Pith. sign in

REVIEW 4 major objections 5 minor 1 cited by

LLM-Land: Large Language Models for Context-Aware Drone Landing

T0 review · 4 major / 5 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 A RAG-grounded LLM raises drone landing success from 34% to 96% in dynamic scenes, and from 28% to 74–80% in cluttered ones.

desk verdict Useful integration of RAG-stabilized LLMs with MPC for drone landing, but the leap from a JSON scalar to a 3D unsafe region is underspecified, so the headline numbers can't yet be assigned to the LLM. read the letter →

arxiv 2505.06399 v1 pith:G36QHZ46 submitted 2025-05-09 cs.RO

Siwei Cai , Yuwei Wu , Lifeng Zhou This is my paper · ORCID
classification cs.RO
keywords autonomousdronelandinglargelanguagemodelsretrieval-augmentedgenerationmodelpredictivecontrolvision-languageencoderdynamicobstacleavoidancesafetybuffersROS-Gazebosimulation
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

LLM-Land claims that a lightweight LLM, grounded by retrieval-augmented generation (RAG), can turn raw camera captions into reliable safety constraints that let an MPC planner land a drone safely in dynamic and cluttered environments. In ROS-Gazebo simulations, the framework achieves 96% landing success with dynamic obstacles and 74–80% in cluttered scenes, compared with 34% and 28–34% for a vision-based MPC baseline and only 6–12% for the same LLM without RAG. The paper positions the semantics it adds as the missing piece: context-aware safety buffers such as 3 meters for pedestrians and 5 meters for vehicles replace fixed margins. If true, it shows a practical route to embedding semantic reasoning in real-time control loops on resource-constrained drones.

What carries the argument

The load-bearing mechanism is a RAG-grounded semantic reasoning stage that outputs exactly two fields, an `is_dynamic` flag and a minimum clearance, from BLIP image captions. The knowledge base stores obstacle classes with nominal safety buffers; retrieval by hybrid dense-keyword search over BLIP captions selects the top three segments, and the LLM is constrained by prompting and a JSON template. The MPC then treats the LLM's unsafe region as a convex set $B$ to be excluded from the feasible corridor $\tilde{\mathcal{P}}^c \subseteq \mathcal{P}^c$, and a heuristic penalty biases the trajectory search away from it. RAG is thus the stabilizing element: without it the same LLM produces inconsistent metadata that lowers success below the no-LLM baseline.

What would settle it

An adversarial sequence where a moving pedestrian crosses the landing path but BLIP's caption does not mention any person—heavy occlusion or an unusual pose—should drive success down toward the baseline MPC's 34%, because the unsafe region is never created. Conversely, running the same captions through the LLM with the knowledge base removed but a well-crafted prompt should reproduce the paper's 6% LLM-MPC failure rate if RAG is doing the causal work.

Watch

Extended reading notes

Core claim

The paper's central discovery is that grounding a small LLM's outputs with a retrieval knowledge base makes the difference between useless and reliable semantic guidance for an MPC landing planner. With bare BLIP captions fed to a 1B-parameter LLM, the planner fails 94% of the time in dynamic scenarios; adding RAG retrieval of obstacle classifications and safety buffers lifts success to 96%. The LLM outputs a structured JSON flag for dynamic obstacles and a clearance height, which the MPC encodes as an additional convex unsafe region to exclude from the corridor, triggering replanning. The same pattern holds across urban and forest-edge scenes, where LLM-Land reaches 74–80% success versus 28–34% for the baseline. The authors attribute the improvement to RAG suppressing hallucinations and supplying context-appropriate safety margins.

Load-bearing premise

The system assumes that a single image caption, filtered through the hand-built retrieval knowledge base, yields the correct dynamic flag and safety clearance for every obstacle in view; a caption that omits a person or mislabels a moving vehicle as static produces a wrong unsafe region and the MPC can plan straight into the hazard.

Editorial extensions

If this is right

  • RAG grounding makes lightweight LLMs practical for safety-critical drone landing on constrained hardware, with about 1.5 GB VRAM and 1.45 seconds latency on a laptop GPU.
  • Context-aware safety buffers, rather than fixed margins, are the main enabler of safe landing in dynamic scenes.
  • The LLM-MPC hybrid architecture transfers across qualitatively different environments (open field, urban, forest edge) with success rates above 74%.
  • Quantized small models (Llama-3.2-1B) outperform larger ones in this loop because faster, steadier JSON output beats slower, more capable inference.

Reading between the lines

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

  • The same RAG-grounded semantic layer could be adapted to other safety-critical robot tasks—navigation, manipulation, or surveillance—where obstacle identity and intent matter, not just position.
  • A testable extension is to update the knowledge base online from operational experience, which the authors list as future work; if effective, it would let the system adjust safety buffers to local behavior.
  • The single-frame caption bottleneck suggests that replacing BLIP with a richer vision-language encoder, or fusing multiple frames, could close the 'missed person' failure mode the paper acknowledges in its Limitations.
  • The authors' comparison isolates RAG as the causal ingredient; a natural follow-up is to test whether a fixed, well-tuned prompt without retrieval achieves similar gains, which would weaken the RAG-specific claim.
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

4 major / 5 minor

Summary. The paper proposes LLM-Land, a hybrid perception-action framework for autonomous drone landing in dynamic, unstructured environments. A vision-language encoder (BLIP) generates captions from a forward camera; a lightweight LLM with retrieval-augmented generation (RAG) classifies obstacles and outputs two JSON fields, is_dynamic and z_min, representing dynamic-object classification and a minimum-altitude/safety-buffer value. These outputs feed an MPC-based planner that excludes an 'unsafe region' from its feasible corridor during landing. The authors evaluate the system in ROS-Gazebo across three scenarios, reporting success rates of 94–96% for LLM-Land versus 28–34% for a depth-based baseline MPC and 6–12% for an LLM-MPC variant without RAG. They also benchmark several quantized lightweight LLMs in terms of latency, VRAM, and landing success, and include a qualitative real-world indoor flight test.

Significance. If the central claim holds, the paper would demonstrate a practical way to inject semantic, context-aware safety reasoning into real-time MPC-based landing, with a modular pipeline that is plausible for onboard deployment using quantized small LLMs. The use of RAG to stabilize lightweight LLM outputs, the reported latency/VRAM measurements, and the multi-environment simulation evaluation are useful empirical contributions. However, the paper currently does not specify how the LLM's two JSON outputs are converted into the three-dimensional unsafe region that the MPC excludes, and the baseline comparison is not sufficiently controlled. Until that mapping is made explicit, the reported large success-rate improvements cannot be causally attributed to the LLM-RAG semantic reasoning, so the paper's headline claim is not yet verifiable in its present form.

major comments (4)
  1. [§3.1–3.2] The perception subsystem is stated to publish exactly two JSON fields, is_dynamic and z_min, but the action subsystem excludes a convex set B described as 'the unsafe region associated with the dynamic obstacle' from the MPC corridor, and the paper never specifies how B is constructed from that two-field output. Section 6 explicitly acknowledges that the LLM reasoning does not embed precise obstacle geometry or spatial relationships, so a single scalar z_min cannot locate a 3D convex set. Because B is the mechanism that actually changes the MPC's feasible set, the reported improvement from 34% to 96% cannot be attributed to the LLM-RAG reasoning until the mapping from the JSON output to B is specified and shown to be the only source of the new constraint.
  2. [§4.2, §4.3] The baseline MPC is described only as relying on geometric obstacle information from a depth sensor, with no specification of how obstacles are detected, how the corridors P_c are built, or how the controller is tuned; the 34% success rate in an open field suggests it fails to avoid a single moving obstacle, which is surprising for a depth-based planner. The authors should provide the baseline implementation details and confirm that the baseline is not inadvertently undertuned; otherwise the comparison against LLM-Land is not a controlled test of the semantic contribution.
  3. [§4.5, Table 3] Table 3 and its accompanying text contradict each other: the text reports mid-sized model latency of 1010.2–1266.5 ms, while the table lists 1510.2 ms for Qwen2.5-1.5B-R1-Distilled and 1889.5 ms for Qwen2.5-3B-Instruct; the table's 'Qwen2.5–3B' also appears as 'Qwen2.5-3B-Instruct' in the text. These discrepancies must be reconciled before the hardware benchmark can be assessed.
  4. [§4.3–4.5] All success rates are reported as point estimates over 50 trials with no confidence intervals or significance tests; for binary outcomes with n=50, the standard error is roughly 4–7 percentage points, so the differences among LLM-Land variants in Table 3 (94%, 70%, 82%, 60%) are not statistically distinguishable without additional analysis, and even comparisons such as 94% versus 96% in Table 2 lack support. The authors should report binomial confidence intervals or raw success counts and restrict comparative claims to comparisons that are statistically separable.
minor comments (5)
  1. [Figures 1 and 3, §3.1] The pipeline figures show JSON containing a 'clearance' field with values like 3 m, but Section 3.1 states that the pipeline outputs exactly two fields, is_dynamic and z_min; the relationship between 'clearance' and 'z_min' and the safety-buffer values (3 m pedestrian, 5 m vehicle) should be clarified.
  2. [§3.1] The text claims 'determinism' with temperature τ≤0.5; a temperature of 0.5 is not deterministic, so the wording should be changed to 'reduced stochasticity' or similar.
  3. [§5, Appendix A.4] The conclusion states that the system was demonstrated to run effectively in real time on resource-constrained hardware such as the NVIDIA Jetson Orin NX, but all experiments in Section 4 used a laptop with an RTX 3070 Ti GPU and the Jetson Orin NX appears nowhere in the reported experiments; this claim should be removed or supported with data.
  4. [§4.3] The success metric uses a 1 m close-call buffer, while the LLM-Land knowledge base specifies safety buffers of 3 m for pedestrians and 5 m for vehicles; the paper should state whether the 1 m metric is separate from the LLM-defined buffers and how the two interact in the evaluation.
  5. [Appendix A.4] The real-world demonstration is a single qualitative indoor flight test with offboard computation; it should be explicitly labeled as a feasibility demonstration and not presented as evidence of system performance or onboard capability.

Circularity Check

0 steps flagged · score 0.0 of 10

No circularity: measured landing-success results, hand-authored RAG buffers, and independent evaluation criteria.

full rationale

The paper makes no derived prediction: its headline numbers (94-96% and 74-80%) are measured simulator outcomes, not quantities obtained by solving equations from stated premises. The safety buffers (3 m for pedestrians, 5 m for vehicles) are hand-authored entries in the RAG knowledge base and are retrieved, not fit to the success metric; the evaluation criteria (0.5 m landing radius, 1 m close-call buffer) are independent of the LLM's emitted clearance values. The central empirical comparison (Baseline MPC vs LLM-MPC vs LLM-Land) is an ablation, and the RAG's value is demonstrated by the drop from 96% to 6% when RAG is removed. The only citations from the author group used in the architecture ([18], [35], [9]) are background, a survey, and an external MPC framework; none is used to forbid alternatives or to justify the semantic-safety claim. Section 6's admission that captions carry no obstacle geometry is a limitation for reproducibility, not a circular reduction: the B-set construction is under-specified, but under-specification is not equivalence-by-construction. No equation in the manuscript reduces to a fitted value or to a self-citation chain.

Assumptions & free parameters 3 free parameters · 4 assumptions · 0 invented entities

The central claims rest on hand-chosen safety buffers, manual knowledge-base entries, caption quality, and simulator representativeness. No new physical entities are introduced. The system does not fit its parameters to the target result, so circularity burden is low; the main exposure is scenario tuning and evaluation metric choice.

free parameters (3)
  • Contextual safety buffers (pedestrian 3 m, vehicle 5 m) = 3 m / 5 m
    Hand-authored in the RAG knowledge base and retrieved by the LLM; their values directly determine how conservative the replanner is, and no sensitivity analysis is reported.
  • Heuristic penalty weighting (lambda) = Not reported
    Appendix A.3 introduces lambda greater than 0 in the modified trajectory-search heuristic; the choice affects avoidance aggressiveness and is not specified.
  • Evaluation thresholds (landing success radius 0.5 m, close-call buffer 1 m) = 0.5 m / 1 m
    Chosen by the authors for the success metric; reported success rates depend on these thresholds and on the LLM output safety buffers.
assumptions (4)
  • domain assumption BLIP captions accurately represent all safety-relevant objects and their dynamics.
    Invoked in Section 3.1 where the perception pipeline converts images to captions; the Limitations section acknowledges the model does not embed precise obstacle geometry.
  • domain assumption RAG knowledge base entries for obstacle classes and safety buffers are correct and complete.
    Section 3.1 builds the knowledge base by hand; incorrect entries would produce incorrect unsafe-region constraints.
  • domain assumption Gazebo/RotorS simulation with the chosen dynamic obstacle models is representative of real landing scenarios.
    Section 4.1; no real-world dynamic-obstacle trials are reported, only an indoor qualitative manikin test.
  • standard math MPC model and trajectory-search framework from Wu et al. [35] are accurate for the simulated quadrotor.
    The action system follows Wu et al. [35], and the paper does not re-derive it; this is a standard prior result.

how reviews work

0 comments
Cite this review

Pith. "Pith review of LLM-Land: Large Language Models for Context-Aware Drone Landing." pith.science (2026). https://pith.science/paper/G36QHZ46

@misc{pith2026250506399,
  author       = {Pith},
  title        = {Pith review of: LLM-Land: Large Language Models for Context-Aware Drone Landing},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/G36QHZ46}},
  note         = {Machine review of arXiv:2505.06399}
}
read the original abstract

Autonomous landing is essential for drones deployed in emergency deliveries, post-disaster response, and other large-scale missions. By enabling self-docking on charging platforms, it facilitates continuous operation and significantly extends mission endurance. However, traditional approaches often fall short in dynamic, unstructured environments due to limited semantic awareness and reliance on fixed, context-insensitive safety margins. To address these limitations, we propose a hybrid framework that integrates large language model (LLMs) with model predictive control (MPC). Our approach begins with a vision-language encoder (VLE) (e.g., BLIP), which transforms real-time images into concise textual scene descriptions. These descriptions are processed by a lightweight LLM (e.g., Qwen 2.5 1.5B or LLaMA 3.2 1B) equipped with retrieval-augmented generation (RAG) to classify scene elements and infer context-aware safety buffers, such as 3 meters for pedestrians and 5 meters for vehicles. The resulting semantic flags and unsafe regions are then fed into an MPC module, enabling real-time trajectory replanning that avoids collisions while maintaining high landing precision. We validate our framework in the ROS-Gazebo simulator, where it consistently outperforms conventional vision-based MPC baselines. Our results show a significant reduction in near-miss incidents with dynamic obstacles, while preserving accurate landings in cluttered environments.

Figures

Figures reproduced from arXiv: 2505.06399 by the authors.

Figure 1
Figure 1. Overview of the LLM-Land framework for semantics-driven safe landing. (A) The Per [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Experiments were conducted using the Gazebo simulator, which provides realistic physics [PITH_FULL_IMAGE:figures/full_fig_p006_2.png] view at source ↗
Figure 3
Figure 3. Illustration of the LLM-land pipeline from visual perception to semantic reasoning and [PITH_FULL_IMAGE:figures/full_fig_p013_3.png] view at source ↗
Figures from the paper (2 more)
Figure 4
Figure 4. Figure 4: Indoor flight test with the custom-built UAV. The UAV detected a manikin near its path, [PITH_FULL_IMAGE:figures/full_fig_p014_4.png]
Figure 5
Figure 5. Figure 5: Lightweight UAV platform based on VOXL and PX4. [PITH_FULL_IMAGE:figures/full_fig_p014_5.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 1 Pith paper

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. Large Language Models for Next-Generation Wireless Network Management: A Survey and Tutorial

    cs.NI 2025-09 conditional novelty 4.0 of 10

    A survey and tutorial that organizes LLM-enabled wireless network optimization into formulation, solution, and verification stages, with case studies drawn from the authors' own prior papers.

Reference graph

Works this paper leans on

39 extracted references · 24 canonical work pages · cited by 1 Pith paper

  1. [1]

    Kingston, R

    D. Kingston, R. W. Beard, and R. S. Holt. Decentralized perimeter surveillance using a team of uavs. IEEE Transactions on Robotics, 24(6):1394–1404, 2008

    work page 2008

  2. [2]

    Saunders, S

    J. Saunders, S. Saeedi, and W. Li. Autonomous aerial robotics for package delivery: A techni- cal review. Journal of Field Robotics, 41(1):3–49, 2024

    work page 2024

  3. [3]

    Aela, H.-L

    P. Aela, H.-L. Chi, A. Fares, T. Zayed, and M. Kim. Uav-based studies in railway infrastructure monitoring. Automation in Construction, 167:105714, 2024

    work page 2024

  4. [4]

    Fang and A

    Z. Fang and A. V . Savkin. Strategies for optimized uav surveillance in various tasks and scenarios: A review. Drones, 8(5):193, 2024

    work page 2024

  5. [5]

    Toscano, C

    F. Toscano, C. Fiorentino, N. Capece, U. Erra, D. Travascia, A. Scopa, M. Drosos, and P. D’Antonio. Unmanned aerial vehicle for precision agriculture: A review. IEEE access , 2024

    work page 2024

  6. [6]

    Gautam, P

    A. Gautam, P. Sujit, and S. Saripalli. A survey of autonomous landing techniques for uavs. In 2014 International Conference on Unmanned Aircraft Systems (ICUAS) , pages 1210–1218,

    work page 2014

  7. [7]

    M. S. Alam and J. Oluoch. A survey of safe landing zone detection techniques for autonomous unmanned aerial vehicles (uavs). Expert Syst. Appl., 179:115091, 2021. URL https://doi. org/10.1016/J.ESWA.2021.115091

    work page arXiv 2021

  8. [8]

    L. Xin, Z. Tang, W. Gai, and H. Liu. Vision-based autonomous landing for the uav: A review. Aerospace, 2022. URL https://api.semanticscholar.org/CorpusId:253134399

    work page 2022

Show all 39 references
  1. [9]

    B. Cai, F. Lu, and L. Zhou. Energy-aware routing algorithm for mobile ground-to-air charging. arXiv preprint arXiv:2310.07729, 2023. 9

    2023 arXiv

  2. [10]

    Bin Junaid, A

    A. Bin Junaid, A. Konoiko, Y . Zweiri, M. N. Sahinkaya, and L. Seneviratne. Autonomous wireless self-charging for multi-rotor unmanned aerial vehicles. energies, 10(6):803, 2017

    2017

  3. [11]

    Alsamhi, A

    S. Alsamhi, A. Shvetsov, S. Kumar, J. Hassan, M. A. Alhartomi, S. Shvetsova, R. Sa- hal, and A. Hawbani. Computing in the sky: A survey on intelligent ubiquitous com- puting for uav-assisted 6g networks and industry 4.0/5.0. Drones, 2022. URL https: //api.semanticscholar.org/C...

    2022

  4. [12]

    S. Yu, Q. Shi, X. Gong, X. Chen, and X. Wang. Ec-sagins: Edge-computing-enhanced space- air-ground-integrated networks for internet of vehicles. IEEE Internet of Things Journal , 9: 5742–5754, 2021. URL https://api.semanticscholar.org/CorpusId:231627644

    2021

  5. [13]

    Hanover, P

    D. Hanover, P. Foehn, S. Sun, E. Kaufmann, and D. Scaramuzza. Performance, preci- sion, and payloads: Adaptive nonlinear mpc for quadrotors. IEEE Robotics and Automation Letters, 7:690–697, 2021. URL http://ieeexplore.ieee.org/stamp/stamp.jsp?tp= &arnumber=9632352

    2021

  6. [14]

    T. Baca, P. Stepan, V . Spurny, D. Hert, R. Penicka, M. Saska, J. Thomas, G. Loianno, and V . Kumar. Autonomous landing on a moving vehicle with an unmanned aerial vehicle.Journal of Field Robotics, 36(5):874–891, 2019. doi:https://doi.org/10.1002/rob.21858. URL https: //onlin...

    2019 doi

  7. [15]

    S. Liu, M. Watterson, K. Mohta, K. Sun, S. Bhattacharya, C. J. Taylor, and V . Kumar. Plan- ning dynamically feasible trajectories for quadrotors using safe flight corridors in 3-d complex environments. IEEE Robotics and Automation Letters (RA-L) , pages 1688–1695, 2017

    2017

  8. [16]

    M. K. Mittal, A. Valada, and W. Burgard. Vision-based autonomous landing in catastrophe- struck environments. ArXiv, abs/1809.05700, 2018. URL https://arxiv.org/pdf/1809. 05700.pdf

    2018 arXiv

  9. [17]

    D. Wang, J. Wang, S. He, J. Huang, B. Zhang, Y . Mao, G. Huang, C. Xu, and F. Gao. Multi-fov- constrained trajectory planning for multirotor safe landing. In 2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) , pages 5356–5363, 2024. doi:10.1109/ I...

    2024

  10. [18]

    P. Li, Z. An, S. Abrar, and L. Zhou. Large language models for multi-robot systems: A survey. arXiv preprint arXiv:2502.03814, 2025

    2025 arXiv

  11. [19]

    M. Yue. A survey of large language model agents for question answering. ArXiv, abs/2503.19213, 2025. URL https://api.semanticscholar.org/CorpusId: 277313791

    2025 arXiv

  12. [20]

    Radford, J

    A. Radford, J. W. Kim, C. Hallacy, A. Ramesh, G. Goh, S. Agarwal, G. Sastry, A. Askell, P. Mishkin, J. Clark, et al. Learning transferable visual models from natural language supervi- sion. In International conference on machine learning , pages 8748–8763. PmLR, 2021

    2021

  13. [21]

    J. Li, D. Li, C. Xiong, and S. Hoi. Blip: Bootstrapping language-image pre-training for uni- fied vision-language understanding and generation. In International conference on machine learning, pages 12888–12900. PMLR, 2022

    2022

  14. [22]

    J. Li, D. Li, S. Savarese, and S. Hoi. Blip-2: Bootstrapping language-image pre-training with frozen image encoders and large language models. In International conference on machine learning, pages 19730–19742. PMLR, 2023

    2023

  15. [23]

    C. Sun, Y . Li, D. Wu, and B. Boulet. Onioneval: An unified evaluation of fact-conflicting hallucination for small-large language models. ArXiv, abs/2501.12975, 2025. URL https: //arxiv.org/pdf/2501.12975.pdf. 10

    2025 arXiv

  16. [24]

    Lewis, E

    P. Lewis, E. Perez, A. Piktus, F. Petroni, V . Karpukhin, N. Goyal, H. Kuttler, M. Lewis, W. tau Yih, T. Rockt¨aschel, S. Riedel, and D. Kiela. Retrieval-augmented generation for knowledge- intensive nlp tasks. ArXiv, abs/2005.11401, 2020. URL https://arxiv.org/pdf/2005. 11401.pdf

    2005 arXiv

  17. [25]

    Gupta, R

    S. Gupta, R. Ranjan, and S. N. Singh. A comprehensive survey of retrieval-augmented gener- ation (rag): Evolution, current landscape and future directions. ArXiv, abs/2410.12837, 2024. URL https://api.semanticscholar.org/CorpusId:273403982

    2024 arXiv

  18. [26]

    O. Ram, Y . Levine, I. Dalmedigos, D. Muhlgay, A. Shashua, K. Leyton-Brown, and Y . Shoham. In-context retrieval-augmented language models. Transactions of the Association for Com- putational Linguistics , 11:1316–1331, 2023. URL https://doi.org/10.1162/tacl_a_ 00605

    2023 doi

  19. [27]

    Bangura and R

    M. Bangura and R. Mahony. Real-time model predictive control for quadrotors. IF AC Pro- ceedings V olumes, 47(3):11773–11780, 2014. ISSN 1474-6670. doi:https://doi.org/10.3182/ 20140824-6-ZA-1003.00203. 19th IFAC World Congress

    2014

  20. [28]

    G. Niu, Q. Yang, Y . Gao, and M.-O. Pun. Vision-based autonomous landing for unmanned aerial and ground vehicles cooperative systems. IEEE robotics and automation letters , 7(3): 6234–6241, 2021

    2021

  21. [29]

    Zhang et al

    B. Zhang et al. Coni-mpc: cooperative non-inertial frame based model predictive control. ieee robot autom lett 8 (12): 8082–8089, 2023

    2023

  22. [30]

    H. Wang, Q. Wang, F. Gao, and S. Shen. Contact-aware motion planning among movable objects. arXiv preprint arXiv:2502.03317, 2025

    2025 arXiv

  23. [31]

    Sinha, A

    R. Sinha, A. Elhafsi, C. Agia, M. Foutter, E. Schmerling, and M. Pavone. Real-time anomaly detection and reactive planning with large language models.arXiv preprint arXiv:2407.08735, 2024

    2024 arXiv

  24. [32]

    J. Liu. LlamaIndex, 11 2022. URL https://github.com/jerryjliu/llama_index

    2022

  25. [33]

    S. Xiao, Z. Liu, P. Zhang, and N. Muennighoff. C-pack: Packaged resources to advance general chinese embedding, 2023

    2023

  26. [34]

    D. Han, M. Han, and Unsloth team. Unsloth, 2023. URL http://github.com/unslothai/ unsloth

    2023

  27. [35]

    Y . Wu, Z. Ding, C. Xu, and F. Gao. External forces resilient safe motion planning for quadro- tor. IEEE Robotics and Automation Letters , 6(4):8506–8513, 2021. doi:10.1109/LRA.2021. 3110316

    2021 doi

  28. [36]

    T. Lee, M. Leok, and N. H. McClamroch. Geometric tracking control of a quadrotor uav on se(3). In 49th IEEE Conference on Decision and Control (CDC) , pages 5420–5425, 2010. doi:10.1109/CDC.2010.5717652

    2010

  29. [37]

    Furrer, M

    F. Furrer, M. Burri, M. Achtelik, and R. Siegwart. Robot Operating System (ROS): The Com- plete Reference (V olume 1) , chapter RotorS—A Modular Gazebo MA V Simulator Frame- work, pages 595–625. Springer International Publishing, Cham, 2016. ISBN 978-3- 319-26054-9. doi:10.100...

    2016 doi

  30. [38]

    is_dynamic

    L. Lian, Y . Ding, Y . Ge, S. Liu, H. Mao, B. Li, M. Pavone, M.-Y . Liu, T. Darrell, A. Yala, and Y . Cui. Describe anything: Detailed localized image and video captioning.arXiv preprint arXiv:2504.16072, 2025. 11 A Appendix This appendix provides additional details on the ove...

    2025 arXiv

  31. [2014]

    doi:10.1109/ICUAS.2014.6842377

    2014

Pith tools

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