Pith. sign in

REVIEW 3 major objections 6 minor 36 references

Decentralized End-to-End Multi-AAV Pursuit Using Predictive Spatio-Temporal Observation via Deep Reinforcement Learning

T0 review · 3 major / 6 minor · reviewed 2026-07-13 · grok-4.5

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

Pith's one-line read A single decentralized policy that reads raw LiDAR plus predicted intent can capture faster than systems that already know the obstacle map.

desk verdict Solid engineering advance: PSTO + end-to-end LiDAR policy beats privileged SOTA on capture time, scales zero-shot, and flies outdoors; the pure-raw-sensor framing is only partial because intent still rides on shared kinematics. read the letter →

arxiv 2603.24238 v2 pith:NYPVSDWV submitted 2026-03-25 cs.RO

Yude Li , Zhexuan Zhou , Huizhe Li , Yanke Sun , Yenan Wu , Yichen Lai , Yiming Wang , Youmin Gong
show 1 more author
Jie Mei
This is my paper · ORCID
classification cs.RO
keywords multi-agentreinforcementlearningpursuit-evasionaerialswarmsend-to-endcontrolLiDARperceptiondecentralizedpoliciesquadrotor
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

Aerial pursuit teams usually cheat: they receive clean obstacle maps or hand-crafted geometric features before they decide how to fly. This paper argues that is unnecessary. It builds a fixed-size egocentric grid called Predictive Spatio-Temporal Observation (PSTO) that stacks raw LiDAR depth with short-horizon predictions of the evader and of teammates, then trains one shared deep policy that maps that grid straight to continuous velocity commands. In simulation the policy captures faster than strong baselines that still enjoy privileged obstacle information, keeps high success rates under faster evaders and denser clutter, and works for 2-, 3-, or 4-pursuer teams without retraining. Outdoor quadrotor flights with only onboard LiDAR and compute confirm the same behavior transfers. The practical claim is that spatial alignment of geometry and intent is enough to close the perception-to-control loop for cooperative aerial pursuit.

What carries the argument

Predictive Spatio-Temporal Observation (PSTO): a fixed-resolution egocentric spherical grid whose two channels align dense LiDAR obstacle proximity with a sparse heatmap of predicted evader attraction and teammate repulsion; this single tensor is the sole perceptual input to a dual-stream convolutional actor.

What would settle it

Run the identical outdoor 2-vs-1 trials with the communication link for teammate states disabled and with the evader tracked solely by each pursuer’s own LiDAR detector; if success rate or capture time then falls below the privileged-obstacle baselines, the end-to-end claim fails.

Watch

Extended reading notes

Core claim

A decentralized multi-agent policy that maps the two-channel PSTO representation—raw LiDAR proximity plus a predictive intent heatmap—directly to continuous body-frame velocity commands achieves higher capture efficiency and competitive success rates versus state-of-the-art learning methods that receive privileged ground-truth obstacle states, while remaining a single network that scales zero-shot across different team sizes and transfers to outdoor quadrotor hardware.

Load-bearing premise

Teammates must share their kinematic states over a communication link, and in the outdoor flights the evader’s relative position is taken from the shared state estimator rather than pure onboard detection; if either stream fails, the intent channel of PSTO collapses.

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 / 6 minor

Summary. The paper proposes a decentralized MARL framework for multi-AAV pursuit–evasion in cluttered arenas that maps a two-channel egocentric representation—Predictive Spatio-Temporal Observation (PSTO)—to continuous body-frame velocity commands. PSTO projects raw LiDAR ranges into a fixed-resolution spherical proximity map and overlays a predictive intent heatmap built from an LSTM forecast of the evader and linear extrapolations of teammate trajectories (Eqs. 3–11). Under CTDE/MAPPO with a shaped reward and progressive curriculum, a single shared policy is trained to navigate obstacles, intercept a faster APF-controlled evader, and maintain encirclement. Simulations (Table I, 150 rollouts) report competitive success rates and shorter capture times than traditional heuristics and the privileged-state OPEN baseline, plus zero-shot scaling across 2/3/4-vs-1 teams (Table II). Fully autonomous outdoor 2-vs-1 flights on quadrotors with onboard LiDAR and NUC compute are presented as zero-shot validation.

Significance. If the claims hold under the paper’s stated sensing and communication assumptions, the work is a solid systems-level contribution to multi-UAV pursuit: it is among the first to close a dense LiDAR-to-control loop for cooperative aerial P–E rather than operating on hand-crafted geometric features or ground-truth obstacle maps. Strengths that should be credited include (i) a carefully designed ablation suite isolating cooperation, intent, and spatial alignment (Table I; Fig. 4), (ii) a curriculum that yields a single policy that generalizes across team sizes without retraining (Table II), and (iii) outdoor hardware transfer with measured onboard latency (~3.87 ms). The spatial unification of dense geometry and sparse predictive intent in a fixed-resolution grid is a reusable representation idea for other multi-agent aerial tasks. The result is incremental rather than foundational, but it is practically relevant for infrastructure-free swarm deployment.

major comments (3)
  1. [Abstract; Sec. I; Sec. III-B (Eqs. 4, 8); Sec. IV-E] The abstract and Sec. I repeatedly frame the method as mapping “raw LiDAR observations directly” to control and as relying on “only onboard sensing and computing.” This is not accurate for the full observation. Sec. III-B-(iii) and Eq. (8) inject perfect world-frame teammate positions and velocities obtained over a communication network into H_Teammate; Eq. (4) treats evader observability as shared (“if observable by any pursuer”); and Sec. IV-E replaces onboard evader detection with shared Swarm-LIO estimates as an “experimental simplification.” Consequently the H_intent channel—and therefore the spatially aligned representation credited for the 10–15-step CT gains—is not generated from local raw sensors alone. Either the abstract/claims must be rewritten to state the communication and shared-localization assumptions explicitly, or the paper must add robustness experiments (noisy, delay
  2. [Sec. IV-E; Abstract; Sec. V] Outdoor validation (Sec. IV-E, Fig. 1, Fig. 5) is valuable but does not close the perception loop claimed in the abstract. The virtual boundary is injected as artificial LiDAR points, and the evader’s relative state is taken from the shared Swarm-LIO estimate rather than an onboard detector. The paper acknowledges this as a simplification, yet still concludes that the framework is validated “using solely onboard sensing and computation.” A major revision should either (a) report at least one trial with an independent onboard evader detector (or a controlled noise model on the proxy) or (b) clearly demote the outdoor result to a coordination/control transfer test under privileged relative-state access, and adjust the abstract and conclusions accordingly. Without that, the hardware experiment does not fully support the perception-to-control narrative.
  3. [Sec. IV-C; Table I; Fig. 4] The efficiency advantage over OPEN (Table I: roughly 10–15 fewer capture steps) is attributed to “projection-level coupling” of intent and geometry (Sec. IV-C). OPEN receives privileged ground-truth obstacle states but processes them as disjoint tokens; PSTO receives dense LiDAR plus shared kinematic intent already projected into the same grid. The comparison therefore confounds representation design with information structure (privileged map vs. sensor grid; vector intent vs. spatially aligned heatmap). To make the causal claim load-bearing, the paper should either (i) give OPEN an equivalent dense obstacle encoding or (ii) add a controlled variant of PSTO that receives the same privileged obstacle map as OPEN while keeping the aligned intent channel. Absent that, the statement that PSTO is superior specifically because of spatial alignment remains only partially isolated, even though t
minor comments (6)
  1. [Title; Abstract] Title and running text inconsistently space “AA V” / “AAV” (e.g., title “Multi-AA V”). Normalize to “AAV” throughout.
  2. [Sec. III-C-1; Fig. 2] Fig. 2 system diagram is described but, from the manuscript text, the dual-stream backbone (dilated conv + SE on the heatmap stream) is only briefly justified as “empirically” better. A short quantitative comparison (shared CNN vs. dual-stream) in an appendix or table would strengthen Sec. III-C-1.
  3. [Sec. III-B; Sec. IV-A] Hyperparameters of PSTO (Vd×Hd = 120×6, λe, λt, Vevader, Vteammate, ddes, σcoord) and domain-randomization noise magnitudes are free design choices. Listing their values and any sensitivity study (even brief) would aid reproducibility.
  4. [Table I; Sec. IV-C] In Table I, at several high-speed/high-obstacle cells OPEN has slightly higher SR while PSTO has lower CT. The text correctly says “competitive success rates” and “superior capture efficiency,” but a one-sentence discussion of the SR–CT trade-off would prevent readers from over-reading the boldface pattern.
  5. [Sec. II] Related work claims “to our knowledge, the first multi-AAV framework to learn a decentralized P–E policy directly from raw LiDAR data.” Soften or qualify given concurrent single-agent LiDAR RL and multi-agent state-based P–E lines already cited; the precise novelty is the unified PSTO for multi-agent aerial pursuit.
  6. [Sec. III-A, Eq. (1)] Eq. (1) writes the objective without the usual expectation over trajectories under the joint policy; a standard Dec-POMDP return notation would be clearer.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: empirical MAPPO policy on constructed PSTO representation, evaluated on external capture/collision metrics; self-cites are background only.

full rationale

The paper's load-bearing claims are empirical performance numbers (success rate and capture time under MAPPO training) of a decentralized policy that consumes a hand-constructed two-channel egocentric grid (raw LiDAR min-range proximity plus an intent heatmap built from an LSTM forecast of the evader and linear extrapolation of shared teammate velocities). Capture is defined by an external geometric threshold (min distance < 1.4 m), collisions and timeouts are likewise external, and baselines receive either privileged ground-truth obstacles or ablated inputs; none of these quantities is obtained by fitting a free parameter to the same data that is later reported as a prediction. The sole self-citation of note ([27], overlapping authors) supplies only the spherical-projection formula already written out as Eq. (2); it is not invoked as a uniqueness theorem, ansatz, or uniqueness result that forces the multi-agent claim. Shared teammate kinematics and the outdoor Swarm-LIO proxy for the evader are experimental assumptions, not definitional reductions of the reported metrics. Consequently the derivation chain contains no self-definitional step, no fitted-input-called-prediction, and no load-bearing self-citation chain.

Assumptions & free parameters 5 free parameters · 4 assumptions · 1 invented entities

The central empirical claim rests on standard MARL machinery plus a handful of hand-chosen geometric and curriculum constants that define the observation and reward. No new physical constants are introduced; the invented entity is the PSTO tensor itself.

free parameters (5)
  • PSTO grid resolution Vd×Hd = 120×6
    Fixed at 120×6 covering 360°×30°; chosen by authors, not derived.
  • Temporal decay factors λe, λt and strengths Vevader, Vteammate
    Hand-tuned scalars that control how strongly predicted points paint the intent heatmap (Eqs. 6,9).
  • Desired inter-agent spacing ddes and coordination bandwidth σcoord
    Gaussian potential parameters in the dense reward; set by curriculum design.
  • Capture radius curriculum schedule
    Starts large and shrinks; together with speed/obstacle ramps it shapes the final policy.
  • Domain-randomization noise magnitudes on proprioception and relative observations
    Gaussian noise levels chosen to aid sim-to-real; not reported numerically.
assumptions (4)
  • domain assumption The multi-pursuer single-evader task is well-modeled as a Dec-POMDP with shared reward and parameter-shared policies.
    Stated in Sec. III-A; standard but not proved for continuous aerial dynamics.
  • domain assumption Linear velocity extrapolation of teammates over the short horizon Tfuture is sufficiently accurate for coordination.
    Sec. III-B-iii; justified by communication of precise velocities but remains an approximation.
  • domain assumption MAPPO with a centralized critic and Beta-distributed continuous actions converges to a useful cooperative policy under the given curriculum.
    Training procedure in Sec. III-C; empirical success is shown but not guaranteed.
  • standard math Spherical projection of 3-D points onto a fixed 2-D grid (Eq. 2) preserves the geometric relationships needed for collision-free interception.
    Taken from prior single-agent work [27]; used without re-derivation.
invented entities (1)
  • Predictive Spatio-Temporal Observation (PSTO)
    purpose: Unified two-channel egocentric tensor that spatially aligns dense LiDAR proximity with sparse predictive heatmaps of evader and teammates so a CNN can reason jointly about geometry and intent.
    Defined in Sec. III-B and Fig. 3; the paper’s central technical contribution. Independent evidence is limited to the ablation gains and outdoor flights reported here; no external replication yet.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Decentralized End-to-End Multi-AAV Pursuit Using Predictive Spatio-Temporal Observation via Deep Reinforcement Learning." pith.science (2026). https://pith.science/paper/NYPVSDWV

@misc{pith2026260324238,
  author       = {Pith},
  title        = {Pith review of: Decentralized End-to-End Multi-AAV Pursuit Using Predictive Spatio-Temporal Observation via Deep Reinforcement Learning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/NYPVSDWV}},
  note         = {Machine review of arXiv:2603.24238}
}
read the original abstract

Decentralized cooperative pursuit in cluttered environments is challenging for autonomous aerial swarms, especially under partial and noisy perception. Existing methods often rely on abstracted geometric features or privileged ground-truth states, and therefore sidestep perceptual uncertainty in real-world settings. We propose a decentralized end-to-end multi-agent reinforcement learning (MARL) framework that maps raw LiDAR observations directly to continuous control commands. Central to the framework is the Predictive Spatio-Temporal Observation (PSTO), an egocentric grid representation that aligns obstacle geometry with predictive adversarial intent and teammate motion in a unified, fixed-resolution projection. Built on PSTO, a single decentralized policy enables agents to navigate static obstacles, intercept dynamic targets, and maintain cooperative encirclement. Simulations demonstrate that the proposed method achieves superior capture efficiency and competitive success rates compared to state-of-the-art learning-based approaches relying on privileged obstacle information. Furthermore, the unified policy scales seamlessly across different team sizes without retraining. Finally, fully autonomous outdoor experiments validate the framework on a quadrotor swarm relying on only onboard sensing and computing.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

36 extracted references

  1. [1]

    Racer: Rapid collaborative exploration with a decentralized multi-uav system,

    B. Zhou, H. Xu, and S. Shen, “Racer: Rapid collaborative exploration with a decentralized multi-uav system,”IEEE Transactions on Robotics, vol. 39, no. 3, pp. 1816–1835, 2023

    2023

  2. [2]

    Swarm of micro flying robots in the wild,

    X. Zhou, X. Wen, Z. Wang, Y . Gao, H. Li, Q. Wang, T. Yang, H. Lu, Y . Cao, C. Xuet al., “Swarm of micro flying robots in the wild,”Science Robotics, vol. 7, no. 66, p. eabm5954, 2022

    2022

  3. [3]

    Proximal cooperative aerial manipulation with vertically stacked drones,

    H. Cao, J. Shen, Y . Zhang, Z. Fu, C. Liu, S. Sun, and S. Zhao, “Proximal cooperative aerial manipulation with vertically stacked drones,”Nature, pp. 1–8, 2025

    2025

  4. [4]

    Primitive-swarm: An ultra-lightweight and scalable planner for large- scale aerial swarms,

    J. Hou, X. Zhou, N. Pan, A. Li, Y . Guan, C. Xu, Z. Gan, and F. Gao, “Primitive-swarm: An ultra-lightweight and scalable planner for large- scale aerial swarms,”IEEE Transactions on Robotics, 2025

    2025

  5. [5]

    Search and pursuit-evasion in mobile robotics: A survey,

    T. H. Chung, G. A. Hollinger, and V . Isler, “Search and pursuit-evasion in mobile robotics: A survey,”Autonomous robots, vol. 31, no. 4, pp. 299–316, 2011

    2011

  6. [6]

    Isaacs,Differential games: a mathematical theory with applications to warfare and pursuit, control and optimization

    R. Isaacs,Differential games: a mathematical theory with applications to warfare and pursuit, control and optimization. Courier Corporation, 1999

    1999

  7. [7]

    Proba- bilistic pursuit-evasion games: theory, implementation, and experimental evaluation,

    R. Vidal, O. Shakernia, H. J. Kim, D. H. Shim, and S. Sastry, “Proba- bilistic pursuit-evasion games: theory, implementation, and experimental evaluation,”IEEE transactions on robotics and automation, vol. 18, no. 5, pp. 662–669, 2002

    2002

  8. [8]

    Multi-robot cooperative pursuit via potential field-enhanced reinforcement learning,

    Z. Zhang, X. Wang, Q. Zhang, and T. Hu, “Multi-robot cooperative pursuit via potential field-enhanced reinforcement learning,” in2022 International Conference on Robotics and Automation (ICRA). IEEE, 2022, pp. 8808–8814

    2022

Show all 36 references
  1. [9]

    Online planning for multi-uav pursuit-evasion in unknown environments using deep reinforcement learning,

    J. Chen, C. Yu, G. Li, W. Tang, S. Ji, X. Yang, B. Xu, H. Yang, and Y . Wang, “Online planning for multi-uav pursuit-evasion in unknown environments using deep reinforcement learning,”IEEE Robotics and Automation Letters, vol. 10, no. 8, pp. 8196–8203, 2025

    2025

  2. [10]

    Game of drones: Intelligent online decision making of multi-uav confrontation,

    D. Liu, Q. Zong, X. Zhang, R. Zhang, L. Dou, and B. Tian, “Game of drones: Intelligent online decision making of multi-uav confrontation,” IEEE Transactions on Emerging Topics in Computational Intelligence, vol. 8, no. 2, pp. 2086–2100, 2024

  3. [11]

    Pursuit-evasion game strategy of usv based on deep reinforcement learning in complex multi-obstacle environment,

    X. Qu, W. Gan, D. Song, and L. Zhou, “Pursuit-evasion game strategy of usv based on deep reinforcement learning in complex multi-obstacle environment,”Ocean Engineering, vol. 273, p. 114016, 2023

    2023

  4. [12]

    Champion-level drone racing using deep reinforcement learning,

    E. Kaufmann, L. Bauersfeld, A. Loquercio, M. M ¨uller, V . Koltun, and D. Scaramuzza, “Champion-level drone racing using deep reinforcement learning,”Nature, vol. 620, no. 7976, pp. 982–987, 2023

    2023

  5. [13]

    Navrl: Learning safe flight in dynamic environments,

    Z. Xu, X. Han, H. Shen, H. Jin, and K. Shimada, “Navrl: Learning safe flight in dynamic environments,”IEEE Robotics and Automation Letters, 2025

    2025

  6. [14]

    You only plan once: A learning-based one-stage planner with guidance learning,

    J. Lu, X. Zhang, H. Shen, L. Xu, and B. Tian, “You only plan once: A learning-based one-stage planner with guidance learning,”IEEE Robotics and Automation Letters, vol. 9, no. 7, pp. 6083–6090, 2024

    2024

  7. [15]

    Learning vision-based agile flight via differentiable physics,

    Y . Zhang, Y . Hu, Y . Song, D. Zou, and W. Lin, “Learning vision-based agile flight via differentiable physics,”Nature Machine Intelligence, pp. 1–13, 2025

    2025

  8. [16]

    Omnidrones: An efficient and flexible platform for reinforcement learning in drone control,

    B. Xu, F. Gao, C. Yu, R. Zhang, Y . Wu, and Y . Wang, “Omnidrones: An efficient and flexible platform for reinforcement learning in drone control,”IEEE Robotics and Automation Letters, vol. 9, no. 3, pp. 2838– 2844, 2024

    2024

  9. [17]

    Distributed pursuit of an evader with collision and obstacle avoidance,

    B. Tian, P. Li, H. Lu, Q. Zong, and L. He, “Distributed pursuit of an evader with collision and obstacle avoidance,”IEEE Transactions on Cybernetics, vol. 52, no. 12, pp. 13 512–13 520, 2021

    2021

  10. [18]

    Intercepting rogue robots: An algorithm for capturing multiple evaders with multiple pursuers,

    A. Pierson, Z. Wang, and M. Schwager, “Intercepting rogue robots: An algorithm for capturing multiple evaders with multiple pursuers,”IEEE Robotics and Automation Letters, vol. 2, no. 2, pp. 530–537, 2016

    2016

  11. [19]

    Real-time obstacle avoidance for manipulators and mobile robots,

    O. Khatib, “Real-time obstacle avoidance for manipulators and mobile robots,”The international journal of robotics research, vol. 5, no. 1, pp. 90–98, 1986

    1986

  12. [20]

    Collective predation and escape strategies,

    L. Angelani, “Collective predation and escape strategies,”Physical review letters, vol. 109, no. 11, p. 118104, 2012

    2012

  13. [21]

    Group chasing tactics: how to catch a faster prey,

    M. Janosov, C. Vir ´agh, G. V ´as´arhelyi, and T. Vicsek, “Group chasing tactics: how to catch a faster prey,”New Journal of Physics, vol. 19, no. 5, p. 053003, 2017

    2017

  14. [22]

    Potential field methods and their inherent limitations for mobile robot navigation

    Y . Koren, J. Borensteinet al., “Potential field methods and their inherent limitations for mobile robot navigation.” inIcra, vol. 2, no. 1991, 1991, pp. 1398–1404

    1991

  15. [23]

    Decentralized multi-agent pursuit using deep reinforcement learning,

    C. De Souza, R. Newbury, A. Cosgun, P. Castillo, B. Vidolov, and D. Kuli ´c, “Decentralized multi-agent pursuit using deep reinforcement learning,”IEEE Robotics and Automation Letters, vol. 6, no. 3, pp. 4552–4559, 2021

    2021

  16. [24]

    Cooperative control for multi-player pursuit-evasion games with reinforcement learning,

    Y . Wang, L. Dong, and C. Sun, “Cooperative control for multi-player pursuit-evasion games with reinforcement learning,”Neurocomputing, vol. 412, pp. 101–114, 2020

    2020

  17. [25]

    Cooperative multi-agent control using deep reinforcement learning,

    J. K. Gupta, M. Egorov, and M. Kochenderfer, “Cooperative multi-agent control using deep reinforcement learning,” inInternational conference on autonomous agents and multiagent systems. Springer, 2017, pp. 66–83

    2017

  18. [26]

    The complexity of decentralized control of markov decision processes,

    D. S. Bernstein, R. Givan, N. Immerman, and S. Zilberstein, “The complexity of decentralized control of markov decision processes,” Mathematics of operations research, vol. 27, no. 4, pp. 819–840, 2002

    2002

  19. [27]

    Agile in the face of delay: Asynchronous end-to-end learning for real-world aerial navigation,

    Y . Li, Z. Zhou, H. Li, Y . Gong, and J. Mei, “Agile in the face of delay: Asynchronous end-to-end learning for real-world aerial navigation,” arXiv preprint arXiv:2509.13816, 2025

    2025

  20. [28]

    What the constant velocity model can teach us about pedestrian motion prediction,

    C. Sch ¨oller, V . Aravantinos, F. Lay, and A. Knoll, “What the constant velocity model can teach us about pedestrian motion prediction,”IEEE Robotics and Automation Letters, vol. 5, no. 2, pp. 1696–1703, 2020

    2020

  21. [29]

    Multi-scale context aggregation by dilated convolutions,

    F. Yu and V . Koltun, “Multi-scale context aggregation by dilated convolutions,” inICLR, 2016

    2016

  22. [30]

    Squeeze-and-excitation networks,

    J. Hu, L. Shen, and G. Sun, “Squeeze-and-excitation networks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 7132–7141

    2018

  23. [31]

    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,” Advances in neural information processing systems, vol. 35, pp. 24 611– 24 624, 2022

    2022

  24. [32]

    Multi-agent actor-critic for mixed cooperative-competitive environ- ments,

    R. Lowe, Y . I. Wu, A. Tamar, J. Harb, O. Pieter Abbeel, and I. Mordatch, “Multi-agent actor-critic for mixed cooperative-competitive environ- ments,”Advances in neural information processing systems, vol. 30, 2017

    2017

  25. [33]

    Improving stochastic policy gradients in continuous control with deep reinforcement learning using the beta distribution,

    P.-W. Chou, D. Maturana, and S. Scherer, “Improving stochastic policy gradients in continuous control with deep reinforcement learning using the beta distribution,” inInternational conference on machine learning. PMLR, 2017, pp. 834–843

    2017

  26. [34]

    Curriculum learning,

    Y . Bengio, J. Louradour, R. Collobert, and J. Weston, “Curriculum learning,” inProceedings of the 26th annual international conference on machine learning, 2009, pp. 41–48

    2009

  27. [35]

    Domain randomization for transferring deep neural networks from simulation to the real world,

    J. Tobin, R. Fong, A. Ray, J. Schneider, W. Zaremba, and P. Abbeel, “Domain randomization for transferring deep neural networks from simulation to the real world,” in2017 IEEE/RSJ international conference on intelligent robots and systems (IROS). IEEE, 2017, pp. 23–30

    2017

  28. [36]

    Swarm-lio2: Decentralized, efficient lidar-inertial odometry for uav swarms,

    F. Zhu, Y . Ren, L. Yin, F. Kong, Q. Liu, R. Xue, W. Liu, Y . Cai, G. Lu, H. Liet al., “Swarm-lio2: Decentralized, efficient lidar-inertial odometry for uav swarms,”IEEE Transactions on Robotics, 2024

    2024

Pith tools

Reviewed July 13, 2026 · model on record in the stance chip above.