Pith. sign in

REVIEW 3 major objections 4 minor 55 references

MEF-Explore: Communication-Constrained Multi-Robot Entropy-Field-Based Exploration

T0 review · 3 major / 4 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read Robot teams explore faster under patchy links via entropy-field goals

desk verdict A practical two-layer communication model for constrained UGV exploration with a real mathematical flaw in the advertised implicit rendezvous mechanism. read the letter →

arxiv 2505.23376 v1 pith:VB466AP2 submitted 2025-05-29 cs.RO

Khattiya Pongsirijinda , Zhiqiang Cao , Billy Pik Lik Lau , Ran Liu , Chau Yuen , U-Xuan Tan This is my paper
classification cs.RO
keywords Communicationconstraintscooperatingrobotsdistributedexplorationentropy-field-basedinformationsharingmulti-robot
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

This paper claims that a team of ground robots can explore an unknown environment faster and more reliably than existing methods even when high-speed communication is intermittent, provided that positions are always exchanged over a low-bandwidth link and exploration goals are chosen by minimizing a combined entropy that mixes frontier uncertainty with robot proximity. The method, MEF-Explore, treats high-speed map merging as opportunistic, happening only when robots come within range, and uses the entropy field to trigger implicit rendezvous when a robot's local frontier pool is nearly exhausted. The authors demonstrate the claim in simulations across two maps, three benchmark methods, team sizes from two to fifteen robots, and communication cut-off ranges from two metres to infinity, and in a real three-robot deployment. In the real-world experiments, MEF-Explore completes exploration 21.32% faster and succeeds 16.67% more often than the potential-field baseline MWF-CN. The practical point is that perfect communication is the exception in real deployments, so a method designed around patchy links addresses the operating condition that actually occurs.

What carries the argument

The load-bearing object is the reformed total entropy $H_{total}(i,p,q)$ defined as the sum of a frontier entropy $H_f(p,q)$ and a robot entropy $H_r(i,p)$. $H_f$ measures the uncertainty of a frontier centroid $q$ at point $p$, scaled by cluster size and an inverse 8-sector modified wavefront distance; $H_r$ measures the presence of other robots inside the sensing range $d_s$, with a relaxation distance $\sigma_r$ and colored noise $\chi$ added to avoid local optima. The Appendix B theorem is the mechanism that turns this scalar field into a rendezvous policy: in the limit of a single tiny frontier cluster, $H_f \to 0$, so the goal becomes the minimizer of $H_r$, which the proof identifies with other robots' positions. The dynamic graph $G=(V,E)$, with edges added when robots come within $r_{comm}$, is the companion mechanism that decides when those rendezvous lead to actual map merging.

What would settle it

Compute the minimizer of $H_r(i,p)$ over $p$ for a robot with a nearly empty frontier pool using the paper's parameters ($k_r=1$, $\sigma_r=0.6$, $\alpha=2$, $\sigma_d=0.035$); if the minimizer is consistently not a teammate's location but a sensor-range boundary or a noise minimum, the rendezvous premise is refuted. Alternatively, compare exploration time with $H_r$ removed from the goal selector: if exploration time does not rise or map-merging frequency does not fall, rendezvous is not the source of the reported gains.

Watch

Extended reading notes

Core claim

The central claim, stated on the paper's own terms, is that the entropy of frontiers and the entropy of robots can be fused into a single goal-selection function that does the work that explicit rendezvous scheduling and centralized coordination usually do. Each robot evaluates $H_{total}(i,p,q)=H_f(p,q)+H_r(i,p)$ over candidate points $p$ in its local map; $H_f$ rewards points near large frontier clusters at short modified wavefront distance, and $H_r$ draws the robot toward regions within its sensor range where other robots are available. When a robot has almost no frontiers left, the frontier term vanishes and the robot's goal becomes the minimizer of $H_r$ alone, which the paper's Theorem 1 argues is the position of another robot, producing an implicit rendezvous that enables map merging. The duration-adaptive goal-assigning module releases a new goal only when the robot has arrived or traveled for a reference duration, avoiding the goal-overloading that makes navigation stacks jerk or stall. On the evidence presented, robots using this scheme explore faster and more consistently than the MWF-CN, MMPF, and GVGExp baselines in every simulated scenario, and 21.32% faster with a 16.67% higher success rate than MWF-CN on real ground robots.

Load-bearing premise

The rendezvous benefit rests on the claim that when a robot has almost no frontiers left, minimizing $H_r$ pulls it to another robot's position; if the real $H_r$ landscape instead favors the edge of sensor range or a noise-created minimum, the theorem's conclusion fails and one of the two credited speedup mechanisms loses its foundation.

Editorial extensions

If this is right

  • A robot team can keep exploring without any centralized coordinator: low-speed position broadcasts preserve a weakest form of connectivity, and map merging happens whenever high-speed links appear.
  • A robot low on frontiers will spontaneously move toward teammates rather than wander, increasing the frequency of map merges without a precomputed meeting schedule.
  • Exploration time and success rate improve as the high-speed communication range $r_{comm}$ grows, and the gains over the GVGExp baseline grow with $r_{comm}$, indicating that better connectivity amplifies the benefit of the entropy-field goals.
  • The goal-assigning module prevents navigation-stack overload, which the paper credits for part of the speedup and for the higher success rate in both simulation and real deployment.
  • The method scales to 10 and 15 simulated robots while retaining an advantage over the potential-field baselines, and it still outperforms them when all inter-robot communication is switched off.

Reading between the lines

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

  • Beyond the paper: if the rendezvous theorem is the source of the gain, a simpler controller that tells a robot to meet when its local frontier count is small might capture part of the benefit; an ablation that disables $H_r$ while keeping $H_f$ and goal assignment would separate the rendezvous contribution from the goal-selection contribution.
  • Beyond the paper: because $H_r$ only activates inside sensor range, the mechanism is local by construction; extending it to a meet-at-a-remembered-position rule would let the same entropy logic handle rendezvous when teammates are out of sensing range, a regime the paper does not test.
  • Beyond the paper: the fixed colored-noise parameters ($\alpha=2$, $\sigma_d=0.035$) were imported from the potential-field baseline; adapting noise variance to the frontier-depletion rate is a natural testable extension that could reduce local-optima effects in larger maps.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 4 minor

Summary. The paper proposes MEF-Explore, a distributed multi-robot exploration method for communication-constrained settings. Robots share positions at all times via low-speed communication and merge local maps only when within a high-speed communication range r_comm. Exploration is driven by a composite entropy field combining a frontier-entropy term H_f and a robot-entropy term H_r, together with a duration-adaptive goal-assignment module. The authors claim that H_r triggers implicit rendezvous when frontiers are scarce, and they provide a formal theorem in Appendix B for this behavior. The method is evaluated in Gazebo on two maps with 2–15 robots against MWF-CN, MMPF, and GVGExp, and in a real-world three-robot experiment, reporting faster exploration time, lower RSD_T, and higher success rate in all compared scenarios.

Significance. If the claims hold, the paper addresses a practically important problem—multi-robot exploration under intermittent communication—and combines several useful ingredients: a two-layer communication model with dynamic graphs, a distributed entropy-field objective, opportunistic map merging, and a real-robot deployment with UWB/WiFi-based communication emulation. The empirical study is comparatively broad: two simulation maps, multiple team sizes, a scalability test with 10 and 15 robots, a communication-loss robustness test, and a real-world experiment. These are genuine strengths. However, the paper's theoretical support for the implicit-rendezvous mechanism (Theorem 1) is not valid as stated, and the statistical support for the headline performance claims is thin because the results rest on 20 simulation rounds per condition and 10 real-world rounds with no significance tests or confidence intervals. The central performance claim is therefore defensible only as an empirical observation, not as a mechanism-verified result.

major comments (3)
  1. [Appendix B, Eqs. (11), (23), (24)] Theorem 1 does not follow from the equations as written. The proof shows only that H_total reduces to H_r when frontiers vanish, but H_r(i,p) in Eq. (11) contains no term involving Pos_cur(j) for any j ≠ i; it is a function only of robot i's own position and the candidate point p. The field is therefore radially symmetric around robot i and carries no information about where teammates are. Moreover, for d_cur(i,p) < d_s the denominator d_cur(i,p) − d_s is negative, so H_r → −∞ as p approaches the sensor-range boundary; with bounded colored noise, the unconstrained minimizer of H_r is on the d_s ring, not at another robot's position. The limit argument in Eqs. (22)–(23) only reduces the objective; it cannot turn a teammate-independent field into a rendezvous field. In addition, the limit (N_C, C_q) → (1, 0) is internally inconsistent: if there is only one centroid (N_C = 1), that centroid must contain all remaining frontiers, so C_q = N_f rather than tending to 0. Since Sections III-A and IV-B explicitly credit 'implicit rendezvous' as a mechanism for the observed speedups, this flaw is load-bearing. The authors should either reformulate H_r so that it genuinely depends on teammate positions, or remove the claim that rendezvous is formally proven and present it as an observed heuristic behavior.
  2. [Section IV-B, Table II, Fig. 5, Fig. 6] The central empirical claim—that MEF-Explore 'surpasses the existing ones regarding exploration time and success rate in all scenarios'—is not supported by statistical evidence. Each condition is based on only 20 simulation rounds, and the real-world experiment uses 10 rounds. No significance tests, confidence intervals, or effect-size uncertainties are reported. Many improvements in Table II are small (e.g., RSD_T improvements of 0.65%, 2.47%, and 3.13% for Map 1 with MWF-CN) and success-rate differences are exactly 0% in several rows. The observed pattern may be real, but with these sample sizes the percentage differences are within the range of random variation. Please add per-condition hypothesis tests (or at least bootstrap confidence intervals) and report them for all metrics, or temper the claim to describe the observed outcomes in the tested runs.
  3. [Section III-B, Eq. (9)] The substitution H_f = log N_f = log(N_C C_q) is only valid when every frontier cluster has the same number of frontiers C_q. In general N_f = Σ_m C_m, and N_C C_q overcounts or undercounts the total number of frontiers whenever cluster sizes differ. Since this entropy term is the core of the exploration objective, the derivation should be clarified: either define H_f for centroid q as the Shannon entropy of that cluster (log C_q) plus a cluster-count term, or justify the approximation and show that the performance comparisons are insensitive to it. As written, the frontier entropy is not the Shannon entropy of the frontiers claimed in Eq. (8).
minor comments (4)
  1. [Algorithm 2 and Eq. (12)] g_new(i) is defined as argmin_p H_total(i,p,q), but q is not an optimization variable and the algorithm does not specify how q is selected for each p. Please clarify whether q is the nearest centroid to p, the centroid associated with the current goal, or evaluated for all centroids.
  2. [Eq. (11)] The colored-noise term χ_p^i(α, σ_d) is not formally defined (distribution, boundedness, or sampling procedure). Because this term is added inside the sensor range only, it creates a discontinuity at d_s and can alter the argmin of the total field; its properties should be stated precisely.
  3. [Section III-A] The statement that H_f ≫ H_r 'in general' is not quantified. With k_f(N_r) = 2N_r − 3 and k_r = 1, the relative magnitudes depend on d*(p,q), cluster size C_q, and the noise term; a parametric condition or empirical verification would make the claim precise.
  4. [Appendix A and Section VI-A] Appendix A assumes a zero-error map-merging operator ⊞, while Section VI-A acknowledges UWB ranging errors below 0.5 m and possible map degradation. Please state explicitly how the ideal-property assumption relates to the real-robot deployment and whether the error magnitude affects the theoretical properties in practice.

Circularity Check

1 steps flagged · score 4.0 of 10

Rendezvous theorem is the design goal restated: Eq. (11) defines H_r using only robot i's own distance to candidate point p, so the argmin in Eq. (24) is a self-centered sensor-range-boundary point, not a teammate.

  1. self definitional [Section III-A, Eq. (11), Eq. (12); Appendix B, Theorem 1, Eqs. (22)-(24)]
    "We also intentionally formulate our entropies to activate some rendezvous at appropriate times ... robot rendezvous will be spontaneously triggered by the effect of Hr when the robots can detect just a small number of frontiers ... We provide formal proof of this rendezvous manner in Appendix B. ... According to eq. (11), since Hr(i,p) exists in roboti's surroundings, roboti itself and other robots within its sensor range ds will then be attracted to travel to each other's positions, i.e., rendezvous."

    The proof reduces H_total to H_r and then sets g_new(i)=argmin_p H_r(i,p). But Eq. (11) defines H_r(i,p) only through d_cur(i,p)=||Pos_cur(i)-p||; there is no term involving Pos_cur(j) for any j≠i. The objective is therefore radially symmetric around robot i and carries no information about where teammates are. With d_cur<d_s, the term k_r σ_r N_r/(d_cur-d_s) logN_r diverges toward -infinity as p approaches the sensor-range boundary, so with bounded colored noise the unconstrained minimizer is on the d_s ring, not at a teammate.

full rationale

The headline performance claims are empirical: simulations and real-robot experiments compare MEF-Explore against MWF-CN, MMPF, and GVGExp using exploration time, RSD_T, and success rate, so the central comparison has independent grounding. The circular element is the supporting 'implicit rendezvous' theorem. The paper states it intentionally formulated the entropies to activate rendezvous, then Appendix B claims to prove the behavior by letting frontiers vanish so H_total reduces to H_r and g_new becomes argmin_p H_r(i,p). However, Eq. (11) contains no other-robot position term; the limiting objective is a field centered on robot i itself, and its minimum lies at the sensor-range boundary rather than at a teammate. The theorem therefore does not derive rendezvous from the model—it re-asserts the design goal. Because the paper credits the speedups partly to these rendezvous rounds, one mechanism is backed by a self-definitional step rather than by the equations. This is partial circularity in a supporting claim, not a collapse of the entire empirical contribution, hence score 4.

Assumptions & free parameters 6 free parameters · 8 assumptions · 0 invented entities

The central performance claim rests on a handful of hand-picked constants and a communication model that is assumed rather than measured. No new physical entities are introduced. The main theoretical props are standard math plus domain assumptions about communication and map merging; the ad hoc entries are the equal-cluster simplification in eq. (9) and the borrowed colored noise.

free parameters (6)
  • k_f (frontier entropy scale) = 2*Nr - 3
    Scale factor in eq. (9) chosen by hand to 'progressively' scale entropy values; no derivation or sensitivity analysis.
  • k_r (robot entropy scale) = 1
    Scale factor in eq. (11) chosen by hand.
  • sigma_r (relaxation distance) = 0.6
    Parameter from prior work [8] controlling the preferred inter-robot distance in eq. (11).
  • alpha (noise color) = 2
    Colored noise exponent from [8] added to Hr to escape local optima.
  • sigma_d (noise variance) = 0.035
    Noise variance from [8] added to Hr.
  • k_ref (goal reassignment fraction) = 0.1
    Scale factor in eq. (16) to set reference travel duration; chosen by hand.
assumptions (8)
  • standard math Standard Shannon entropy definitions (eqs. 8 and 10).
    The total entropy in eq. (7) is built from Shannon entropy of frontiers and robots; the paper treats these as established.
  • standard math L'Hopital's rule is used in Appendix B to evaluate the limit as C_q approaches 0.
    Invoked without proof; standard calculus result.
  • domain assumption All robots are homogeneous with identical sensors, velocity, and map resolution.
    Stated in Section III: 'a group of N_r homogeneous robots with identical capabilities and sensor range d_s'.
  • domain assumption Low-speed communication is always available between every pair of robots; high-speed communication exists exactly when d_cur < r_comm (eqs. 1-2).
    This is the communication model on which the whole method rests; it is asserted, not validated against a radio model.
  • domain assumption The map merging operator bottomplus is error-free, idempotent, commutative, and associative (Appendix A, eqs. 19-21).
    The paper states 'we assume that there is no error from merging' for the properties; real map stitching from [50] has alignment errors.
  • ad hoc to paper The special form H_f = log(N_C * C_q) in eq. (9) represents the frontier entropy, implying all clusters have equal frontier count C_q.
    Substituting N_f = N_C * C_q into eq. (8) is only valid under equal cluster sizes; the paper does not justify this.
  • ad hoc to paper Colored noise chi_p^i(alpha, sigma_d) in eq. (11) helps avoid local optima.
    Borrowed from prior work [8] without analysis of its effect on the rendezvous or goal selection.
  • domain assumption Frontier centroids are generated by the continuity-based clustering method of [9].
    Section III-A states this is applied to detected frontiers; the paper does not analyze sensitivity to clustering.

how reviews work

0 comments
Cite this review

Pith. "Pith review of MEF-Explore: Communication-Constrained Multi-Robot Entropy-Field-Based Exploration." pith.science (2026). https://pith.science/paper/VB466AP2

@misc{pith2026250523376,
  author       = {Pith},
  title        = {Pith review of: MEF-Explore: Communication-Constrained Multi-Robot Entropy-Field-Based Exploration},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/VB466AP2}},
  note         = {Machine review of arXiv:2505.23376}
}
read the original abstract

Collaborative multiple robots for unknown environment exploration have become mainstream due to their remarkable performance and efficiency. However, most existing methods assume perfect robots' communication during exploration, which is unattainable in real-world settings. Though there have been recent works aiming to tackle communication-constrained situations, substantial room for advancement remains for both information-sharing and exploration strategy aspects. In this paper, we propose a Communication-Constrained Multi-Robot Entropy-Field-Based Exploration (MEF-Explore). The first module of the proposed method is the two-layer inter-robot communication-aware information-sharing strategy. A dynamic graph is used to represent a multi-robot network and to determine communication based on whether it is low-speed or high-speed. Specifically, low-speed communication, which is always accessible between every robot, can only be used to share their current positions. If robots are within a certain range, high-speed communication will be available for inter-robot map merging. The second module is the entropy-field-based exploration strategy. Particularly, robots explore the unknown area distributedly according to the novel forms constructed to evaluate the entropies of frontiers and robots. These entropies can also trigger implicit robot rendezvous to enhance inter-robot map merging if feasible. In addition, we include the duration-adaptive goal-assigning module to manage robots' goal assignment. The simulation results demonstrate that our MEF-Explore surpasses the existing ones regarding exploration time and success rate in all scenarios. For real-world experiments, our method leads to a 21.32% faster exploration time and a 16.67% higher success rate compared to the baseline.

Figures

Figures reproduced from arXiv: 2505.23376 by the authors.

Figure 1
Figure 1. Overview of our proposed MEF-Explore I. INTRODUCTION C OLLABORATIVE multi-robot exploration systems have emerged as a prevailing approach for autonomous un￾known area exploration owing to significant advancements and practical capabilities in various domains, such as search and rescue operations [1], [2], environmental monitoring and patrolling [3], [4], and several types of automation. So far, diverse methods have … view at source ↗
Figure 2
Figure 2. Graphical representation of the proposed information-sharing strategy [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Graphical representation of three-robot exploration by the proposed entropy-field-based exploration strategy from Robot 1’s point of view when robots [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (12 more)
Figure 4
Figure 4. Figure 4: Environments for simulation. The red-outlined, green-filled, blue [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Simulation results on Map 1 of our proposed MEF-Explore, the MWF-CN, the MMPF, and the GVGExp [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 5
Figure 5. Figure 5: Simulation results on Map 1 of our proposed MEF-Explore, the MWF-CN, the MMPF, and the GVGExp [PITH_FULL_IMAGE:figures/full_fig_p010_5.png]
Figure 6
Figure 6. Figure 6: Simulation results on Map 2 of our proposed MEF-Explore, the MWF-CN, the MMPF, and the GVGExp [PITH_FULL_IMAGE:figures/full_fig_p010_6.png]
Figure 6
Figure 6. Figure 6: Simulation results on Map 2 of our proposed MEF-Explore, the MWF-CN, the MMPF, and the GVGExp [PITH_FULL_IMAGE:figures/full_fig_p011_6.png]
Figure 6
Figure 6. Figure 6: Simulation results on Map 2 of our proposed MEF-Explore, the MWF-CN, the MMPF, and the GVGExp [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]
Figure 7
Figure 7. Figure 7: Initial positions for exploration with 10 and 15 robots [PITH_FULL_IMAGE:figures/full_fig_p013_7.png]
Figure 8
Figure 8. Figure 8: Simulation results on Map 1 of our proposed MEF-Explore, the [PITH_FULL_IMAGE:figures/full_fig_p013_8.png]
Figure 9
Figure 9. Figure 9: Simulation results on Map 2 of our proposed MEF-Explore, the [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]
Figure 10
Figure 10. Figure 10: Simulation results on Map 1 and Map 2 of our proposed MEF [PITH_FULL_IMAGE:figures/full_fig_p014_10.png]
Figure 11
Figure 11. Figure 11: Robot and environment setup for three-robot exploration in real-world environment [PITH_FULL_IMAGE:figures/full_fig_p015_11.png]
Figure 12
Figure 12. Figure 12: Experimental results on Real Environment of our proposed MEF-Explore and the MWF-CN [PITH_FULL_IMAGE:figures/full_fig_p015_12.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

55 extracted references · 54 canonical work pages

  1. [1]

    Generating task reallocation suggestions to handle contingencies in human-supervised multi-robot missions,

    S. Al-Hussaini, J. M. Gregory, and S. K. Gupta, “Generating task reallocation suggestions to handle contingencies in human-supervised multi-robot missions,”IEEE Trans. Autom. Sci. Eng., vol. 21, no. 1, pp. 367–381, 2024

    work page 2024

  2. [3]

    Near-optimal area- coverage path planning of energy-constrained aerial robots with appli- cation in autonomous environmental monitoring,

    K. R. Jensen-Nau, T. Hermans, and K. K. Leang, “Near-optimal area- coverage path planning of energy-constrained aerial robots with appli- cation in autonomous environmental monitoring,”IEEE Trans. Autom. Sci. Eng., vol. 18, no. 3, pp. 1453–1468, 2021

    work page 2021

  3. [4]

    Decen- tralized control of a heterogeneous human–robot team for exploration and patrolling,

    M. Aggravi, G. Sirignano, P. R. Giordano, and C. Pacchierotti, “Decen- tralized control of a heterogeneous human–robot team for exploration and patrolling,”IEEE Trans. Autom. Sci. Eng., vol. 19, no. 4, pp. 3109– 3125, 2022

    work page 2022

  4. [5]

    Autonomous robotic exploration based on multiple rapidly-exploring randomized trees,

    H. Umari and S. Mukhopadhyay, “Autonomous robotic exploration based on multiple rapidly-exploring randomized trees,” in2017 IEEE/RSJ Int. Conf. Intell. Robots Syst., Sept. 2017

    work page 2017

  5. [6]

    Multi-AGV's temporal memory-based RRT explo- ration in unknown environment,

    B. P. L. Lauet al., “Multi-AGV's temporal memory-based RRT explo- ration in unknown environment,”IEEE Robot. Autom. Lett., vol. 7, no. 4, pp. 9256–9263, Oct. 2022

    work page 2022

  6. [7]

    Multi-robot multi-room exploration with geometric cue extraction and circular decomposition,

    S. Kim, M. Corah, J. Keller, G. Best, and S. Scherer, “Multi-robot multi-room exploration with geometric cue extraction and circular decomposition,”IEEE Robot. Autom. Lett., vol. 9, no. 2, pp. 1190–1197, Feb. 2024

    work page 2024

  7. [8]

    Distributed multi-robot potential-field-based exploration with submap-based mapping and noise-augmented strategy,

    K. Pongsirijindaet al., “Distributed multi-robot potential-field-based exploration with submap-based mapping and noise-augmented strategy,” Robot. Auton. Syst., vol. 179, p. 104752, 2024

    work page 2024

  8. [9]

    SMMR-explore: SubMap-based multi-robot exploration system with multi-robot multi-target potential field exploration method,

    J. Yuet al., “SMMR-explore: SubMap-based multi-robot exploration system with multi-robot multi-target potential field exploration method,” in2021 IEEE Int. Conf. Robot. Automat., May 2021

    work page 2021

Show all 55 references
  1. [10]

    Cooperative exploration of networked multi-robot systems using minimal information entropy,

    C.-F. Chang and C.-C. Tsai, “Cooperative exploration of networked multi-robot systems using minimal information entropy,” in2015 IEEE Int. Conf. Netw. Sens. Control, Apr. 2015

    2015

  2. [11]

    Multi-robot exploration and coverage: Entropy- based adaptive maps with adjacency control laws,

    M. Scott and K. Jerath, “Multi-robot exploration and coverage: Entropy- based adaptive maps with adjacency control laws,” in2018 Annu. Am. Control Conf., June 2018

    2018

  3. [12]

    Coordination methods for entropy-based multi-agent exploration under sparsity constraints,

    C. Manss, D. Shutin, A. Viseras, and G. Leus, “Coordination methods for entropy-based multi-agent exploration under sparsity constraints,” in 2019 IEEE Int. Workshop Comput. Adv. Multi-Sensor Adapt. Process., Dec. 2019

    2019

  4. [13]

    Mapping and exploration in a hierarchical heterogeneous multi-robot system using limited capability robots,

    P. Chand and D. A. Carnegie, “Mapping and exploration in a hierarchical heterogeneous multi-robot system using limited capability robots,”Rob. Auton. Syst., vol. 61, no. 6, pp. 565–579, 2013

    2013

  5. [14]

    Cure: A hierarchical framework for multi-robot au- tonomous exploration inspired by centroids of unknown regions,

    Q. Biet al., “Cure: A hierarchical framework for multi-robot au- tonomous exploration inspired by centroids of unknown regions,”IEEE Trans. Autom. Sci. Eng., vol. 21, no. 3, pp. 3773–3786, 2024

    2024

  6. [15]

    Learning to explore using active neural slam,

    D. S. Chaplot, D. Gandhi, S. Gupta, A. Gupta, and R. Salakhutdinov, “Learning to explore using active neural slam,” inInt. Conf. Learn. Represent., 2020

    2020

  7. [16]

    The surprising effectiveness of PPO in cooperative multi agent games,

    C. Yuet al., “The surprising effectiveness of PPO in cooperative multi agent games,” in2022 Conf. Adv. Neural Inf. Process Syst., 2022

    2022

  8. [17]

    Communication-efficient multi-robot exploration using coverage-biased distributed q-learning,

    E. Latif and R. Parasuraman, “Communication-efficient multi-robot exploration using coverage-biased distributed q-learning,”IEEE Robot. Autom. Lett., vol. 9, no. 3, pp. 2622–2629, Mar. 2024

    2024

  9. [18]

    IR 2: Implicit rendezvous for robotic exploration teams under sparse intermittent connectivity,

    D. M. S. Tan, Y . Ma, J. Liang, Y . Cao, and G. Sartoretti, “IR 2: Implicit rendezvous for robotic exploration teams under sparse intermittent connectivity,” in2024 IEEE/RSJ Int. Conf. Intell. Robots Syst., 2024

    2024

  10. [19]

    Enabling large-scale heterogeneous collaboration with op- portunistic communications,

    F. Cladera, Z. Ravichandran, I. D. Miller, M. Ani Hsieh, C. J. Taylor, and V . Kumar, “Enabling large-scale heterogeneous collaboration with op- portunistic communications,” in2024 IEEE Int. Conf. Robot. Automat., 2024, pp. 2610–2616

    2024

  11. [20]

    Representation gran- ularity enables time-efficient autonomous exploration in large, complex worlds,

    C. Cao, H. Zhu, Z. Ren, H. Choset, and J. Zhang, “Representation gran- ularity enables time-efficient autonomous exploration in large, complex worlds,”Sci. Robot., vol. 8, no. 80, July 2023

    2023

  12. [21]

    Fast multi-UA V decentralized exploration of forests,

    L. Bartolomei, L. Teixeira, and M. Chli, “Fast multi-UA V decentralized exploration of forests,”IEEE Robot. Autom. Lett., vol. 8, no. 9, pp. 5576–5583, Sept. 2023

    2023

  13. [22]

    RACER: Rapid collaborative exploration with a decentralized multi-UA V system,

    B. Zhou, H. Xu, and S. Shen, “RACER: Rapid collaborative exploration with a decentralized multi-UA V system,”IEEE Trans. Robot., pp. 1–20, 2023

    2023

  14. [23]

    Decentralized multi-agent exploration with limited inter-agent communications,

    H. J. He, A. Koppel, A. S. Bedi, D. J. Stilwell, M. Farhood, and B. Biggs, “Decentralized multi-agent exploration with limited inter-agent communications,” in2023 IEEE Int. Conf. Robot. Automat., May 2023

    2023

  15. [24]

    Coordinated multi-agent ex- ploration, rendezvous, & task allocation in unknown environments with limited connectivity,

    L. Bramblett, R. Peddi, and N. Bezzo, “Coordinated multi-agent ex- ploration, rendezvous, & task allocation in unknown environments with limited connectivity,” in2022 IEEE/RSJ Int. Conf. Intell. Robots Syst., Oct. 2022

    2022

  16. [25]

    Meeting-merging-mission: A multi-robot coordinate framework for large-scale communication-limited exploration,

    Y . Gaoet al., “Meeting-merging-mission: A multi-robot coordinate framework for large-scale communication-limited exploration,” in2022 IEEE/RSJ Int. Conf. Intell. Robots Syst., Oct. 2022

    2022

  17. [26]

    MR-TopoMap: Multi- robot exploration based on topological map in communication restricted environment,

    Z. Zhang, J. Yu, J. Tang, Y . Xu, and Y . Wang, “MR-TopoMap: Multi- robot exploration based on topological map in communication restricted environment,”IEEE Robot. Autom. Lett., pp. 1–8, 2022

    2022

  18. [27]

    GVGExp: Communication-constrained multi- robot exploration system based on generalized voronoi graphs,

    K. Masaba and A. Q. Li, “GVGExp: Communication-constrained multi- robot exploration system based on generalized voronoi graphs,” in2021 Int. Symp. Multi-Robot Multi-Agent Syst., Nov. 2021

    2021

  19. [28]

    Intermittent connectivity for exploration in communication-constrained multi-agent systems,

    F. Klaesson, P. Nilsson, A. D. Ames, and R. M. Murray, “Intermittent connectivity for exploration in communication-constrained multi-agent systems,” in2020 ACM/IEEE Int. Conf. Cyber-Phys. Syst., 2020, pp. 196–205

    2020

  20. [29]

    Multirobot exploration of communication-restricted environments: A survey,

    F. Amigoni, J. Banfi, and N. Basilico, “Multirobot exploration of communication-restricted environments: A survey,”IEEE Intell. Syst., vol. 32, no. 6, pp. 48–57, Nov. 2017

    2017

  21. [30]

    A critical review of communi- cations in multi-robot systems,

    J. Gielis, A. Shankar, and A. Prorok, “A critical review of communi- cations in multi-robot systems,”Curr . Robot. Rep., vol. 3, no. 4, pp. 213–225, Aug. 2022

    2022

  22. [31]

    Distributed state estimation using intermittently connected robot networks,

    R. Khodayi-mehr, Y . Kantaros, and M. M. Zavlanos, “Distributed state estimation using intermittently connected robot networks,”IEEE Trans. Robot., vol. 35, no. 3, pp. 709–724, 2019

    2019

  23. [32]

    Distributed multi-robot coor- dination in area exploration,

    W. Sheng, Q. Yang, J. Tan, and N. Xi, “Distributed multi-robot coor- dination in area exploration,”Robot. Auton. Syst., vol. 54, no. 12, pp. 945–955, 2006

    2006

  24. [33]

    Multirobot tree and graph exploration,

    P. Brass, F. Cabrera-Mora, A. Gasparri, and J. Xiao, “Multirobot tree and graph exploration,”IEEE Trans. Robot., vol. 27, no. 4, pp. 707–717, 2011

    2011

  25. [34]

    Multi-robot taboo-list exploration of unknown structured environments,

    M. Andries and F. Charpillet, “Multi-robot taboo-list exploration of unknown structured environments,” in2015 IEEE/RSJ Int. Conf. Intell. Robots Syst., 2015, pp. 5195–5201

    2015

  26. [35]

    Collaboration in multi-robot exploration: To meet or not to meet?

    T. Andre and C. Bettstetter, “Collaboration in multi-robot exploration: To meet or not to meet?”J. Intell. Robot. Syst., vol. 82, pp. 325–337, 2016

    2016

  27. [36]

    A mathematical theory of communication,

    C. E. Shannon, “A mathematical theory of communication,”Bell Syst. Tech. J., vol. 27, no. 3, pp. 379–423, 1948

    1948

  28. [37]

    A mathematical theory of communication,

    C. E. Shannon, “A mathematical theory of communication,”Bell Syst. Tech. J., vol. 27, no. 4, pp. 623–656, 1948

    1948

  29. [38]

    Potential information fields for mobile robot exploration,

    J. Vallv ´e and J. Andrade-Cetto, “Potential information fields for mobile robot exploration,”Robot. Auton. Syst., vol. 69, pp. 68–79, July 2015

    2015

  30. [39]

    Fast frontier-based information-driven autonomous exploration with an mav,

    A. Dai, S. Papatheodorou, N. Funk, D. Tzoumanikas, and S. Leuteneg- ger, “Fast frontier-based information-driven autonomous exploration with an mav,” in2020 IEEE Int. Conf. Robot. Automat., May 2020

    2020

  31. [40]

    Autonomous robotic exploration using occupancy grid maps and graph slam based on shannon and r ´enyi entropy,

    H. Carrillo, P. Dames, V . Kumar, and J. A. Castellanos, “Autonomous robotic exploration using occupancy grid maps and graph slam based on shannon and r ´enyi entropy,” in2015 IEEE Int. Conf. Robot. Automat., May 2015

    2015

  32. [41]

    Autonomous robotic exploration using a utility function based on r ´enyi’s general theory of entropy,

    H. Carrillo, P. Dames, V . Kumar, and J. A. Castellanos, “Autonomous robotic exploration using a utility function based on r ´enyi’s general theory of entropy,”Auton. Robots, vol. 42, no. 2, pp. 235–256, Aug. 2017

    2017

  33. [42]

    High resolution maps from wide angle sonar,

    H. Moravec and A. Elfes, “High resolution maps from wide angle sonar,” in1985 IEEE Int. Conf. Robot. Automat., vol. 2, 1985, pp. 116–121

    1985

  34. [43]

    Explore-bench: Data sets, metrics and evaluations for frontier-based and deep-reinforcement-learning-based autonomous ex- ploration,

    Y . Xuet al., “Explore-bench: Data sets, metrics and evaluations for frontier-based and deep-reinforcement-learning-based autonomous ex- ploration,” in2022 IEEE Int. Conf. Robot. Automat., May 2022

    2022

  35. [44]

    A decentralized hybrid method for goal assignment in multi-robot exploration,

    B. Tong, Q. Liu, and C. Dai, “A decentralized hybrid method for goal assignment in multi-robot exploration,” in2020 IEEE Int. Conf. Inf. Technol. Big Data Artif. Intell., Nov. 2020

    2020

  36. [45]

    Autonomous exploration using a tree structure for goal selection,

    B. Abonyi-T ´oth and A. Nagy, “Autonomous exploration using a tree structure for goal selection,” in2023 IEEE Int. Symp. Appl. Comput. Intell. Inf., May 2023

    2023

  37. [46]

    Decentralized goal assignment and safe trajectory generation in multirobot networks via multiple lyapunov functions,

    D. Panagou, M. Turpin, and V . Kumar, “Decentralized goal assignment and safe trajectory generation in multirobot networks via multiple lyapunov functions,”IEEE Trans. Autom. Control, vol. 65, no. 8, pp. 3365–3380, Aug. 2020

    2020

  38. [47]

    Graph neural network for decentralized multi-robot goal assignment,

    M. Goarin and G. Loianno, “Graph neural network for decentralized multi-robot goal assignment,”IEEE Robot. Autom. Lett., vol. 9, no. 5, pp. 4051–4058, May 2024

    2024

  39. [48]

    Design and use paradigms for gazebo, an open-source multi-robot simulator,

    N. Koenig and A. Howard, “Design and use paradigms for gazebo, an open-source multi-robot simulator,” in2004 IEEE/RSJ Int. Conf. Intell. Robots Syst., vol. 3, 2004, pp. 2149–2154

    2004

  40. [49]

    A quantitative study of tuning ros gmapping parameters and their effect on performing indoor 2D SLAM,

    Y . Abdelrasoul, A. B. S. H. Saman, and P. Sebastian, “A quantitative study of tuning ros gmapping parameters and their effect on performing indoor 2D SLAM,” in2016 IEEE Int. Symp. Robot. Manuf. Autom., 2016, pp. 1–6

    2016

  41. [50]

    map merge 2d,

    A. Athukorala, “map merge 2d,” https://github.com/acachathuranga/ map merge 2d/tree/ros melodic, 2023

    2023

  42. [51]

    Map-merging for multi-robot system,

    J. H ¨orner, “Map-merging for multi-robot system,” Bachelor’s thesis, Charles University, Prague, 2016

    2016

  43. [53]

    NLOS identification and compensation for UWB ranging based on obstruction classification,

    K. Wen, K. Yu, and Y . Li, “NLOS identification and compensation for UWB ranging based on obstruction classification,” in2017 25th Eur . Signal Process Conf., 2017, pp. 2704–2708

    2017

  44. [54]

    A survey on blockchain in robotics: Issues, opportunities, challenges and future directions,

    U. S. P. S. Aditya, R. Singh, P. K. Singh, and A. Kalla, “A survey on blockchain in robotics: Issues, opportunities, challenges and future directions,”J. Netw. Comput. Appl., vol. 196, p. 103245, 2021

    2021

  45. [55]

    Congestion- aware policy synthesis for multirobot systems,

    C. Street, S. P ¨utz, M. M ¨uhlig, N. Hawes, and B. Lacerda, “Congestion- aware policy synthesis for multirobot systems,”IEEE Trans. Robot., vol. 38, no. 1, pp. 262–280, 2022

    2022

  46. [56]

    Multi-robot exploration in task allocation problem,

    R. J. Alitappeh and K. Jeddisaravi, “Multi-robot exploration in task allocation problem,”Appl. Intell., vol. 52, p. 2189–2211, 2022. APPENDIX In this Appendix, we provide the mathematical properties and proof of our proposed method’s components mentioned in the previous sectio...

    2022

  47. [2014]

    He is currently working as a Senior Research Fellow at Nanyang Technological University

    From 2015 to 2024, he was a Research Fellow at Singapore University of Technology and Design. He is currently working as a Senior Research Fellow at Nanyang Technological University. His research interests include robotics and SLAM. 17 This paper has been accepted for publicat...

    2024

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.