensor planning in computer vision,” IEEE Transactions on Robotics and Automation, vol. 11, no. 1, pp. 86–104, 1995

1995
[22]

Submodular trajectory optimization for aerial 3d scanning

M. Roberts, S. Shah, D. Dey, A. Truong, S. N. Sinha, A. Kapoor, P. Hanrahan, and N. Joshi, “Submodular trajectory optimization for aerial 3d scanning.” in International Conference on Computer Vision (ICCV), 2017, pp. 5334–5343

2017
[23]

Asymp- totically optimal inspection planning using systems with differential constraints,

G. Papadopoulos, H. Kurniawati, and N. M. Patrikalakis, “Asymp- totically optimal inspection planning using systems with differential constraints,” in IEEE International Conference on Robotics and Au- tomation. IEEE, 2013, pp. 4126–4133

2013
[24]

S. M. LaValle, Planning algorithms . Cambridge university press, 2006

2006
[25]

Prob- abilistic roadmaps for path planning in high-dimensional conﬁguration spaces,

L. E. Kavraki, P. ˇSvestka, J.-C. Latombe, and M. H. Overmars, “Prob- abilistic roadmaps for path planning in high-dimensional conﬁguration spaces,” IEEE TRANSACTIONS ON ROBOTICS AND AUTOMATION, vol. 12, no. 4, 1996

1996
[26]

Bubble mesh: automated triangular meshing of non-manifold geometry by sphere packing,

K. Shimada and D. C. Gossard, “Bubble mesh: automated triangular meshing of non-manifold geometry by sphere packing,” inProceedings of the third ACM symposium on Solid modeling and applications . ACM, 1995, pp. 409–419

1995
[27]

Drake: A planning, control, and analysis toolbox for nonlinear dynamical systems,

R. Tedrake and the Drake Development Team, “Drake: A planning, control, and analysis toolbox for nonlinear dynamical systems,” 2016. [Online]. Available: https://drake.mit.edu

2016
[28]

OctoMap: An efﬁcient probabilistic 3D mapping framework based on octrees,

A. Hornung, K. M. Wurm, M. Bennewitz, C. Stachniss, and W. Burgard, “OctoMap: An efﬁcient probabilistic 3D mapping framework based on octrees,” Autonomous Robots , 2013, software available at http://octomap.github.com. [Online]. Available: http: //octomap.github.com

2013
[29]

VisualSFM: A visual structure from motion system,

C. Wu, “VisualSFM: A visual structure from motion system,” URL http://ccwu. me/vsfm/, 2011

2011
[30]

Multi-view reconstruction preserving weakly-supported surfaces,

M. Jancosek and T. Pajdla, “Multi-view reconstruction preserving weakly-supported surfaces,” in IEEE Conference on Computer Vision and Pattern Recognition. IEEE, 2011, pp. 3121–3128

2011
[31]

Model-based coverage motion planning for industrial 3d shape in- spection applications,

W. Jing, J. Polden, P. Y . Tao, C. F. Goh, W. Lin, and K. Shimada, “Model-based coverage motion planning for industrial 3d shape in- spection applications,” in IEEE Conference on Automation Science and Engineering. IEEE, 2017, pp. 1293–1300

2017

Pith tools

Reviewed August 14, 2026 · model on record in the stance chip above.

[1] [1]

Sampling-based view planning for 3d visual coverage task with unmanned aerial vehicle,

W. Jing, J. Polden, W. Lin, and K. Shimada, “Sampling-based view planning for 3d visual coverage task with unmanned aerial vehicle,” in IEEE/RSJ International Conference on Intelligent Robots and Systems. IEEE, 2016, pp. 1808–1815

work page 2016

[2] [2]

Structural inspection path planning via iterative viewpoint resampling with application to aerial robotics,

A. Bircher, K. Alexis, M. Burri, P. Oettershagen, S. Omari, T. Mantel, and R. Siegwart, “Structural inspection path planning via iterative viewpoint resampling with application to aerial robotics,” in IEEE International Conference on Robotics and Automation . IEEE, 2015, pp. 6423–6430

work page 2015

[3] [3]

Heterogeneous vehicles routing for water canal damage assessment,

D. Deng, T. Pang, P. Palli, F. Shu, and K. Shimada, “Heterogeneous vehicles routing for water canal damage assessment,” in IEEE/RSJ International Conference on Intelligent Robots and Systems . IEEE, 2018, pp. 2375–2382

work page 2018

[4] [4]

A survey on coverage path planning for robotics,

E. Galceran and M. Carreras, “A survey on coverage path planning for robotics,” Robotics and Autonomous Systems , vol. 61, no. 12, pp. 1258–1276, 2013

work page 2013

[5] [5]

View planning for 3d shape reconstruction of buildings with unmanned aerial vehicles,

W. Jing, J. Polden, P. Y . Tao, W. Lin, and K. Shimada, “View planning for 3d shape reconstruction of buildings with unmanned aerial vehicles,” in International Conference on Control, Automation, Robotics and Vision . IEEE, 2016, pp. 1–6

work page 2016

[6] [6]

Three-dimensional coverage planning for an underwater inspection robot,

B. Englot and F. S. Hover, “Three-dimensional coverage planning for an underwater inspection robot,”The International Journal of Robotics Research, vol. 32, no. 9-10, pp. 1048–1073, 2013

work page 2013

[7] [7]

Model-based view planning,

W. R. Scott, “Model-based view planning,” Machine Vision and Applications, vol. 20, no. 1, pp. 47–69, 2009

work page 2009

[8] [8]

View planning for automated 3d object reconstruction inspection,

W. Scott, G. Roth, and J.-F. Rivest, “View planning for automated 3d object reconstruction inspection,” ACM Computing Surveys , vol. 35, no. 1, 2003

work page 2003