Pith. sign in

REVIEW 4 major objections 5 minor 28 references

Bidirectional planner grafts around bad edges to find paths faster

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 ↗

G3T* combines edge-grafting, greedy GuILD-style beacon subsets, and an improvement-based adaptive sampler, reporting faster convergence and lower path costs than prior planners.

T0 review reviewed 2026-08-05 challenge →

load-bearing objection A novel grafting idea with solid experiments, but the asymptotic optimality claim is unsupported and there's a real bug in Algorithm 4. the 4 major comments →

arxiv 2508.19776 v1 pith:WIVNEEZH submitted 2025-08-27 cs.RO

Tree-Based Grafting Approach for Bidirectional Motion Planning with Local Subsets Optimization

Liding Zhang , Yao Ling , Zhenshan Bing , Fan Wu , Sami Haddadin , Alois Knoll This is my paper
classification cs.RO
keywords bidirectional motion planningtree graftinggreedy GuILD subsetsLebesgue measure beacon selectionhistorical samplingasymptotic optimalitysampling-based planningsingle-query planners
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

G3T* is a bidirectional sampling-based motion planner that claims to solve the tree-connection bottleneck in asymmetric searches. When the forward and reverse trees meet through an edge that collides, older planners raise the lazy-check resolution and restart the reverse tree; G3T* instead gathers common neighbors of the edge's two endpoints and tries to bypass the obstacle with a two-edge graft, so the search continues without a restart. The paper also shrinks the sampling focus to a pair of hyperellipsoidal "greedy GuILD" subsets built around a beacon vertex chosen to minimize total Lebesgue measure, and adjusts how many samples go into those subsets versus the broader informed set based on current and historical cost improvements. In R2-to-R8 benchmarks and a real 8-DoF kitchen-manipulation task, it reports faster time-to-first-solution, lower median path costs, and higher success rates than six existing planners. If these claims hold, G3T* offers a way to get the speed of bidirectional search without paying for failed connection attempts.

Core claim

The paper proposes that a failed lazy-checked connection between the forward and reverse trees need not abort the reverse search: the two endpoints' common neighbors inside the current random geometric graph form candidate edge-pairs, sorted by a cost-plus-effort heuristic, and the best collision-free pair grafts the trees back together. Around this, it defines greedy GuILD subsets whose beacon vertex minimizes the sum of the Lebesgue measures (hypervolumes) of the two front/back hyperellipsoids, so local densification targets the smallest admissible region. To avoid being trapped in that local region, a historical sampling rule compares the current cost-improvement ratio against the historical average.

What carries the argument

The load-bearing objects are (1) the common-neighbor queue QcomNeigh of an invalid edge's two endpoints, which yields graft edge-pairs replacing one bad edge with two good ones; (2) the greedy GuILD subsets XG2f and XG2b, two hyperellipsoids whose beacon vertex is selected by minimizing total Lebesgue measure and whose transverse diameters are greedy costs from front/back path vertices; and (3) the historical distributed sampling rule that allocates batch samples between XG2 and the informed set Xbf using current-versus-historical cost improvement ratios. The random-geometric-graph radius $r(m)$ supplies the neighborhoods and the density needed for connectivity and, in principle, optimality.

Load-bearing premise

The asymptotic-optimality guarantee rests on an unproven assertion that, over many improvements, the current cost improvement will eventually exceed the historical average, even though sampling is deliberately non-uniform and skewed away from the informed set; if that assertion fails, the guarantee is unsupported.

What would settle it

Set up a benchmark where the strictly shorter optimal corridor lies inside the informed set but outside the greedy GuILD subsets, then track the number of samples placed into $\mathrm{Xbf} \setminus \mathrm{XG2}$ per batch as the number of cost improvements grows. If Eq. (24) is wrong, the planner will starve the informed set of samples, final cost will plateau above $c^*$, and repeated seeds will fail to show $\limsup P(\text{CCI} > \text{HCI}) = 1$. That plateau, or a direct counterexample where adaptive non-uniform sampling violates the uniform-density premise of Lemmas 56/71/72 of [10], would settle the optimality claim.

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

If this is right

  • Bidirectional planners can recover from a failed connection without a full reverse-tree restart, which removes a known source of variance in single-query planning times.
  • Sampling effort concentrates in the smallest admissible local subsets, so convergence of path cost should be faster in narrow-passage and high-dimensional problems.
  • The historical sampling allocation is the mechanism claimed to keep asymptotic optimality alive despite the non-uniform, locally biased distribution.
  • In time-limited settings, G3T* should produce a usable initial path more often than planners that restart the reverse search, giving higher success rates for real robot tasks.
  • The implementation fits the standard random-geometric-graph planner interface, so it can be adopted as a drop-in single-query planner.

Where Pith is reading between the lines

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

  • Beyond the paper, grafting is a local repair operator that could in principle be applied whenever any graph edge fails validation, not only at bidirectional tree joins, making it a general edge-replacement primitive for sampling-based planners.
  • The "minimum Lebesgue measure" beacon criterion is a geometric-tightness proxy; a natural extension is to weight the measure by obstacle occupancy or by an anisotropic cost to handle non-Euclidean planning metrics.
  • The historical sampling rule resembles a bandit-style allocation between exploration (informed set) and exploitation (greedy GuILD); one testable extension is to measure how quickly the exploration signal arrives in problems with multiple disconnected narrow passages.
  • A direct stress test would compare G3T* against FIT* on a long-horizon manipulation task where the shortest path lies outside the current greedy GuILD subsets, to see whether the historical sampling rule rediscovers the global region before time runs out.
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

4 major / 5 minor

Summary. The paper proposes G3T*, a bidirectional sampling-based motion planner that combines a tree-grafting mechanism for invalid edge connections, a greedy GuILD subset sampler based on Lebesgue measure, and an adaptive historical sampling strategy. The authors claim that the historical sampling strategy 'ensures asymptotic optimality' while preserving fast practical convergence, and they support this with benchmark experiments in R2–R8 and a real-world 8-DoF manipulation task. The central theoretical assertion is that G3T* inherits the asymptotic optimality guarantees of Karaman–Frazzoli via Lemmas 56, 71, and 72, and this assertion is the main load-bearing element of the paper. The experimental methodology is generally sound, but the theoretical argument is not.

Significance. If the asymptotic optimality claim were rigorously established, G3T* would be a useful contribution to high-dimensional bidirectional planning, combining grafting, adaptive local subsets, and non-uniform sampling. The empirical side has real strengths: 100 runs per benchmark, use of Planner Developer Tools, 99% nonparametric confidence intervals, experiments across R2–R8, and a public OMPL implementation. However, the paper's headline theoretical contribution—'ensuring asymptotic optimality'—is not supported by the arguments presented. The cited Karaman–Frazzoli lemmas are applied to an adaptive non-uniform sampler they do not cover, and the key probability limit in Eq. (24) is asserted without proof. These are not presentation issues; they are load-bearing gaps in the central claim.

major comments (4)
  1. [Section III-C, Eq. (24)] The linchpin of the asymptotic optimality argument is the asserted limit lim sup_{Nimp→∞} P(Υ_CCI > Υ_HCI) := 1. No derivation is given. The accompanying text argues that Υ_HCI 'gradually approaches zero,' but since Υ_HCI is the empirical average of the same Υ_CCI sequence, this presupposes that the sequence of current-cost improvements has positive mass or that path costs converge—which is exactly what the asymptotic optimality proof is supposed to establish. This is circular unless a separate probabilistic lower bound on future improvement is proven. The claim 'This mechanism guarantees asymptotic optimality' therefore does not follow from Eq. (24).
  2. [Section III-D, Eq. (26)] Eq. (26) is justified by 'By Lemmas 56, 71, and 72 in [10].' Those lemmas are for the uniform-sampling RRT* algorithm and its rewiring analysis. G3T*, however, uses an adaptive, non-uniform sampling distribution over XG2 and X_bf in Algorithm 4, and the sampler changes over time based on cost improvements. The hypotheses of the cited lemmas are not shown to hold for this adaptive sampler, so the asymptotic optimality conclusion is unsupported. Similarly, Eq. (25) is justified by 'leveraging uniform sampling,' but after an initial path is found Algorithm 4 does not perform uniform sampling over Xfree; it samples only inside XG2 and X_bf. The proof needs a direct argument that the resulting sample distribution eventually provides the needed cover of the informed set, rather than an invocation of lemmas from a different sampling regime.
  3. [Algorithm 4, lines 12–14] The allocation formulas are internally inconsistent with the experimental configuration. With θ_thold=3 (as stated in Section IV), ⌊1/θ_thold⌋ = 0, so both line 12 (M_XG2 ← ⌊1/θ_thold⌋·llevel·M_curr) and line 14 (M_XG2 ← (θ_thold·⌊1/θ_thold⌋)·M_curr) give M_XG2=0 in the marginal-improvement branch. This means no samples are allocated to the GuILD subsets in that branch, contradicting the text and Figure 3, which describe greedy sampling in XG2. Since the experiments set θ_thold=3, the implemented planner likely differs from the algorithm as written, or the formulas need correction. This mismatch undermines the connection between the theoretical description and the empirical results.
  4. [Section III-D, Eq. (25)] Probabilistic completeness is asserted rather than proven. The equation states lim_{i→∞} P({V_F ∪ V_R} ∩ X_goal ≠ ∅)=1 merely from 'By leveraging uniform sampling,' but the algorithm's sampling is non-uniform after the first solution is found, and the grafting/rewiring operations are not shown to preserve the conditions needed for completeness. This is a secondary but still load-bearing gap: the paper's overall theoretical guarantee includes both completeness and asymptotic optimality, and neither is established by the presented arguments.
minor comments (5)
  1. [Algorithm 4 input] The input line lists 'cprev - Cost of previous solution path, cprev - Cost of current solution path'; the second occurrence should be 'ccurr'.
  2. [Section III-B, Eq. (20)–(21)] The notation 'arg max' is used where a maximum value is intended (cgreedyf and cgreedyb are scalar costs). Use 'max' instead of 'arg max' for these definitions.
  3. [Section III-B, text near Eq. (15)] Typo: 'font-path cost' should be 'front-path cost.'
  4. [Figure 4 caption] The caption contains grammatical errors ('Start configration Goal configration - PlanningTime 1.0s') and should be corrected.
  5. [Section IV-A, Table I] The table reports median initial-time improvements but not confidence intervals for those medians; given that PDT provides confidence intervals, reporting them would strengthen the comparative claim.

Circularity Check

0 steps flagged

No significant circularity: the asymptotic-optimality proof is incomplete and misapplies Karaman-Frazzoli lemmas, but no claim reduces by construction to its own inputs.

full rationale

The paper's central empirical claims are benchmarked against external planners and are not fitted predictions. The theoretical AO claim is the only candidate for circularity. Section III-C defines the current and historical improvement ratios (Eqs. 22-23) from observed cost improvements and then asserts Eq. (24) without proof; Section III-D invokes Lemmas 56/71/72 of [10] to conclude Eq. (26). This chain is not a definitional reduction: Eq. (24) is not derived from the definitions, and it does not by itself state or imply that the path cost converges to c*. Even if the historical improvement rate tends to zero (which is automatic for any nonincreasing cost sequence), that does not force the adaptive non-uniform sampling distribution to be dense in the informed set, nor does it make the Karaman-Frazzoli uniform-sampling lemmas applicable to Algorithm 4's sampler. Thus the argument is unsupported and the cited lemmas are misapplied, but the conclusion is not equivalent to the inputs by construction. The self-citations (e.g., FIT* [20]) supply the algorithmic scaffold but are not used as the sole evidence for AO; the external [10] lemmas are substantive rather than self-referential. Missing proof and mathematical invalidity are correctness risks, not circularity, under the given rules.

Axiom & Free-Parameter Ledger

4 free parameters · 6 axioms · 0 invented entities

The paper introduces no new physical entities, forces, or dimensions. The free parameters are standard algorithm hyperparameters plus the ad hoc theta_thold. The load-bearing axioms are the unproven Eq. (24) and the transfer of the Karaman-Frazzoli AO theorem to a setting where its hypotheses are not checked.

free parameters (4)
  • theta_thold (level-up threshold) = 3
    Controls when marginal improvements escalate the sampling level. Set ad hoc to 3, and the AO argument depends on it, though the floor(1/theta_thold) formula makes it vacuous.
  • RGG constant eta = 1.001
    Used in the connection radius formula, Eq. (3). Chosen by benchmark convention, not derived.
  • rewire factor = 1.2
    Used in experiments for rewiring-based planners. Standard value, not justified theoretically.
  • max edge length per dimension = 0.5 (R2), 1.25 (R4), 3.0 (R8)
    Set for the planning benchmarks; affects the graph structure and solution quality.
axioms (6)
  • standard math Karaman-Frazzoli asymptotic optimality of RRT* (Lemmas 56, 71, 72 in [10])
    Invoked in Section III-D Eq. (26) to conclude AO. Those lemmas require uniform sampling and rewiring, which the adaptive non-uniform sampler does not satisfy without proof.
  • standard math RGG connectivity via radius r(m) (Penrose [17], Solovey et al. [24])
    Used in Eqs. (2)-(4) to define graph structure and justify near-certain connectivity as m grows.
  • standard math Informed set Xbf contains the optimal path (Gammell et al.)
    Used in Eqs. (10)-(13) to focus sampling; standard property of admissible heuristics.
  • ad hoc to paper Eq. (24): lim sup P(CCI > HCI) = 1
    Asserted in Section III-C as the mechanism guaranteeing asymptotic optimality. It assumes the cost sequence converges, which is the target property, and is not derived.
  • ad hoc to paper Historical sampling preserves asymptotic optimality
    The claim that adaptively shifting samples between XG2 and Xbf yields an asymptotically dense sampler over a set containing the optimum is stated without proof in Sections III-C and III-D.
  • domain assumption Exact collision checking and valid lazy checking
    Full and lazy collision checks are treated as exact; the lazy reverse search is assumed to produce a tree that can eventually be connected. Standard in this literature.

reviewed 2026-08-05 · how reviews work

0 comments
Cite this review

Pith. "Pith review of Tree-Based Grafting Approach for Bidirectional Motion Planning with Local Subsets Optimization." pith.science (2026). https://pith.science/paper/WIVNEEZH

@misc{pith2026250819776,
  author       = {Pith},
  title        = {Pith review of: Tree-Based Grafting Approach for Bidirectional Motion Planning with Local Subsets Optimization},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WIVNEEZH}},
  note         = {Machine review of arXiv:2508.19776}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Bidirectional motion planning often reduces planning time compared to its unidirectional counterparts. It requires connecting the forward and reverse search trees to form a continuous path. However, this process could fail and restart the asymmetric bidirectional search due to the limitations of lazy-reverse search. To address this challenge, we propose Greedy GuILD Grafting Trees (G3T*), a novel path planner that grafts invalid edge connections at both ends to re-establish tree-based connectivity, enabling rapid path convergence. G3T* employs a greedy approach using the minimum Lebesgue measure of guided incremental local densification (GuILD) subsets to optimize paths efficiently. Furthermore, G3T* dynamically adjusts the sampling distribution between the informed set and GuILD subsets based on historical and current cost improvements, ensuring asymptotic optimality. These features enhance the forward search's growth towards the reverse tree, achieving faster convergence and lower solution costs. Benchmark experiments across dimensions from R^2 to R^8 and real-world robotic evaluations demonstrate G3T*'s superior performance compared to existing single-query sampling-based planners. A video showcasing our experimental results is available at: https://youtu.be/3mfCRL5SQIU

Figures

Figures reproduced from arXiv: 2508.19776 by Alois Knoll, Fan Wu, Liding Zhang, Sami Haddadin, Yao Ling, Zhenshan Bing.

Figure 1
Figure 1. Figure 1: Illustrates the grafting process in asymmetric bidirectional search. [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Illustrates the concept of GuILD and greedy GuILD. The problem [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Flowchart of G3T* integrating historical sampling into grafting and [PITH_FULL_IMAGE:figures/full_fig_p005_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Experimental results from Section IV-B are summarized above. It illustrates the [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Simulated planning problems were illustrated using a 2D represen [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
Figure 1
Figure 1. Figure 1: Top: Percentage of runs that found a solution at any given time with a Clopper￾Pearson (nonparametric) 99% confidence interval. Bottom: Median cost evolution and median of initial solution with nonparametric 99% confidence intervals. 1 (a) Dividing Wall-gaps (DW) in R2 - MaxTime: 0.06s (c) Dividing Wall-gaps (DW) in R4 - MaxTime: 0.16s (e) Dividing Wall-gaps (DW) in R8 - MaxTime: 0.22s (b) Random Rectangle… 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

28 extracted references · 13 canonical work pages

  1. [1]

    Asymptotically optimal sampling-based motion planning methods,

    J. D. Gammell and M. P. Strub, “Asymptotically optimal sampling-based motion planning methods,” Annual Review of Control, Robotics, and Autonomous Systems, vol. 4, pp. 295–318, 2021

    work page 2021

  2. [2]

    Motion planning for robotics: A review for sampling- based planners,

    L. Zhang, K. Cai, Z. Sun, Z. Bing, C. Wang, L. Figueredo, S. Haddadin, and A. Knoll, “Motion planning for robotics: A review for sampling- based planners,” Biomimetic Intelligence and Robotics , vol. 5, no. 1, p. 100207, 2025

    2025

  3. [3]

    Sampling-based motion planning: A comparative review,

    A. Orthey, C. Chamzas, and L. E. Kavraki, “Sampling-based motion planning: A comparative review,” Annual Review of Control, Robotics, and Autonomous Systems , vol. 7, no. 1, 2024

    work page 2024

  4. [4]

    Bidirectional sampling-based motion plan- ning without two-point boundary value solution,

    S. Nayak and M. W. Otte, “Bidirectional sampling-based motion plan- ning without two-point boundary value solution,” IEEE Transactions on Robotics, vol. 38, no. 6, pp. 3636–3654, 2022

    2022

  5. [5]

    A note on two problems in connexion with graphs,

    E. Dijkstra, “A note on two problems in connexion with graphs,” Numerische Mathematik, vol. 1, no. 1, pp. 269–271, 1959

    1959

  6. [6]

    A formal basis for the heuristic determination of minimum cost paths,

    P. E. Hart, N. J. Nilsson, and B. Raphael, “A formal basis for the heuristic determination of minimum cost paths,” IEEE transactions on Systems Science and Cybernetics , vol. 4, no. 2, pp. 100–107, 1968

    1968

  7. [7]

    Rapidly-exploring random trees: A new tool for path planning,

    S. LaValle, “Rapidly-exploring random trees: A new tool for path planning,” Research Report 9811 , 1998

    1998

  8. [8]

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

    L. E. Kavraki, P. Svestka, 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, pp. 566–580, 1996

    1996

  9. [9]

    RRT-connect: An efficient approach to single-query path planning,

    J. Kuffner and S. LaValle, “RRT-connect: An efficient approach to single-query path planning,” in Proceedings 2000 ICRA. Millennium Conference. IEEE International Conference on Robotics and Automa- tion. Symposia Proceedings , vol. 2, 2000, pp. 995–1001 vol.2

    work page 2000

  10. [10]

    Sampling-based algorithms for optimal motion planning,

    S. Karaman and E. Frazzoli, “Sampling-based algorithms for optimal motion planning,” The international journal of robotics research, vol. 30, no. 7, pp. 846–894, 2011

    2011

  11. [11]

    Elliptical k-nearest neighbors: Path optimization via coulomb’s law and invalid vertices in c-space obstacles,

    L. Zhang, Z. Bing, Y . Zhang, K. Cai, L. Chen, F. Wu, S. Haddadin, and A. Knoll, “Elliptical k-nearest neighbors: Path optimization via coulomb’s law and invalid vertices in c-space obstacles,”2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pp. 12 032–12 039, 2024

    2024

  12. [12]

    Informed RRT*: Optimal sampling-based path planning focused via direct sampling of an admissible ellipsoidal heuristic,

    J. D. Gammell, S. S. Srinivasa, and T. D. Barfoot, “Informed RRT*: Optimal sampling-based path planning focused via direct sampling of an admissible ellipsoidal heuristic,” in 2014 IEEE/RSJ international conference on intelligent robots and systems . IEEE, 2014

    2014

  13. [13]

    Informed sam- pling for asymptotically optimal path planning,

    J. D. Gammell, T. D. Barfoot, and S. S. Srinivasa, “Informed sam- pling for asymptotically optimal path planning,” IEEE Transactions on Robotics, vol. 34, no. 4, pp. 966–984, 2018

    2018

  14. [14]

    Batch informed trees (BIT*): Sampling-based optimal planning via the heuristically guided search of implicit random geometric graphs,

    J. D. Gammell, S. S. Srinivasa, and T. D. Barfoot, “Batch informed trees (BIT*): Sampling-based optimal planning via the heuristically guided search of implicit random geometric graphs,” in 2015 IEEE international conference on robotics and automation (ICRA). IEEE, 2015, pp. 3067– 3074

    work page 2015

  15. [15]

    Batch informed trees (BIT*): Informed asymptotically optimal anytime search,

    J. D. Gammell, T. D. Barfoot, and S. S. Srinivasa, “Batch informed trees (BIT*): Informed asymptotically optimal anytime search,” The International Journal of Robotics Research , vol. 39, no. 5, 2020

    2020

  16. [16]

    Fast marching tree: A fast marching sampling-based method for optimal motion planning in many dimensions,

    L. Janson, E. Schmerling, A. Clark, and M. Pavone, “Fast marching tree: A fast marching sampling-based method for optimal motion planning in many dimensions,” The International Journal of Robotics Research , vol. 34, no. 7, pp. 883–921, 2015

    work page 2015

  17. [17]

    Penrose, Random geometric graphs

    M. Penrose, Random geometric graphs . OUP Oxford, 2003, vol. 5

    2003

  18. [18]

    Adaptively informed trees (AIT*) and effort informed trees (EIT*): Asymmetric bidirectional sampling-based path planning,

    M. P. Strub and J. D. Gammell, “Adaptively informed trees (AIT*) and effort informed trees (EIT*): Asymmetric bidirectional sampling-based path planning,” The International Journal of Robotics Research, vol. 41, no. 4, pp. 390–417, 2022

    work page 2022

  19. [19]

    Adaptively informed trees (AIT*): Fast asymptotically optimal path planning through adaptive heuristics,

    M. P. Strub and J. D. Gammell, “Adaptively informed trees (AIT*): Fast asymptotically optimal path planning through adaptive heuristics,” in 2020 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2020, pp. 3191–3198

    work page 2020

  20. [20]

    Flexible informed trees (FIT*): Adaptive batch-size approach in informed sampling-based path planning,

    L. Zhang, Z. Bing, K. Chen, L. Chen, K. Cai, Y . Zhang, F. Wu, P. Krumbholz, Z. Yuan, S. Haddadin, and A. Knoll, “Flexible informed trees (FIT*): Adaptive batch-size approach in informed sampling-based path planning,” 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pp. 3146–3152, 2024

    work page 2024

  21. [21]

    Nearest-neighbourless asymptotically optimal motion planning with fully connected informed trees (FCIT*),

    T. S. Wilson, W. Thomason, Z. Kingston, L. E. Kavraki, and J. D. Gammell, “Nearest-neighbourless asymptotically optimal motion planning with fully connected informed trees (FCIT*),” arXiv, 2024. [Online]. Available: https://arxiv.org/abs/2411.17902

    work page arXiv 2024

  22. [22]

    Energy-efficient path planning of reconfigurable robots in complex environments,

    P. T. Kyaw, A. V . Le, P. Veerajagadheswar, M. R. Elara, T. T. Thu, N. H. K. Nhan, P. Van Duc, and M. B. Vu, “Energy-efficient path planning of reconfigurable robots in complex environments,” IEEE Transactions on Robotics , vol. 38, no. 4, pp. 2481–2494, 2022

    work page 2022

  23. [23]

    GuILD: Guided incremental local densification for accelerated sampling-based motion planning,

    R. Scalise, A. Mandalika, B. Hou, S. Choudhury, and S. S. Srini- vasa, “GuILD: Guided incremental local densification for accelerated sampling-based motion planning,” in 2023 IEEE International Confer- ence on Robotics and Automation (ICRA) , 2023, pp. 10 212–10 218

    work page 2023

  24. [24]

    New perspective on sampling-based motion planning via random geometric graphs,

    K. Solovey, O. Salzman, and D. Halperin, “New perspective on sampling-based motion planning via random geometric graphs,” The International Journal of Robotics Research , vol. 37, no. 10, pp. 1117– 1133, 2018

    work page 2018

  25. [25]

    Benchmarking motion planning algorithms: An extensible infrastructure for analysis and vi- sualization,

    M. Moll, I. A. Sucan, and L. E. Kavraki, “Benchmarking motion planning algorithms: An extensible infrastructure for analysis and vi- sualization,” IEEE Robotics & Automation Magazine , vol. 22, no. 3, pp. 96–102, 2015

    2015

  26. [26]

    Planner developer tools (PDT): Reproducible experiments and statistical analysis for de- veloping and testing motion planners,

    J. D. Gammell, M. P. Strub, and V . N. Hartmann, “Planner developer tools (PDT): Reproducible experiments and statistical analysis for de- veloping and testing motion planners,” in Proceedings of the Workshop on Evaluating Motion Planning Performance (EMPP), IEEE/RSJ Inter- national Conference on Intelligent Robots and Systems (IROS) , 2022

    work page 2022

  27. [27]

    Moveit! task construc- tor for task-level motion planning,

    M. G ¨orner, R. Haschke, H. Ritter, and J. Zhang, “Moveit! task construc- tor for task-level motion planning,” in IEEE International Conference on Robotics and Automation (ICRA) , 2019, pp. 190–196

    2019

  28. [28]

    The open motion planning library,

    I. A. Sucan, M. Moll, and L. E. Kavraki, “The open motion planning library,” IEEE Robotics & Automation Magazine , vol. 19, no. 4, pp. 72–82, 2012

    2012

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