Pith. sign in

REVIEW 3 major objections 4 minor 48 references

A Multi-Agent Reinforcement Learning Approach for Cooperative Air-Ground-Human Crowdsensing in Emergency Rescue

T0 review · 3 major / 4 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read This paper claims that a multi-agent reinforcement learning algorithm, HECTA4ER, using a 'Hard-Cooperative' policy where UGVs recharge low-battery UAVs, raises the average task completion rate in emergency crowdsensing by 18.42% over…

desk verdict A useful MARL application with a broad experimental sweep, but the 18.42% headline only matches a subset of Table II, the convergence proof is not a proof, and no code is released to audit the numbers. read the letter →

arxiv 2505.06997 v1 pith:EA5FPC7I submitted 2025-05-11 cs.AI

Wenhao Lu , Zhengqiu Zhu , Yong Zhao , Yonglin Tian , Junjie Zeng , Jun Zhang , Zhong Liu , Fei-Yue Wang This is my paper · ORCID
classification cs.AI
keywords MobileCrowdsensingCollaborativeSensingEmergencyRescueTaskAllocationPartiallyObservableEnvironmentalStatesMulti-AgentReinforcementLearningHard-CooperativepolicyHECTA
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 task allocation among humans, UAVs, and UGVs in emergency rescue can be modeled as a decentralized partially observable Markov decision process, and that a multi-agent reinforcement learning agent trained with centralized information can allocate tasks effectively under partial observability. The key novel mechanism is a 'Hard-Cooperative' policy in which UGVs drop their own sensing to recharge low-battery UAVs, which the authors argue is essential for sustaining UAV sensing. In simulations the proposed algorithm HECTA4ER achieves on average 18.42% higher task completion rate than baseline methods, and a real-world case study in a community suggests it maintains a higher completion rate in a dynamic setting. A sympathetic reader would care because emergency response requires decisions under limited communication and time pressure, and the paper offers an end-to-end learning approach that does not require a hand-coded planner.

What carries the argument

The load-bearing mechanism is the 'Hard-Cooperative' policy: if a UGV detects that a UAV cannot move due to insufficient power, the UGV immediately goes to replace the UAV's battery instead of executing its own sensing task. This policy is expressed as a deterministic transition rule for the UGV's location. Around this policy, the algorithm HECTA4ER uses a Dec-POMDP formulation with belief states that satisfy the Markov property, a recurrent decision module that summarizes action-observation history in hidden states, an action-filtering mechanism that masks actions outside the movable range to counter sparse rewards, and a mixing network that enforces the monotonicity condition between individual and joint action values.

What would settle it

A controlled field test in which a low-battery UAV is placed outside all UGV detection ranges and obstacles or tasks move between time steps should reproduce the 18.42% completion-rate advantage; if the advantage disappears in that setting, the Hard-Cooperative policy's guaranteed-recharge assumption is doing the work rather than the learning algorithm.

Watch

Extended reading notes

Core claim

The paper's central claim is that the Heterogeneous-Entity Collaborative-Sensing Task Allocation (HECTA) problem in emergency rescue, which is NP-hard, can be solved effectively by a multi-agent reinforcement learning algorithm that explicitly models partial observability and enforces a cooperative energy-management rule. The authors formulate the problem as a Dec-POMDP, prove that belief states based on action-observation history satisfy the Markov property, and use this to justify a recurrent decision module. They then introduce HECTA4ER, built on a Centralized Training with Decentralized Execution architecture, with a convolutional feature extractor, a recurrent policy with action filtering, and a mixing network that combines global and local information under an Individual-Global-Max consistency condition. In simulations across varied scenarios, HECTA4ER achieves an average 18.42% higher task completion rate than four baselines, and in a real-world case study in a 12 $km^{2}$ community it maintains the highest completion rate across different sensing time limits.

Load-bearing premise

The model assumes a static 2D grid with discrete time steps, that each task can be done by exactly one entity type, and that recharging always succeeds whenever a UGV detects a low-battery UAV; if real rescue scenes involve moving obstacles, communication breakdowns, or tasks needing multiple capabilities, the central performance gain may not carry over.

Editorial extensions

If this is right

  • If the 18.42% average improvement replicates, emergency responders can complete substantially more sensing tasks within a fixed time window by using the learned policy rather than greedy or independent reinforcement learning baselines.
  • The 'Hard-Cooperative' policy implies that UGV capacity is reserved for UAV battery support, so systems with a higher UGV-to-UAV ratio should see larger gains, while fleets with too few UGVs may leave UAVs stranded without recharging.
  • Because the problem is formulated as a Dec-POMDP with belief states, the same algorithm structure can be reapplied to other partially observable collaborative sensing tasks with different entity types and energy constraints.
  • The real-world case study suggests that even with only six sensing entities, the learned policy can outperform baselines, indicating that the approach does not require a large fleet to be useful.

Reading between the lines

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

  • The paper's model ties each task to a single entity type; a natural extension would allow multi-capability tasks and ask whether the same architecture can learn synchronized multi-agent arrivals, a setting the current formulation explicitly rules out.
  • The failed replenishment at time T=6 in the case study (UAV outside UGV detection range) suggests a concrete algorithmic improvement the authors do not explore: letting UGVs predict UAV energy depletion before it becomes critical, rather than reacting only when the UAV is already immobile.
  • The convergence proof in Appendix D uses textbook Robbins-Monro conditions with a specific decreasing step size; the practical RMSprop schedule in the experiments is not shown to satisfy those conditions, so the 'finite-step convergence' is best read as an assurance argument rather than a measured guarantee.
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.

Referee Report

3 major / 4 minor

Summary. The paper studies the Heterogeneous-Entity Collaborative-Sensing Task Allocation (HECTA) problem in emergency rescue, modeling it as a Dec-POMDP over a discrete grid and proposing a MARL algorithm called HECTA4ER. The algorithm uses a Centralized Training with Decentralized Execution architecture with convolutional feature extraction, recurrent hidden states for action-observation history, and a mixing network that fuses global and local information. A key design is a 'Hard-Cooperative' policy in which UGVs interrupt their own sensing to recharge low-battery UAVs. The claims include an NP-hardness proof, a Markov property proof for belief states, a convergence theorem, extensive simulations over ten scenarios with four baselines, robustness tests, ablations, and a real-world case study, with a headline result of an average 18.42% increase in task completion rate (TCR) over baseline methods.

Significance. If the claims are supported, the paper offers a useful application of MARL to a realistic heterogeneous crowdsensing problem, combining partial observability, heterogeneous capabilities, energy constraints, and time limits. The most valuable parts are the explicit problem formalism, the concrete baseline comparisons (including MANF-RL-RP and FD-MAPPO), the ablation study, and the attempt to move beyond simulation with a real-world case study. The paper also states its limiting assumptions clearly. However, the central quantitative result is not supported by the tables in the form stated, and the convergence proof contains substantial gaps, so the current version needs revision before the empirical and theoretical contributions can be accepted.

major comments (3)
  1. [Abstract; §V-B, Table II; §VI] The headline claim of an average 18.42% increase in TCR over 'baseline methods' is not supported by the reported tables as written. Recomputing Table II, the mean difference in the 'Training performance' column is about 18.5% across the 17 scenario rows, which can reproduce the 18.42% figure only if that single column and the single baseline MANF-RL-RP are used as the comparison set. Averaging all five columns of Table II (85 comparisons) gives about 11.1%, while the case-study advantage over MANF-RL-RP in Table IV averages about 9.9%; the other baselines (Greedy-SC-RP, FD-MAPPO, HECTA4ER-Voluntary) are not part of the claimed denominator. Furthermore, §V-B states that HECTA4ER 'consistently outperforms MANF-RL-RP in each case,' but Table II rows 6-1 and 8-2 in the 'Changing sensing entity position' column show MANF-RL-RP with higher TCR (43.2 vs. 43.0 and 30.5 vs. 30.4, respectively). The paper must either correct the headline to name the exact comparison set or provide a properly aggregated average over all baselines and scenarios, and it must reconcile or annotate the two negative cells.
  2. [§V-B, Table II] The robustness protocol is under-specified. The text says that after training on a base scenario the algorithm is 'evaluated across 50 new random scenarios,' but Table II does not show 50 scenarios, per-seed results, or confidence intervals; each cell appears to be a single TCR value, and the column labeled 'Training performance' is never defined in the caption or the body. It is therefore impossible to determine whether the reported robustness results are means over 50 random scenarios, means over seeds, or single runs. The paper should state the aggregation procedure, the number of seeds, and the variance or confidence intervals for every cell in Table II.
  3. [Appendix D, Theorem 4] The convergence proof is not valid for the algorithm as implemented. The induction assumes at initialization that the Q-values are already within ε0 of the optimal Q* (Eq. 33), which presupposes the conclusion. The proof then applies the tabular Q-learning update rule (Eqs. 36-37), although HECTA4ER uses neural-network function approximation, target networks, an experience replay buffer, and the mixing losses of Eq. 24, none of which fit the stochastic-approximation framework invoked. Moreover, the Robbins-Monro conditions give asymptotic convergence, not 'convergence within a finite number of steps' as claimed at the end of the proof, and the inequality chain in Eqs. 43-44 requires σ<εt, which is not established. The authors should either remove the finite-step claim and supply a rigorous asymptotic result under stated assumptions, or present the convergence section as empirical and qualitative.
minor comments (4)
  1. [§III-B, Eq. 1] The 'Hard-Cooperative' override is specified only for the case where the UGV detects a UAV in its range with insufficient power; please state explicitly how the rule composes with the action-filtering mechanism of Algorithm 1 when the UGV cannot reach the UAV within one time step or the target cell is occupied by an obstacle.
  2. [Appendix C, Eq. 28] The conditioning set in the normalization factor is written as 'I_{t-1}^t' in the denominator but should be 'I_k^{t-1}'; the same notational issue appears in Eqs. 30-31 and should be corrected.
  3. [§V-A1] The complexity expression O(W×(B+batchsize+TimeLimit)) is not derived and appears to omit per-agent factors and environment-step costs; either provide a derivation or remove the formula.
  4. [Title page and biography] The affiliation block lists 'Yongling Tian' while the author biography uses 'Yonglin Tian'; the authors should harmonize the spelling.

Circularity Check

1 steps flagged · score 2.0 of 10

Self-contained empirical comparison; only minor tautology in 'full cooperativity' theorem.

  1. self definitional [Appendix A, Theorem 1; invoked in Section IV-C (Mixing module)]
    "Theorem 1. Human workers, UAVs, and UGVs operate in a fully cooperative manner. Proof. All entity types share the common overarching objective of maximizing the number of completed sensing tasks within the time limit. Furthermore, the specific “Hard-Cooperative” policy establishes an explicit cooperative interaction between UGVs and UAVs, where UGVs deviate from their sensing tasks to support the UAV battery swapping. Given the shared global objective and defined cooperative interactions, the heterogeneous entities operate under a fully cooperative paradigm."

    The proof's premises already contain the conclusion: 'fully cooperative' is defined by the shared global TCR objective, and the Hard-Cooperative rule is a cooperative interaction designed into the model. No independent property is derived; the theorem is a restatement of the modeling assumption. Because Section IV-C uses this theorem to justify monotonic value decomposition, the support is tautological. However, the monotonicity/IGM machinery itself is cited to external QMIX/QTRAN work, and the paper's headline results are direct empirical comparisons to external baselines, so this step is minor and not load-bearing for the central claim.

full rationale

The derivation chain is otherwise self-contained. The HECTA problem is an externally stated optimization (Eq. 10) with NP-hardness shown by reduction to the orienteering problem; the Dec-POMDP formulation and belief-state Markov proof are standard and do not assume the algorithm's performance. HECTA4ER is evaluated against external baselines (greedy, MANF-RL-RP [30] from a different group, FD-MAPPO [44]) and an internal ablation (HECTA4ER-Voluntary); no parameter is fitted to the comparison metric, and the Hard-Cooperative rule is tested by ablation rather than asserted. Citations to the authors' prior work ([10], [28], [34]) appear only in related-work or parameter-setting contexts and are not load-bearing. The headline 18.42% improvement is under-specified and appears to match only the Training-performance column of Table II against MANF-RL-RP, not the full robustness table or all baselines; this is a reporting/scoping issue, not a reduction by construction. The paper's Discussion explicitly admits modeling limitations, which cuts against overclaiming but does not create circularity. Overall, only one minor self-definitional step exists: Theorem 1's 'proof' of full cooperativity merely restates the shared-objective premise.

Assumptions & free parameters 3 free parameters · 6 assumptions · 0 invented entities

The central claim rests on several modeling premises that the reader does not pay for upstream: the world is a static 2D grid with discrete time steps, tasks are locked to entity types, the recharge rule is deterministic and always beneficial, and the environment follows a fully cooperative Dec-POMDP. These premises are introduced by the authors as design choices, not independently validated. There are no fitted physical constants; the learned neural network parameters are trained on rewards, and the main hand-set hyperparameters (discount factor, loss weights, learning schedule) are listed but their sensitivity is not analyzed. No new entities are invented.

free parameters (3)
  • Discount factor gamma = 0.7
    Set to 0.7 to prioritize immediate task execution; the sensitivity of results to this choice is not studied.
  • Loss weights lambda_opt and lambda_nopt = Not stated
    The total loss (Eq. 24) includes these weights but their values are never given, so the reported performance may depend on unreported tuning.
  • Learning rate schedule = 1e-4, decay 10% per 1000 episodes
    Chosen to meet Robbins-Monro conditions; the paper asserts convergence without proving the step sizes satisfy the conditions globally.
assumptions (6)
  • domain assumption The emergency rescue environment is accurately represented by a static grid with discrete time steps and known obstacle cells.
    Section III-A models the sensing area as grid cells with binary obstacle/task flags; if real disaster scenes have dynamic obstacles or continuous terrain, the policy may not transfer.
  • domain assumption Each task type can only be completed by the corresponding entity type (UAV, UGV, or human).
    Section III-B fixes a one-to-one mapping between task types and entity capabilities; the paper does not consider tasks that can be done by multiple entity types or require joint action.
  • domain assumption UGVs always detect and recharge low-battery UAVs in range, deterministically (Eq. 1 and Eq. 2).
    The 'Hard-Cooperative' rule is hard-coded into the environment; real sensor failures, mis-detection, or competing UGV tasks are not modeled.
  • standard math The Dec-POMDP belief-state update (Appendix C) is sufficient for optimal action selection.
    This is standard POMDP theory; the paper's proof is a recursive belief update, not a new result.
  • standard math The orienteering problem is NP-hard and the reduction in Appendix B is valid.
    The NP-hardness proof reduces Problem 1 to a single-UAV orienteering instance; the reduction is plausible but does not fully specify depot, time budgets, and node scores, and therefore relies on standard OP hardness.
  • standard math Q-learning with Robbins-Monro step sizes converges to the optimal Q-function.
    Invoked in Theorem 4; standard stochastic approximation, but the proof does not correctly apply the conditions.

how reviews work

0 comments
Cite this review

Pith. "Pith review of A Multi-Agent Reinforcement Learning Approach for Cooperative Air-Ground-Human Crowdsensing in Emergency Rescue." pith.science (2026). https://pith.science/paper/EA5FPC7I

@misc{pith2026250506997,
  author       = {Pith},
  title        = {Pith review of: A Multi-Agent Reinforcement Learning Approach for Cooperative Air-Ground-Human Crowdsensing in Emergency Rescue},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EA5FPC7I}},
  note         = {Machine review of arXiv:2505.06997}
}
read the original abstract

Mobile crowdsensing is evolving beyond traditional human-centric models by integrating heterogeneous entities like unmanned aerial vehicles (UAVs) and unmanned ground vehicles (UGVs). Optimizing task allocation among these diverse agents is critical, particularly in challenging emergency rescue scenarios characterized by complex environments, limited communication, and partial observability. This paper tackles the Heterogeneous-Entity Collaborative-Sensing Task Allocation (HECTA) problem specifically for emergency rescue, considering humans, UAVs, and UGVs. We introduce a novel ``Hard-Cooperative'' policy where UGVs prioritize recharging low-battery UAVs, alongside performing their sensing tasks. The primary objective is maximizing the task completion rate (TCR) under strict time constraints. We rigorously formulate this NP-hard problem as a decentralized partially observable Markov decision process (Dec-POMDP) to effectively handle sequential decision-making under uncertainty. To solve this, we propose HECTA4ER, a novel multi-agent reinforcement learning algorithm built upon a Centralized Training with Decentralized Execution architecture. HECTA4ER incorporates tailored designs, including specialized modules for complex feature extraction, utilization of action-observation history via hidden states, and a mixing network integrating global and local information, specifically addressing the challenges of partial observability. Furthermore, theoretical analysis confirms the algorithm's convergence properties. Extensive simulations demonstrate that HECTA4ER significantly outperforms baseline algorithms, achieving an average 18.42% increase in TCR. Crucially, a real-world case study validates the algorithm's effectiveness and robustness in dynamic sensing scenarios, highlighting its strong potential for practical application in emergency response.

Figures

Figures reproduced from arXiv: 2505.06997 by the authors.

Figure 1
Figure 1. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Grid-based modeling of the emergency rescue environment. data collection. Zhao et al. [10] introduced gMADRL-VCS, applying diffusion models and hierarchical MARL to optimize navigation and path planning strategies for sensing entities. Furthermore, to address complex task relationships, Zhao et al. [34] employed heterogeneous graph RL in HGRL-TA for handling various complex relationships in interdependent multi-task… view at source ↗
Figure 3
Figure 3. State space. (a)-(c) show the distribution of obstacles, tasks and entities respectively; (d) represents the type of task, indicating which category of entity can perform the task; (e) depicts the remaining execution time of the task, which will gradually decrease when an agent is performing the task; (f) shows the matching status between tasks and sensing entities. by the position of the bit with a value of 1 [PIT… view at source ↗
Figures from the paper (9 more)
Figure 4
Figure 4. Figure 4: Local observation space. (a) represents entity’s current position; (b) shows the movable range, the cells highlighted in green, of entity k within a time step; (c) depicts the power information of UAVs, including current power and power consumption per step; (d) shows …
Figure 5
Figure 5. Figure 5: Action space. (a) shows the traditional action space in route planning problem, which includes discrete movement directions. (b) represents the action space designed in our problem, denoted by the movable range of any entities at the next time step. a t k = (w/d/g)Loct…
Figure 6
Figure 6. Figure 6: Overview of the HECTA4ER algorithm. The workflow begins with sensing entities interacting with the environment to obtain global states and local observations, which are then processed by the EIEM to extract relevant features. These features are used by the SEDM to make…
Figure 7
Figure 7. Figure 7: Structure of experience replay. D. CTDE training process and loss function The algorithm employs the CTDE architecture [40]. During the centralized training phase, the algorithm leverages global information s t and data aggregated from all entities to train the critic …
Figure 8
Figure 8. Figure 8: The training curves of different algorithms. (a) Scenario 1 (b) Scenario 2 (c) Scenario 3 (d) Scenario 4 [PITH_FULL_IMAGE:figures/full_fig_p011_8.png]
Figure 9
Figure 9. Figure 9: Overall performance of different algorithms with scenario 1-4. performance of MANF-RL-RP and FD-MAPPO is similar to each other, but significantly lower than HECTA4ER’s. HECTA4ER-Voluntary closely matches HECTA4ER’s results in most scenarios, though it slightly under-pe…
Figure 10
Figure 10. Figure 10: Experimental results of different scenario parameters. (a) Satellite map of Zhongfu Commu￾nity (b) Visualization of the task scenario [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]
Figure 11
Figure 11. Figure 11: Real-world scene of the case study. rising from around 65% to nearly 87%. MANF-RL-RP also improves significantly (TCR around 55% to 77%) but consis￾tently lags behind HECTA4ER. The TCR of the FD-MAPPO algorithm increases from around 46% to 71%. HECTA4ER￾Voluntary’s pe…
Figure 12
Figure 12. Figure 12: Visualization results of the movement trajectories for human workers, UGVs, and UAVs under the time limit of 12 [PITH_FULL_IMAGE:figures/full_fig_p014_12.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

48 extracted references · 46 canonical work pages

  1. [9]

    Air-ground spatial crowdsourcing with uav carriers by geometric graph convolutional multi-agent deep reinforcement learning,

    Y . Wang, J. Wu, X. Hua, C. H. Liu, G. Li, J. Zhao, Y . Yuan, and G. Wang, “Air-ground spatial crowdsourcing with uav carriers by geometric graph convolutional multi-agent deep reinforcement learning,” in2023 IEEE 39th Int. Conf. Data Eng. (ICDE). IEEE, 2023, pp. 1790–1802

    work page 2023

  2. [10]

    Energy-efficient ground-air-space vehicular crowdsensing by hierarchical multi-agent deep reinforcement learning with diffusion models,

    Y . Zhao, C. H. Liu, T. Yi, G. Li, and D. Wu, “Energy-efficient ground-air-space vehicular crowdsensing by hierarchical multi-agent deep reinforcement learning with diffusion models,”IEEE J. Sel. Areas Commun., 2024

    work page 2024

  3. [1]

    A comprehensive survey on mobile crowdsensing systems,

    D. Suhag and V . Jha, “A comprehensive survey on mobile crowdsensing systems,”J. Syst. Archit., vol. 142, p. 102952, 2023

    work page 2023

  4. [2]

    Privacy-preserving mechanisms for location privacy in mobile crowdsensing: A survey,

    J. W. Kim, K. Edemacu, and B. Jang, “Privacy-preserving mechanisms for location privacy in mobile crowdsensing: A survey,”J. Netw. Comput. Appl., vol. 200, p. 103315, 2022

    work page 2022

  5. [3]

    Drift: A dynamic crowd inflow control system using lstm-based deep reinforcement learn- ing,

    X. Liao, W. Chen, X. Guo, J. Zhong, and D. Wang, “Drift: A dynamic crowd inflow control system using lstm-based deep reinforcement learn- ing,”IEEE Trans. Syst. Man Cybern. Syst., 2025

    work page 2025

  6. [4]

    A survey on task assignment in crowdsourcing,

    D. Hettiachchi, V . Kostakos, and J. Goncalves, “A survey on task assignment in crowdsourcing,”ACM Comput. Surv., vol. 55, no. 3, pp. 1–35, 2022

    work page 2022

  7. [5]

    Towards crowd- sourcing internet of things (crowd-iot): Architectures, security and applications,

    K. L. M. Ang, J. K. P. Seng, and E. Ngharamike, “Towards crowd- sourcing internet of things (crowd-iot): Architectures, security and applications,”Future Internet, vol. 14, no. 2, p. 49, 2022

    work page 2022

  8. [6]

    Task-oriented wireless communications for collaborative perception in intelligent un- manned systems,

    S. Zhou, Y . Jia, R. Mao, Z. Nan, Y . Sun, and Z. Niu, “Task-oriented wireless communications for collaborative perception in intelligent un- manned systems,”IEEE Network, 2024

    work page 2024

Show all 48 references
  1. [7]

    Unified perception and collaborative mapping for connected and autonomous vehicles,

    Z. Yang and C. Liu, “Unified perception and collaborative mapping for connected and autonomous vehicles,”IEEE Network, vol. 37, no. 4, pp. 273–281, 2023

    2023

  2. [8]

    A survey and framework of cooperative perception: From heterogeneous singleton to hierarchical cooperation,

    Z. Bai, G. Wu, M. J. Barth, Y . Liu, E. A. Sisbot, K. Oguchi, and Z. Huang, “A survey and framework of cooperative perception: From heterogeneous singleton to hierarchical cooperation,”IEEE Trans. Intell. Transp. Syst., 2024

    2024

  3. [11]

    The system framework and application research of the parallel emergency management system (pems),

    F. Wang, “The system framework and application research of the parallel emergency management system (pems),”Chin. Emerg. Manage., no. 12, pp. 22–27, 2007

    2007

  4. [12]

    Parallel emergency management of incidents by integrating ooda and prea loops: The c2 mechanism and modes,

    D. Yang, Q. Li, F. Zhu, H. Cui, W. Yi, and J. Qin, “Parallel emergency management of incidents by integrating ooda and prea loops: The c2 mechanism and modes,”IEEE Trans. Syst. Man Cybern. Syst., vol. 53, no. 4, pp. 2160–2172, 2022

    2022

  5. [13]

    Towards time- constrained task allocation in semi-opportunistic mobile crowdsensing,

    W. Wei, H. Chen, X. Liu, G. Ma, Y . Liu, and X. Liu, “Towards time- constrained task allocation in semi-opportunistic mobile crowdsensing,” Ad Hoc Networks, vol. 150, p. 103282, 2023

    2023

  6. [14]

    Optimal resource allocation for uav-relay-assisted mobile crowdsensing,

    X. Yang, Y . Fu, J. Zheng, Z. Xu, R. Shao, and Y . Wu, “Optimal resource allocation for uav-relay-assisted mobile crowdsensing,”IEEE Trans. Commun., 2024

    2024

  7. [15]

    Mappo-based cooperative uav trajectory design with long-range emergency communications in disaster areas,

    Y . Guan, S. Zou, K. Li, W. Ni, and B. Wu, “Mappo-based cooperative uav trajectory design with long-range emergency communications in disaster areas,” in2023 IEEE 24th Int. Symp. World Wireless Mob. Multimedia Netw. (WoWMoM). IEEE, 2023, pp. 376–381

    2023

  8. [16]

    Disaster-resilient emergency communication with intelligent air–ground cooperation,

    X. Tang, F. Chen, F. Wang, and Z. Jia, “Disaster-resilient emergency communication with intelligent air–ground cooperation,”IEEE Internet Things J., vol. 11, no. 3, pp. 5331–5346, 2023

    2023

  9. [17]

    Task search and alloca- tion strategy for heterogeneous multiagent systems under communication constraints,

    Z. Mao, D. Liu, K. Ju, B. Jiang, and X.-G. Yan, “Task search and alloca- tion strategy for heterogeneous multiagent systems under communication constraints,”IEEE Trans. Syst. Man Cybern. Syst., 2024

    2024

  10. [18]

    Cooperative multi- aav path planning for discovering and tracking multiple radio-tagged targets,

    Y . Cui, J. Chen, H. Lin, Z. Shu, and T. Huang, “Cooperative multi- aav path planning for discovering and tracking multiple radio-tagged targets,”IEEE Trans. Syst. Man Cybern. Syst., 2025

    2025

  11. [19]

    Robust training in multiagent deep reinforcement learning against optimal adversary,

    W. Guo, G. Liu, Z. Zhou, J. Wang, Y . Tang, and M. Wang, “Robust training in multiagent deep reinforcement learning against optimal adversary,”IEEE Trans. Syst. Man Cybern. Syst., pp. 1–12, 2025

    2025

  12. [20]

    Sensors on the internet of things systems for urban disaster management: a systematic literature review,

    F. Zeng, C. Pang, and H. Tang, “Sensors on the internet of things systems for urban disaster management: a systematic literature review,”Sensors, vol. 23, no. 17, p. 7475, 2023

    2023

  13. [21]

    Optimization of emergency rescue routes after a violent earthquake,

    X. Wang, S. Wu, Z. Zhao, H. Guo, and W. Chen, “Optimization of emergency rescue routes after a violent earthquake,”Nat. Hazards, pp. 1–29, 2024

    2024

  14. [22]

    Autonomous unmanned aerial vehicles in bushfire management: Challenges and opportunities,

    S. Partheepan, F. Sanati, and J. Hassan, “Autonomous unmanned aerial vehicles in bushfire management: Challenges and opportunities,”Drones, vol. 7, no. 1, p. 47, 2023

    2023

  15. [23]

    Gacf: Ground-aerial collaborative framework for large-scale emergency rescue scenarios,

    Y . Zhang, J. Yu, Y . Tang, Y . Deng, X. Tian, Y . Yue, and Y . Yang, “Gacf: Ground-aerial collaborative framework for large-scale emergency rescue scenarios,” in2023 IEEE Int. Conf. Unmanned Syst. (ICUS). IEEE, 2023, pp. 1701–1707

    2023

  16. [24]

    Cooperative unmanned surface vehicles and unmanned aerial vehicles platform as a tool for coastal monitoring activities,

    J. Wu, R. Li, J. Li, M. Zou, and Z. Huang, “Cooperative unmanned surface vehicles and unmanned aerial vehicles platform as a tool for coastal monitoring activities,”Ocean Coast. Manage., vol. 232, p. 106421, 2023

    2023

  17. [25]

    Human detection and action recognition for search and rescue in disasters using yolov3 algorithm,

    B. Valarmathi, J. Kshitij, R. Dimple, N. Srinivasa Gupta, Y . Harold Robinson, G. Arulkumaran, and T. Mulu, “Human detection and action recognition for search and rescue in disasters using yolov3 algorithm,”J. Electr. Comput. Eng., vol. 2023, no. 1, p. 5419384, 2023

    2023

  18. [26]

    A comprehensive survey of unmanned ground vehicle terrain traversability for unstructured environments and sensor technology insights,

    S. Beycimen, D. Ignatyev, and A. Zolotas, “A comprehensive survey of unmanned ground vehicle terrain traversability for unstructured environments and sensor technology insights,”Eng. Sci. Technol. Int. J., vol. 47, p. 101457, 2023

    2023

  19. [27]

    Unmanned aerial vehicles for search and rescue: A survey,

    M. Lyu, Y . Zhao, C. Huang, and H. Huang, “Unmanned aerial vehicles for search and rescue: A survey,”Remote Sensing, vol. 15, no. 13, p. 3266, 2023

    2023

  20. [28]

    A crowd- aided vehicular hybrid sensing framework for intelligent transportation systems,

    Z. Zhu, Y . Zhao, B. Chen, S. Qiu, Z. Liu, K. Xie, and L. Ma, “A crowd- aided vehicular hybrid sensing framework for intelligent transportation systems,”IEEE Trans. Intell. Veh., vol. 8, no. 2, pp. 1484–1497, 2022

    2022

  21. [29]

    Cooperative uav trajectory design for disaster area emergency communications: A multiagent ppo method,

    Y . Guan, S. Zou, H. Peng, W. Ni, Y . Sun, and H. Gao, “Cooperative uav trajectory design for disaster area emergency communications: A multiagent ppo method,”IEEE Internet Things J., vol. 11, no. 5, pp. 8848–8859, 2023

    2023

  22. [30]

    Collab- orative route planning of uavs, workers, and cars for crowdsensing in disaster response,

    L. Han, C. Tu, Z. Yu, Z. Yu, W. Shan, L. Wang, and B. Guo, “Collab- orative route planning of uavs, workers, and cars for crowdsensing in disaster response,”IEEE/ACM Trans. Netw., 2024

    2024

  23. [31]

    Decentralized task assignment for mobile crowd- sensing with multi-agent deep reinforcement learning,

    C. Xu and W. Song, “Decentralized task assignment for mobile crowd- sensing with multi-agent deep reinforcement learning,”IEEE Internet Things J., vol. 10, no. 18, pp. 16 564–16 578, 2023

    2023

  24. [32]

    Ensuring threshold aoi for uav-assisted mobile crowdsensing by multi-agent deep reinforcement learning with transformer,

    H. Wang, C. H. Liu, H. Yang, G. Wang, and K. K. Leung, “Ensuring threshold aoi for uav-assisted mobile crowdsensing by multi-agent deep reinforcement learning with transformer,”IEEE/ACM Trans. Netw., vol. 32, no. 1, pp. 566–581, 2023

    2023

  25. [33]

    Exploring both individuality and cooperation for air-ground spatial 16 crowdsourcing by multi-agent deep reinforcement learning,

    Y . Ye, C. H. Liu, Z. Dai, J. Zhao, Y . Yuan, G. Wang, and J. Tang, “Exploring both individuality and cooperation for air-ground spatial 16 crowdsourcing by multi-agent deep reinforcement learning,” in2023 IEEE 39th Int. Conf. Data Eng. (ICDE). IEEE, 2023, pp. 205–217

    2023

  26. [34]

    Hetero- geneous graph reinforcement learning for dependency-aware multi-task allocation in spatial crowdsourcing,

    Y . Zhao, Z. Zhu, C. Gao, E. Wang, J. Huang, and F.-Y . Wang, “Hetero- geneous graph reinforcement learning for dependency-aware multi-task allocation in spatial crowdsourcing,”arXiv preprint arXiv:2410.15449, 2024

    2024 arXiv

  27. [35]

    Energy-efficient 3d vehicular crowdsourcing for disaster response by distributed deep reinforcement learning,

    H. Wang, C. H. Liu, Z. Dai, J. Tang, and G. Wang, “Energy-efficient 3d vehicular crowdsourcing for disaster response by distributed deep reinforcement learning,” inProc. 27th ACM SIGKDD Conf. Knowl. Discov. Data Min., 2021, pp. 3679–3687

    2021

  28. [36]

    Pomdp inference and robust solution via deep reinforce- ment learning: An application to railway optimal maintenance,

    G. Arcieri, C. Hoelzl, O. Schwery, D. Straub, K. G. Papakonstantinou, and E. Chatzi, “Pomdp inference and robust solution via deep reinforce- ment learning: An application to railway optimal maintenance,”Mach. Learn., vol. 113, no. 10, pp. 7967–7995, 2024

    2024

  29. [37]

    Informed pomdp: Lever- aging additional information in model-based rl,

    G. Lambrechts, A. Bolland, and D. Ernst, “Informed pomdp: Lever- aging additional information in model-based rl,”arXiv preprint arXiv:2306.11488, 2023

    2023 arXiv

  30. [38]

    An effective cnn and transformer com- plementary network for medical image segmentation,

    F. Yuan, Z. Zhang, and Z. Fang, “An effective cnn and transformer com- plementary network for medical image segmentation,”Pattern Recogni- tion, vol. 136, p. 109228, 2023

    2023

  31. [39]

    A comprehensive overview and comparative analysis on deep learning models,

    T. Perumal, N. Mustapha, R. Mohamed, and F. M. Shiri, “A comprehensive overview and comparative analysis on deep learning models,”J. Artif. Intell., vol. 6, no. 1, pp. 301–360, 2024. [Online]. Available: http://www.techscience.com/jai/v6n1/58699

    2024

  32. [40]

    An introduction to centralized training for decentralized execution in cooperative multi-agent reinforcement learning,

    C. Amato, “An introduction to centralized training for decentralized execution in cooperative multi-agent reinforcement learning,”arXiv preprint arXiv:2409.03052, 2024

    2024 arXiv

  33. [41]

    Intelligent routing method based on dueling dqn reinforcement learning and net- work traffic state prediction in sdn,

    L. Huang, M. Ye, X. Xue, Y . Wang, H. Qiu, and X. Deng, “Intelligent routing method based on dueling dqn reinforcement learning and net- work traffic state prediction in sdn,”Wireless Networks, vol. 30, no. 5, pp. 4507–4525, 2024

    2024

  34. [42]

    Monotonic value function factorisation for deep multi- agent reinforcement learning,

    T. Rashid, M. Samvelyan, C. S. De Witt, G. Farquhar, J. Foerster, and S. Whiteson, “Monotonic value function factorisation for deep multi- agent reinforcement learning,”J. Mach. Learn. Res., vol. 21, no. 178, pp. 1–51, 2020

    2020

  35. [43]

    Qtran: Learning to factorize with transformation for cooperative multi-agent reinforcement learning,

    K. Son, D. Kim, W. J. Kang, D. E. Hostallero, and Y . Yi, “Qtran: Learning to factorize with transformation for cooperative multi-agent reinforcement learning,” inProc. Int. Conf. Mach. Learn.PMLR, 2019, pp. 5887–5896

    2019

  36. [44]

    The surprising effectiveness of ppo in cooperative multi-agent games,

    C. Yu, A. Velu, E. Vinitsky, J. Gao, Y . Wang, A. Bayen, and Y . Wu, “The surprising effectiveness of ppo in cooperative multi-agent games,” Adv. Neural Inf. Process. Syst., vol. 35, pp. 24 611–24 624, 2022

    2022

  37. [45]

    Online organizing large-scale heterogeneous tasks and multi-skilled participants in mobile crowdsensing,

    L. Han, Z. Yu, Z. Yu, L. Wang, H. Yin, and B. Guo, “Online organizing large-scale heterogeneous tasks and multi-skilled participants in mobile crowdsensing,”IEEE Trans. Mobile Comput., vol. 22, no. 5, pp. 2892– 2909, 2023

    2023

  38. [46]

    Human-drone collaborative spatial crowdsourcing by memory- augmented and distributed multi-agent deep reinforcement learning,

    Y . Wang, C. H. Liu, C. Piao, Y . Yuan, R. Han, G. Wang, and J. Tang, “Human-drone collaborative spatial crowdsourcing by memory- augmented and distributed multi-agent deep reinforcement learning,” in 2022 IEEE 38th Int. Conf. Data Eng. (ICDE). IEEE, 2022, pp. 459– 471

    2022

  39. [47]

    Modeling human steering behavior in teleoperation of unmanned ground vehicles with varying speed,

    C. Li, Y . Tang, Y . Zheng, P. Jayakumar, and T. Ersal, “Modeling human steering behavior in teleoperation of unmanned ground vehicles with varying speed,”Hum. Factors, vol. 64, no. 3, pp. 589–600, 2022

    2022

  40. [48]

    3d building model generation from mls point cloud and 3d mesh using multi-source data fusion,

    W. Liu, Y . Zang, Z. Xiong, X. Bian, C. Wen, X. Lu, C. Wang, J. M. Junior, W. N. Gonc ¸alves, and J. Li, “3d building model generation from mls point cloud and 3d mesh using multi-source data fusion,”Int. J. Appl. Earth Obs. Geoinformation, vol. 116, p. 103171, 2023. ABOUT THE...

    2023

Pith tools

Reviewed August 15, 2026 · model on record in the stance chip above.