Pith. sign in

REVIEW 4 major objections 5 minor 45 references

Dynamic Airspace Management for UAVs in Evolving Urban Environments: Collaborative Coordination and Human Safety

T0 review · 4 major / 5 minor · reviewed 2026-07-11 · grok-4.5

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

Pith's one-line read Pharos coordinates non-overlapping exclusive spaces so multiple UAVs can fly in parallel through cities while lowering a human-fear metric and raising airspace utilization.

desk verdict Solid engineering idea for exclusive-space multi-UAV coordination with a real urban sim and code, but the abstract’s headline fear claim is inverted relative to Table 2 and needs fixing before anyone cites the numbers. read the letter →

arxiv 2607.04825 v1 pith:63PXH2WK submitted 2026-07-06 cs.MA

Lin Sun , Yuhang Wang , Fan Meng Hong , Haopeng Chen , Yan Jiao , Yongming Xu This is my paper
classification cs.MA
keywords AirspacemanagementUAVcollaborativecoordinationExclusivespaceHumansafetySpatialentropyMulti-agentreinforcementlearningMAPPO
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

Dense low-altitude UAV traffic over cities risks both mid-air collisions and frightening people on the ground. The paper argues that neither purely local onboard sensing nor fully centralized trajectory control solves the problem cleanly. Pharos sits in the middle: a central coordinator learns to hand each UAV a private cuboid of airspace at every time step; the vehicles then fly autonomously inside those boxes. Training uses multi-agent proximal policy optimization whose reward penalizes collisions and a geometric human-fear term while still rewarding progress toward destinations. In a 3-D simulator built from real Shanghai building data, the learned policy keeps vehicles safe, cuts the paper's fear metric by more than half relative to a classical path planner, and improves a new spatial-entropy measure of how evenly the sky is used. The approach therefore claims to give cities a practical way to share the low-altitude sky with both drones and pedestrians.

What carries the argument

Exclusive cuboid spaces: each MAPPO actor outputs a discrete direction that defines a safe box around its UAV; a shared critic and a composite reward (collision volume, human-fear factor based on distance-angle-velocity, and progress bonuses) train the joint policy that produces those boxes.

What would settle it

Fly the same set of origin-destination pairs under Pharos-generated exclusive spaces versus the Ipopt baseline in a controlled outdoor trial with instrumented pedestrians; if self-reported fear or physiological stress does not drop as the simulation predicts, the central human-safety claim is falsified.

Watch

Extended reading notes

Core claim

A multi-agent reinforcement learner can assign non-overlapping exclusive cuboids to many UAVs at once so that the vehicles operate in parallel without collisions or excessive human fear, outperforming both a mathematical optimizer and classical A-star on the paper's safety and utilization metrics inside a real-city 3-D simulation.

Load-bearing premise

The hand-crafted formula that turns distance, relative heading and speed into a human-fear number is assumed to be a good enough proxy for real pedestrian psychology that optimizing it keeps people safe and comfortable.

Editorial extensions

If this is right

  • UAV fleets can share airspace without heavy onboard sensors or manufacturer-opened control APIs, needing only lightweight telemetry exchange.
  • Human psychological impact becomes a first-class, optimizable constraint rather than an after-the-fact check.
  • Spatial entropy can serve as a standard post-hoc score for how uniformly any airspace-management scheme uses the available volume.
  • Pre-trained policies can be reused across cities whose building footprints are discretized the same way, enabling faster deployment.
  • The same exclusive-space interface can later support cross-zone hand-offs and on-demand admission of additional vehicles.

Reading between the lines

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

  • The exclusive-cuboid idea could transfer to mixed fleets of ground robots and low-flying drones that must share the same street-level volume without a single manufacturer controlling every vehicle.
  • Turning spatial entropy into an online admission gate (as the authors flag for future work) would let a city refuse new UAV requests when local entropy falls, automatically preventing congestion hotspots.
  • Field calibration of the fear formula against real anxiety scores would either validate the reward design or force a data-driven replacement, clarifying how much of the reported gain is simulation artifact.
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

4 major / 5 minor

Summary. The paper proposes Pharos, a collaborative multi-UAV airspace management system that sits between distributed local-perception and centralized fine-grained trajectory control. It coordinates non-overlapping exclusive cuboid spaces for UAVs via a Dec-POMDP formalization solved by MAPPO, with a multi-objective reward that penalizes inter-UAV/obstacle collisions and a novel human-fear term (Eq. 2) while rewarding flight progression. Evaluation uses a 3D simulator built on real Shanghai OSM building data, stochastic pedestrians, and visualizations of intersection and obstacle scenarios. Reported results include faster MAPPO convergence and higher rewards than HAPPO/HATRPO, a 52.72% human-fear reduction versus Ipopt, and spatial-entropy gains of 70.82% and 2.03% versus Ipopt and A-star, respectively, with anonymized source code released.

Significance. If the exclusive-space coordination paradigm and fear-aware reward hold under tighter evaluation, Pharos would supply a practical middle ground for scalable low-altitude multi-UAV operations that avoids heavy onboard sensors and proprietary motion-control interfaces. Credit is due for the open simulation code, real-map 3D environment, explicit Dec-POMDP formulation, spatial-entropy utilization metric, and systematic MAPPO-versus-HAPPO/HATRPO comparison. The human-fear modeling and parallel exclusive-space inference are useful additions to human–UAV coexistence work. The quantitative safety and utilization claims, however, currently rest on an unvalidated proxy formula, single-run averages, and an inverted abstract baseline, so significance is conditional on correction and stronger statistical support.

major comments (4)
  1. [Abstract / §5.3.3 / Table 2] The abstract asserts “Pharos reduced human fear by 52.72% compared to the benchmark Ipopt.” Table 2 and §5.3.3 show the opposite ordering: Ipopt average fear (0.1720 / 0.3953 / 0.3131) is substantially lower than Pharos (0.4228 / 0.9273 / 1.4304) for 10/20/30 UAVs. The 52.72% figure matches only the average reduction versus A-star. §5.3.3 itself notes that Ipopt’s lower fear arises from hovering (a suboptimal policy that sacrifices progression). This factual inversion of the paper’s primary quantitative human-safety claim must be corrected in the abstract, highlights, and any corresponding text; the progression–fear trade-off should be stated explicitly rather than presented as a pure fear win over Ipopt.
  2. [§3.2.2 Eq. (2) / Table 1] The human-fear factor (Eq. 2: (1−cos θ)·max(0,cos φ)/d inside radius r̄=5 m) is load-bearing for the claimed human-safety contribution and is folded into the shared reward with weight βf=2.5 (Table 1). No empirical calibration, pedestrian study, or sensitivity analysis against alternative angular/distance formulations is supplied. Because r̄, the angular terms, and the reward weights are free parameters, the safety benefit is only as strong as this ad-hoc proxy. The manuscript should acknowledge the model’s provisional status and report how fear rankings and spatial entropy change under reasonable variations of r̄ and the trigonometric factors.
  3. [§5.3 / Table 2 / Figs. 6–8] All tabulated fear and spatial-entropy numbers (Table 2), reward curves (Figs. 6–7), and scalability plots (Fig. 8) appear to be single-run averages without variance, confidence intervals, or multi-seed statistics. For a stochastic MARL system whose environment randomizes human trajectories and UAV destinations, this precludes assessing whether the 2.03% entropy edge over A-star or the MAPPO ranking over HAPPO/HATRPO is statistically reliable. Multi-seed means and error bars (or equivalent) are required for the central performance claims.
  4. [§5.2 / §5.3.1 / Fig. 7 / §5.3.3] Ipopt is positioned as an accuracy benchmark for the global objective R (§5.2, §5.3.1, Fig. 7), yet §5.3.3 observes that it frequently selects hovering, producing low short-horizon fear and low spatial entropy while failing to progress. Comparing a myopic continuous solver against a learned long-horizon policy on fear alone is therefore misleading. Either constrain Ipopt to a multi-step or progression-aware formulation, or reframe the comparison as evidence for the value of long-horizon planning rather than pure optimality of R.
minor comments (5)
  1. [Title / throughout] Title and running text repeatedly insert a space in “UA Vs”; standardize to “UAVs”.
  2. [Abstract / §5.3.4 Eq. (9)] Clarify that higher spatial entropy is defined as better utilization; the abstract’s phrasing “improved performance by 70.82%” is otherwise ambiguous to readers unfamiliar with the metric.
  3. [Fig. 3 / Eq. (2)] Fig. 3 is projected onto the xOz plane while the model uses 3D vectors; a short note that the four cases are illustrative projections would avoid confusion.
  4. [§1 / §4] The assumption that a lightweight data-exchange interface suffices for exclusive-space enforcement on heterogeneous commercial UAVs is stated but not stress-tested (latency, packet loss, non-compliance). A brief limitations paragraph would help.
  5. [Table 1 / Algorithm 1] Hyperparameter table (Table 1) and algorithm box (Algorithm 1) are clear; a one-sentence note on whether actor networks share parameters or are fully heterogeneous would aid reproducibility.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: explicit hand-crafted reward and post-hoc metrics evaluated against independent baselines; no derivation reduces to its own inputs by construction.

full rationale

The paper is an empirical multi-agent RL system. The Dec-POMDP (Sec. 4.1), observation features (including the APF-style repulsive term and the seven-value fear prediction vector F_i built from Eq. 2), discrete exclusive-space actions (Eq. 8), and shared reward (Eq. 4 = weighted sum of the collision indicator P^s, the authors’ own fear formula P^f, and the two progression bonuses) are all stated explicitly as design choices. MAPPO is trained to maximize that reward; evaluation then re-measures the same quantities (average human-fear values and the newly defined spatial entropy H of Eq. 9) on held-out roll-outs and compares them to Ipopt and A-star. Nothing is fitted to a data subset and then “predicted,” no uniqueness theorem is imported from prior self-work, and no intermediate quantity is defined in terms of the final reported metric. The only self-reference is the anonymized GitHub repository of the authors’ own simulator, which is ordinary engineering practice and does not close a logical loop. The factual inversion of the 52.72 % claim in the abstract (it matches A-star, not Ipopt) is a correctness error, not a circularity. Consequently the derivation chain is self-contained and non-circular.

Assumptions & free parameters 4 free parameters · 4 assumptions · 3 invented entities

The central performance claims rest on a discrete cuboid model, a hand-crafted fear formula, fixed reward weights, and the assumption that MAPPO policies trained in a simplified Shanghai map transfer to real heterogeneous fleets. These are free parameters and domain assumptions rather than derived quantities.

free parameters (4)
  • reward weights βs, βf, αl, αg = 100.0, 2.5, 3.0, 1.0
    Set by hand to 100, 2.5, 3.0, 1.0 (Table 1) to enforce safety priority; no sensitivity analysis or learning of the weights is reported.
  • human-fear radius r̄ and height h̄ = 5.0 m, 1.7 m
    Fixed at 5.0 m and 1.7 m (Table 1); directly scales the fear penalty that drives the claimed 52% reduction.
  • spatial discretization e, μ, δ = 1, 0.5, 0.02
    Meter-scale grid (e=1, μ=0.5, δ=0.02) chosen from latency/speed arguments; changes the exclusive-space geometry and collision checks.
  • MAPPO hyperparameters (γ, clip ς, learning rates, episode length) = γ=0.95, ς=0.2, lr=0.005/0.002, 5e6 steps
    Standard PPO values (Table 1) that control convergence; different settings alter the final policy used for all reported metrics.
assumptions (4)
  • domain assumption Discrete exclusive cuboids of integer meter extent are a sufficient and safe abstraction of continuous 3-D airspace for collision-free flight.
    Stated in §3.1 as the modeling choice that reduces computation; never proved to preserve safety under real dynamics or wind.
  • ad hoc to paper The scalar fear formula (1-cos θ)·max(0,cos φ)/d inside r̄ is an adequate quantitative model of human psychological impact.
    Introduced in §3.2.2 without human-subject validation; used both as reward and as evaluation metric.
  • domain assumption A lightweight data-exchange interface is sufficient to enforce exclusive spaces without access to proprietary UAV flight controllers.
    Core architectural claim of the “middle paradigm” (§1, Fig. 1c); assumed rather than demonstrated on real hardware.
  • standard math MAPPO with centralized critic and decentralized actors converges to a near-optimal joint policy for the finite-horizon Dec-POMDP.
    Invoked via citation to Yu et al. 2022; empirical convergence plots (Fig. 6) support it inside the training distribution.
invented entities (3)
  • human fear factor / penalty P^f_u
    purpose: Quantify psychological impact of low-altitude flight so that MARL can trade it off against collision and progress.
    Defined by Eq. 2; no external psychometric validation or independent measurement protocol is supplied.
  • spatial entropy H
    purpose: Post-hoc scalar that measures uniformity of UAV occupancy across discrete cells.
    Eq. 9 is a direct Shannon entropy of occupancy frequencies; useful but not independently validated as a capacity metric.
  • exclusive space cuboid c_u
    purpose: Coarse-grained safe volume that replaces fine trajectory planning.
    Core of the Pharos paradigm (§3.1, Eq. 8); safety relies on the unproven assumption that any path inside the cuboid remains collision-free under real dynamics.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Dynamic Airspace Management for UAVs in Evolving Urban Environments: Collaborative Coordination and Human Safety." pith.science (2026). https://pith.science/paper/63PXH2WK

@misc{pith2026260704825,
  author       = {Pith},
  title        = {Pith review of: Dynamic Airspace Management for UAVs in Evolving Urban Environments: Collaborative Coordination and Human Safety},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/63PXH2WK}},
  note         = {Machine review of arXiv:2607.04825}
}
read the original abstract

The low-altitude economy is an emerging industry with significant development potential, in which the safety of unmanned aerial vehicle (UAV) operations is a critical challenge. Particularly within complex urban topographies and human-populated environments, UAV airspace management must prioritize collision avoidance and human safety. We propose Pharos, a collaborative multi-UAV airspace management system. Pharos lies between the distributed local perception paradigm and the centralized fine-grained control paradigm. Pharos coordinates the safe parallel execution of UAVs in shared airspace while innovatively accounting for the impact of human fear. Pharos is implemented using the MAPPO algorithm due to its faster convergence and higher rewards than other typical MARL algorithms (HAPPO and HATRPO). To evaluate Pharos, we developed a 3D simulation system using real urban data. Visualization results demonstrate its effective airspace coordination capability. Regarding performance verification, Pharos reduced human fear by 52.72% compared to the benchmark Ipopt. Moreover, we designed spatial entropy as a system evaluation metric to quantify space utilization, which improved performance by 70.82% and 2.03% compared to the benchmarks Ipopt and A-star, respectively. The source code is available at an anonymized repository: https://github.com/pharos-anonymized/source-code.git.

Figures

Figures reproduced from arXiv: 2607.04825 by the authors.

Figure 1
Figure 1. Three airspace management paradigms: (a) Dis￾tributed local perception paradigm may lead to collisions of 𝐴,𝐵 at blind spots. (b) Centralized fine-grained control para￾digm ensures safety for 𝐴 and 𝐵 via sequential instructions by timeline ①-②. (c) Pharos coordinates the exclusive spaces for 𝐴 and 𝐵 through coarse-grained global management to enable safe parallel passage. and urban last-mile delivery. However, when … view at source ↗
Figure 2
Figure 2. An illustration of the multi-UAV management in [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. The human fear model with highlighting four spe [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Pharos’ core algorithm based on MAPPO. dynamically changing environment to achieve global objectives. Therefore, we adopt the Decentralized Partially Observable Markov Decision Process (Dec-POMDP) as the formal algorithm frame￾work of Pharos. This model accurately capt…
Figure 5
Figure 5. Figure 5: (a) Overview of Pharos visualization system. (b)(c) Snapshots of two test cases for system capabilities. Full dynamic [PITH_FULL_IMAGE:figures/full_fig_p007_5.png]
Figure 6
Figure 6. Figure 6: Comparison of training convergence with other typical MARL algorithms under different numbers of UAVs ( [PITH_FULL_IMAGE:figures/full_fig_p008_6.png]
Figure 7
Figure 7. Figure 7: Comparison of inference performance between [PITH_FULL_IMAGE:figures/full_fig_p008_7.png]
Figure 8
Figure 8. Figure 8: Scalability testing of Pharos based on MAPPO: Train [PITH_FULL_IMAGE:figures/full_fig_p010_8.png]

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

45 extracted references

  1. [1]

    Mohammad Amin Akbari, Mohsen Zare, Rasoul Azizipanah-Abarghooee, Seyedali Mirjalili, and Mohamed Deriche. 2022. The cheetah optimizer: A nature- inspired metaheuristic algorithm for large-scale optimization problems.Scientific reports12, 1 (2022), 10953

    2022

  2. [2]

    Noorulden Basil, Abdullah Fadhil Mohammed, Bayan Mahdi Sabbar, Hamzah M Marhoon, Adis Abebaw Dessalegn, Mohammad Alsharef, Enas Ali, and Sherif SM Ghoneim. 2025. Performance analysis of hybrid optimization approach for UAV path planning control using FOPID-TID controller and HAOAROA algorithm. Scientific Reports15, 1 (2025), 4840

    2025

  3. [3]

    Daniel S Bernstein, Robert Givan, Neil Immerman, and Shlomo Zilberstein

  4. [4]

    Mathematics of operations research27, 4 (2002), 819–840

    The complexity of decentralized control of Markov decision processes. Mathematics of operations research27, 4 (2002), 819–840

    2002

  5. [5]

    2018.Safety Flight Test Report for Networked Low-Altitude UA Vs

    CAAC. 2018.Safety Flight Test Report for Networked Low-Altitude UA Vs. https: //www.caac.gov.cn/XXGK/XXGK/GFXWJ/201811/t20181127_193186.html Ac- cessed: January 2026

    2018

  6. [6]

    Minghao Chen, Feng Shu, Min Zhu, Di Wu, Yu Yao, and Qi Zhang. 2025. Reinforcement-Learning-Based UAV 3-D Target Tracking and Digital-Twin- Assisted Collision Avoidance with Integrated Sensing and Communication.IEEE Internet of Things Journal(2025)

    2025

  7. [7]

    2023.Interim Regulations on the Flight Management of Un- manned Aerial Vehicles

    PRC State Council. 2023.Interim Regulations on the Flight Management of Un- manned Aerial Vehicles. https://www.gov.cn/gongbao/2023/issue_10586/202307/ content_6893000.html Accessed: January 2026

    2023

  8. [8]

    Ghulam Farid, Silvio Cocuzza, Talha Younas, Asghar Abbas Razzaqi, Waqas Ah- mad Wattoo, Ferdinando Cannella, and Hongwei Mo. 2022. Modified A-star (A*) approach to plan the motion of a quadrotor UAV in three-dimensional obstacle-cluttered environment.Applied Sciences12, 12 (2022), 5791

    2022

Show all 45 references
  1. [9]

    Shengwei Fu, Ke Li, Haisong Huang, Chi Ma, Qingsong Fan, and Yunwei Zhu

  2. [10]

    Red-billed blue magpie optimizer: a novel metaheuristic algorithm for 2D/3D UAV path planning and engineering design problems.Artificial Intelligence Review57, 6 (2024), 134

    2024

  3. [11]

    Jiali He, Liangdong Qu, Pei Wang, and Zhaowen Li. 2024. An oscillatory particle swarm optimization feature selection algorithm for hybrid data based on mutual information entropy.Applied Soft Computing152 (2024), 111261

    2024

  4. [12]

    Xiao-Bing Hu, Chang-Shu Yang, Jun Zhou, Ying-Fei Zhang, and Yi-Ming Ma

  5. [13]

    Research on 3D layered visibility graph route network model and multi- objective path planning for UAVs in complex urban environments.Aerospace Science and Technology159 (2025), 109947

    2025

  6. [14]

    Zihan Hu, Weikai Yang, Meng Zhang, Chengdong Lei, Hongtao Liang, Fuhui Zhou, and Qihui Wu. 2024. Toward Embodied Intelligence: An Autonomous Exploration System for Multimodal UAV. In2024 International Conference on Ubiquitous Communication (Ucom). IEEE, 444–448

    2024

  7. [15]

    Shuangyao Huang, Haibo Zhang, and Zhiyi Huang. 2024. CoDe: A Cooperative and Decentralized Collision Avoidance Algorithm for Small-Scale UAV Swarms Considering Energy Efficiency. In2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 13152–13159

    2024

  8. [16]

    S ISO. 2012. Robots and Robotic Devices-Safety Requirements for Industrial Robots-Part 1: Robots (ISO 10218-1: 2012).International Organization for Stan- dardization(2012)

    2012

  9. [17]

    JG Kuba, R Chen, M Wen, Y Wen, F Sun, J Wang, and Y Yang. 2022. Trust Region Policy Optimisation in Multi-Agent Reinforcement Learning. InICLR 2022-10th International Conference on Learning Representations. The International Conference on Learning Representations (ICLR), 1046

    2022

  10. [18]

    Jiacheng Li, Yangwang Fang, Haoyun Cheng, Zhikai Wang, Zihao Wu, and Mengjie Zeng. 2022. Large-scale fixed-wing UAV swarm system control with collision avoidance and formation maneuver.IEEE Systems Journal17, 1 (2022), 744–755

    2022

  11. [19]

    Shaowei Li, Yuhong Jia, Fan Yang, Qingyang Qin, Hui Gao, and Yaoming Zhou

  12. [20]

    Collaborative decision-making method for multi-UAV based on multiagent reinforcement learning.IEEE Access10 (2022), 91385–91396

    2022

  13. [21]

    Peng Liang, Yangtao Chen, Yafeng Sun, Ying Huang, and Wei Li. 2024. An information entropy-driven evolutionary algorithm based on reinforcement learning for many-objective optimization.Expert Systems with Applications238 (2024), 122164

    2024

  14. [22]

    Haicheng Liao, Zhenning Li, Huanming Shen, Wenxuan Zeng, Dongping Liao, Guofa Li, and Chengzhong Xu. 2024. Bat: Behavior-aware human-like trajectory prediction for autonomous driving. InProceedings of the AAAI Conference on Artificial Intelligence, Vol. 38. 10332–10340

    2024

  15. [23]

    Zhiyuan Liu, Yunshan Wang, Qixiu Cheng, and Hai Yang. 2022. Analysis of the in- formation entropy on traffic flows.IEEE Transactions on Intelligent Transportation Systems23, 10 (2022), 18012–18023

    2022

  16. [24]

    Qicheng Meng, Kai Chen, and Qingjun Qu. 2024. Ppswarm: Multi-uav path planning based on hybrid pso in complex scenarios.Drones8, 5 (2024), 192

    2024

  17. [25]

    Kelly Merckaert, Bryan Convens, Marco M Nicotra, and Bram Vanderborght. 2024. Real-time constraint-based planning and control of robotic manipulators for safe human–robot collaboration.Robotics and Computer-Integrated Manufacturing87 (2024), 102711

    2024

  18. [26]

    Tomáš Musil, Matěj Petrlík, and Martin Saska. 2022. SphereMap: Dynamic multi- layer graph structure for rapid safety-aware UAV planning.IEEE Robotics and Automation Letters7, 4 (2022), 11007–11014

    2022

  19. [27]

    2016.A concise introduction to decentralized POMDPs

    Frans A Oliehoek, Christopher Amato, et al . 2016.A concise introduction to decentralized POMDPs. Vol. 1. Springer

    2016

  20. [28]

    OpenStreetMap contributors. 2012. OpenStreetMap. http://www.openstreetmap. org (Accessed 1 March 2026)

    2012

  21. [29]

    Zhenhua Pan, Chengxi Zhang, Yuanqing Xia, Hao Xiong, and Xiaodong Shao

  22. [30]

    An improved artificial potential field method for path planning and forma- tion control of the multi-UAV systems.IEEE Transactions on Circuits and Systems II: Express Briefs69, 3 (2021), 1129–1133

    2021

  23. [31]

    Aaron Pereira, Mareike Baumann, Jonas Gerstner, and Matthias Althoff. 2022. Improving efficiency of human-robot coexistence while guaranteeing safety: Theory and user study.IEEE Transactions on Automation Science and Engineering 20, 4 (2022), 2706–2719

    2022

  24. [32]

    Manh Duong Phung and Quang Phuc Ha. 2021. Safety-enhanced UAV path planning with spherical vector-based particle swarm optimization.Applied Soft Computing107 (2021), 107376

    2021

  25. [33]

    Joohwan Seo, Nikhil Potu Surya Prakash, Alexander Rose, Jongeun Choi, and Roberto Horowitz. 2023. Geometric impedance control on SE (3) for robotic manipulators.IFAC-PapersOnLine56, 2 (2023), 276–283

    2023

  26. [34]

    Claude E Shannon. 1948. A mathematical theory of communication.The Bell system technical journal27, 3 (1948), 379–423

    1948

  27. [35]

    2024.Ipopt v3.14.17

    Vigerske Stefan. 2024.Ipopt v3.14.17. https://github.com/coin-or/Ipopt Accessed: January 2026

    2024

  28. [36]

    2025.ThirdParty-Mumps v3.0.10

    Vigerske Stefan. 2025.ThirdParty-Mumps v3.0.10. https://github.com/coin-or- tools/ThirdParty-Mumps Accessed: January 2026

    2025

  29. [37]

    Sanne Van Waveren, Rasmus Rudling, Iolanda Leite, Patric Jensfelt, and Christian Pek. 2023. Increasing perceived safety in motion planning for human-drone interaction. InProceedings of the 2023 ACM/IEEE international conference on human-robot interaction. 446–455

    2023

  30. [38]

    Chao Yan, Chang Wang, Xiaojia Xiang, Kin Huat Low, Xiangke Wang, Xin Xu, and Lincheng Shen. 2023. Collision-avoiding flocking with multiple fixed- wing UAVs in obstacle-cluttered environments: A task-specific curriculum-based MADRL approach.IEEE Transactions on Neural Networks...

    2023

  31. [39]

    Chao Yan, Xiaojia Xiang, Chang Wang, Fei Li, Xiangke Wang, Xin Xu, and Lincheng Shen. 2023. PASCAL: PopulAtion-Specific Curriculum-based MADRL for collision-free flocking with large-scale fixed-wing UAV swarms.Aerospace Science and Technology133 (2023), 108091

    2023

  32. [40]

    Zheng Yao, Qiwu Zhu, Yanhui Zhang, Haibo Huang, and Min Luo. 2025. Min- imizing Long-Term Energy Consumption in RIS-Assisted UAV-Enabled MEC Network.IEEE Internet of Things Journal(2025)

    2025

  33. [41]

    Chao Yu, Akash Velu, Eugene Vinitsky, Jiaxuan Gao, Yu Wang, Alexandre Bayen, and Yi Wu. 2022. The surprising effectiveness of ppo in cooperative multi-agent games.Advances in neural information processing systems35 (2022), 24611–24624

    2022

  34. [42]

    Weiwei Zhan, Yanjie Chen, Bingwei He, Zhiqiang Miao, Hui Zhang, and Yaonan Wang. 2022. Geometric-based prescribed performance control for unmanned aerial manipulator system under model uncertainties and external disturbances. ISA transactions128 (2022), 367–379

    2022

  35. [43]

    Pengfei Zhang, Yin He, Zhongliu Wang, Shujie Li, and Qinyang Liang. 2024. Research on Multi-UAV Obstacle Avoidance with Optimal Consensus Control Dynamic Airspace Management for UA Vs in Evolving Urban Environments and Improved APF.Drones8, 6 (2024), 248

    2024

  36. [44]

    Chenyang Zhao, Duanfeng Chu, Zejian Deng, and Liping Lu. 2024. Human-like decision making for autonomous driving with social skills.IEEE Transactions on Intelligent Transportation Systems25, 9 (2024), 12269–12284

    2024

  37. [45]

    Yuekuan Zhou. 2025. Unmanned aerial vehicles based low-altitude economy with lifecycle techno-economic-environmental analysis for sustainable and smart cities.Journal of Cleaner Production(2025), 145050

    2025

Pith tools

Reviewed July 11, 2026 · model on record in the stance chip above.