Embodied AI-Enhanced Vehicular Networks: An Integrated Large Language Models and Reinforcement Learning Method

Changyuan Zhao; Dong In Kim; Dusit Niyato; Hongyang Du; Jiacheng Wang; Ruichen Zhang; Suttinee Sawadsitang; Xuemin Shen

REVIEW 5 major objections 5 minor 3 cited by

Embodied AI-Enhanced Vehicular Networks: An Integrated Large Language Models and Reinforcement Learning Method

T0 review · 5 major / 5 minor · reviewed 2026-08-10 · deepseek-v4-flash

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

Pith's one-line read An LLAVA-plus-GAE-PPO framework shrinks vehicular image transmission by over 90% while improving QoE by up to 36%.

desk verdict The paper's QoE metric is inverted—it rewards lower semantic similarity—so the headline 36% gain measures degradation, not quality. read the letter →

arxiv 2501.01141 v1 pith:T5SAK3KZ submitted 2025-01-02 cs.NI

Ruichen Zhang , Changyuan Zhao , Hongyang Du , Dusit Niyato , Jiacheng Wang , Suttinee Sawadsitang , Xuemin Shen , Dong In Kim This is my paper · ORCID

classification cs.NI

keywords embodiedAIvehicularnetworkssemanticcommunicationLLAVAdeepreinforcementlearningqualityofexperienceWeber-FechnerlawGAE-PPO

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

The paper proposes an embodied-AI framework for vehicular networks in which each vehicle uses the LLAVA vision-language model to turn camera images into short structured text messages, and a generalized-advantage-estimation PPO agent decides transmission power, spectrum sharing, and semantic symbol length. The central claim is that this combination keeps decision-relevant content while shrinking transmitted data by more than 90%, and that the learned transmission policy improves quality of experience by up to 36% over DDPG while converging in fewer steps than pure PPO. The optimization target is a Weber-Fechner-law-inspired QoE metric that balances semantic similarity (BERT cosine) against bandwidth use. The work's significance is showing that LLM-based semantic compression and DRL-based resource control can be jointly trained in the same loop, which would matter if 6G vehicular networks must carry massive sensor data through limited spectrum.

What carries the argument

The load-bearing object is the Weber-Fechner-inspired QoE objective, $\text{QoE} = \sum_{w=1}^{W}\sum_{q=1}^{Q} \Theta_a(1-\xi_q[w])\,\Theta_b\!\left(b_q[w]P_q^{\mathrm{V2V}}[w]\right)$, with $\Theta_a(x)=x$ and $\Theta_b(x)=1/(1+e^{-x})$, where $\xi$ is the BERT cosine similarity between transmitted and recovered messages. This scalar metric converts semantic fidelity and transmit power into a single reward, making the non-convex resource allocation problem (14) amenable to reinforcement learning. Around it, LLAVA acts as the semantic encoder that compresses images to text, and GAE-PPO is the policy optimizer that maximizes cumulative QoE under SINR and similarity thresholds.

What would settle it

Replace the BERT cosine similarity in Eq. (12) with a task-level measure: have the receiver run an obstacle-detection or lane-change decision from the recovered text and compare success rates against decisions made from the original image. If a panel or automated benchmark shows no better-than-baseline task accuracy while Eq. (13) reports high QoE, the metric and the claimed gains are falsified.

Watch

Extended reading notes

Core claim

On the paper's own terms, the discovery is that LLAVA-1.5-7b-hf compresses a 614 KB camera frame into a 12.1 KB text message while preserving the facts needed for driving decisions (vehicle counts, parking occupancy, lane incidents), and that a GAE-PPO agent optimizing the QoE objective reaches higher accumulated return and faster convergence than pure PPO or DDPG baselines. The quantitative claims are a QoE gain up to 36% over DDPG at eight vehicles, a 61.4% QoE improvement when scaling from four to eight vehicles, and convergence-step reductions of 10, 23, and 54 for three of four simulated vehicles compared with pure PPO. The authors present these as evidence that semantic communication plus reinforcement learning can jointly optimize bandwidth use and decision accuracy in dynamic vehicular environments.

Load-bearing premise

The entire performance comparison stands or falls on the assumption that the QoE formula in Eq. (13) captures real driving-related quality of experience; the paper asserts this via the Weber-Fechner law but does not derive the formula from it or validate it against human perception.

Editorial extensions

If this is right

If the framework's results hold, V2I links can transmit semantic text rather than raw video, cutting spectrum load by an order of magnitude while retaining the content that matters for driving decisions.
GAE-PPO's faster, more stable convergence than pure PPO would make DRL-based transmission control feasible in real time, where training episodes are cheap but decision latency matters.
The reported 61.4% QoE gain from four to eight vehicles suggests the method scales as network load grows, which is exactly the regime where spectrum is scarce.
Adapting semantic symbol or sentence length to SINR would give operators a tunable knob between fidelity and bandwidth, as Figure 10 indicates.

Reading between the lines

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

The QoE metric itself is the least-tested link: a subjective or task-based validation study would likely be needed before the exact 36% number can be trusted.
The 90% compression claim presumes text retains all decision-critical information; a natural extension is to compare end-to-end driving-task accuracy under text-only versus raw-image transmission.
The same LLM-plus-DRL architecture could be transferred to roadside units, drone relays, or other embodied agents, since the framework does not depend on vehicle-specific physics.
The attention-map analysis hints at a further bandwidth saving: transmitting only the highlighted image regions, rather than full frames, could push compression beyond 90%.

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.

Desk Editor's Note

read the letter

First, the thing to know before you spend time on this: the central QoE metric in Eq. (13) is backwards. With Θ_a(x)=x and Θ_b(x)=1/(1+e^{-x}), the term Θ_a(1−ξ) equals 1−ξ, so QoE increases as semantic similarity ξ decreases. Perfect reconstruction scores zero; a meaninglessly different message scores the maximum. The penalty in Eq. (28) only keeps ξ above the 0.3 threshold, so the RL agent learns to deliver the worst allowed quality, and the reported 36% QoE improvement over DDPG is an improvement on a metric that rewards destroying semantic content. That is an internal correctness problem, not a missing external validation. What is genuinely new: the integration of LLAVA for image-to-text semantic extraction with GAE-PPO for transmission control in a vehicular setting is, as far as I can tell, not in the prior work. The benchmark of five vision-language models on parking-lot queries is a small but real empirical contribution, and the raw data-size reduction (614 KB image to 12.1 KB text) is a concrete, plausibly reproducible result. The convergence-speed gain of GAE-PPO over pure PPO is standard but cleanly demonstrated. The soft spots beyond the inverted metric: Eq. (18) uses an L2 norm on text messages, which is not a vector space—that loss is undefined. Section 5.1.1 calls ξ_th the SSIM threshold even though ξ was defined as BERT cosine similarity, and Fig. 10 plots SSIM as if it were the same quantity. No code or data are released, and the simulation figures have no error bars, so the magnitude claims (47%, 61.4%) rest entirely on the authors' own runs. I would not send this to peer review in its present form. The central performance measure is self-contradictory, and even a fix to the sign would require re-running the entire experiment. The framework idea has some merit, and a cleaned version with a correctly defined QoE could become a modest extension worth a second look. But as it stands, the headline conclusion is not supported.

Referee Report

5 major / 5 minor

Summary. The manuscript proposes an embodied-AI framework for vehicular networks in which LLAVA converts camera images into compact textual semantic messages and a GAE-PPO agent selects V2V power, subband, and semantic-symbol allocations. The authors formulate an optimization problem over a QoE metric that they state is inspired by the Weber-Fechner law, and they report that the proposed transmission strategy improves QoE by up to 36% over DDPG, converges faster than pure PPO, and reduces transmitted data size by more than 90% via LLAVA. The paper also compares LLAVA-1.5-7b against four vision-language baselines on parameter count, inference time, and visual-question-answering accuracy, and analyzes convergence, QoE scaling, and the effect of semantic sentence length.

Significance. If validated, the framework would address a real problem: reducing raw image transmission in bandwidth-limited vehicular networks while retaining task-relevant semantic content. The LLAVA compression figures and the systematic comparison of vision-LLM baselines are useful empirical data points, and using GAE to stabilize PPO in this setting is a sensible design choice. However, the central quantitative claims rest on Eq. (13), which is both unvalidated against any external QoE or perceptual standard and internally inconsistent, since it rewards decreasing semantic similarity. The absence of error bars and the undefined fine-tuning loss further prevent the reported improvements from being accepted as stated. The paper therefore does not currently establish its main claims.

major comments (5)

[Sec. 3.3, Eq. (13); Sec. 4.2.2, Eq. (27)] The QoE and reward are defined as the double sum of Theta_a(1 - xi_q[w]) Theta_b(b_q[w] P_q^V2V[w]), with Theta_a(x) = x and Theta_b(x) = 1 / (1 + e^{-x}) set in Sec. 5.1.1. Since xi = 1 is perfect similarity and xi = 0 is no similarity (Eq. (12)), this quantity is strictly increasing in (1 - xi) for any active link with positive power: perfect reconstruction contributes zero, and completely dissimilar content contributes the maximum sigmoid(power). The DRL agent is therefore rewarded for driving xi down to the constraint boundary xi_th = 0.3, and the headline 36% QoE gain compares policies under an objective that penalizes semantic fidelity. Moreover, because the sum includes links with b_q[w] = 0, inactive links contribute Theta_b(0) = 0.5 times (1 - xi), so the metric is nonzero even when nothing is transmitted. The metric should be a non-decreasing function of xi, or the paper must provide a derivation that justifies the inversion; as written, the central claim is internally inconsistent.
[Sec. 3.3; Sec. 5.1.1] The paper states that Eq. (13) is inspired by the Weber-Fechner law, but it never derives Eq. (13) from that law, and the formula is not in the logarithmic Weber-Fechner form. The shaping functions Theta_a and Theta_b are declared in Sec. 5.1.1 without justification or sensitivity analysis, and no comparison against human-perception data or a standard QoE dataset is provided. Consequently, the reported 36% and 61.4% improvements are improvements on an unvalidated, author-defined objective; the citation to psychological QoE work does not establish validity.
[Sec. 4.1.2, Eq. (18)] The LLAVA fine-tuning loss is written as min over theta of sum_i ||M_i - M_hat_i||^2, but M_i and M_hat_i are natural-language semantic messages (text descriptions) and no embedding, tokenizer, or vector representation is specified in the paper. An L2 norm on raw text is therefore undefined, making the fine-tuning procedure non-reproducible. The authors should specify the representation (for example, BERT embeddings) and the exact loss used in the implementation.
[Sec. 5, Figs. 7-10] The simulations are reported as single trajectories with no error bars, confidence intervals, or seeds, so the claimed gains (61% accumulated-return gain, 36% QoE gain, up to 47% fewer convergence steps, and 61.4% scaling improvement) cannot be distinguished from random variation. In addition, the abstract and contributions claim improved decision accuracy, but no experiment measures a downstream decision-making task; Fig. 6 only reports visual-question-answering output correctness. The authors should define and measure decision accuracy, or remove that claim.
[Sec. 5.1.1, Fig. 10] The threshold xi_th is described as an SSIM threshold and Fig. 10 plots SSIM versus SINR and sentence length, whereas Eq. (12) defines xi as BERT cosine similarity between text messages. SSIM is an image-quality metric and cannot be applied directly to the text-based semantic similarity used in Problem (14); the relationship shown in Fig. 10 therefore does not validate constraint (14f). The metric terminology should be unified and the actual quantity plotted should match the quantity used in the optimization.

minor comments (5)

[Sec. 4.1.2; Algorithm 1] The phrase 'To adopt LLAVA' should be 'To adapt LLAVA', and the Algorithm 1 title contains a stray space in 'V ehicular'.
[Sec. 5.1.1; Table 1] The V2V power is described as continuous in [-100, 23] dBm in one place and as the discrete set {-100, 5, 15, 23} dBm in Table 1; please reconcile whether power control is continuous or discrete.
[Fig. 7; Fig. 8] The y-axes are labelled 'Accumulated return' and 'Achievable QoE' with no units or confidence bands; report the mean and standard deviation over multiple random seeds.
[Sec. 4.2.1, Eq. (24)] The text says the target value uses n-step returns, but the displayed formula V_tar(s_t) = r(s_t,a_t) + gamma V_theta_C(s_{t+1}) is a one-step bootstrap; align the description with the implementation.
[Sec. 5.2, Fig. 9] The text reports 'up to 47%' fewer convergence steps, while Fig. 9 lists absolute step reductions of 10, 23, and 54; report percentages together with the baseline totals so the comparison is unambiguous.

Circularity Check

1 steps flagged · score 4.0 of 10

Headline QoE gains are measured on the paper's own reward function (Eq. 27 = Eq. 13), and that metric is decreasing in the semantic similarity it claims to preserve.

self definitional [Section 3.3 Eq. (13); Section 4.2.2 Eq. (27); Section 5.1.1; Section 5.2 Fig. 8]
"The reward at time step t is given by r_t = sum_{w=1}^{W} sum_{q=1}^{Q} Theta_a(1-xi_q[w](t)) Theta_b(b_q[w]P_q^{V2V}[w](t)). ... Theta_a(x) is set to x and Theta_b is expressed as Theta_b(x)=1/(1+e^{-x})."

Eq. (27) is exactly Eq. (13)'s QoE expression with a time index, so the quantity reported in Fig. 8 as 'Achievable QoE' is the same function the GAE-PPO agent is trained to maximize. The headline 'improves QoE by up to 36%' therefore compares policies on the paper's own reward, not on an externally validated quality measure; it is a self-referential report of the optimized objective. The self-definition is aggravated by the chosen Theta_a(x)=x: each term is proportional to (1-xi), so QoE decreases as the BERT semantic similarity xi increases, pushing the policy toward the xi_th=0.3 constraint boundary rather than toward higher reconstruction fidelity. The claimed QoE result thus reduces by construction to the reward design in Eq. (27).

full rationale

The paper's central QoE claim is partially circular: the optimized reward (Eq. 27) is identical to the reported performance metric (Eq. 13), and because Theta_a(x)=x, that metric is decreasing in the semantic similarity xi that Section 3.3 equates with reconstruction quality. This makes the headline QoE improvement a report of the training objective rather than an independent measure of experience quality, and the Weber-Fechner-law grounding is asserted rather than derived. Some independent empirical content does exist: the LLAVA data-size reduction (>90%), the vision-LLM inference-time comparison (Fig. 6), and the convergence-step counts (Fig. 9) are not defined circularly. I found no load-bearing self-citation chain; the many self-citations are contextual. The correctness risk from the inverted QoE objective and the SSIM/xi terminology conflation in Section 5.1.1 and Fig. 10 should be assessed separately, but they do not change the circularity verdict beyond the self-referential metric.

Assumptions & free parameters 8 free parameters · 5 assumptions · 0 invented entities

The central claim rests on an author-defined QoE metric, an unverified semantic-preservation assumption, and an undefined text loss. The paper introduces no new physical entities, but it leans on several hand-picked thresholds and functional forms.

free parameters (8)

QoE shaping functions Θ_a(x)=x, Θ_b(x)=1/(1+e^{-x}) = x and sigmoid(x)
Chosen ad hoc in Eq. (13) and Section 5.1.1; not derived from the Weber-Fechner law or calibrated to user experience.
semantic similarity threshold ξ_th = 0.3
Set in Section 5.1.1 as a constraint threshold; no evidence it reflects acceptable communication quality.
SINR threshold γ_th = 10 dB
Set in Section 5.1.1; standard but hand-picked for the simulation.
penalty coefficients λ1 and λ2 = 1 and 1
Set in Section 5.1.2; the reward function's constraint handling depends on these values.
PPO clipping parameter ε = 0.5
Table 1; unusually large compared to the common 0.2, with no sensitivity analysis.
semantic symbol length bounds u_q ∈ {1,...,u_qmax} = u_qmax not specified
Constraint (14e) and the action space rely on an unspecified upper bound.
V2V power levels = [-100, 5, 15, 23] dBm
Table 1; discrete power levels chosen from prior work that constrain the action space.
learning rates = 1e-4 to 1e-8
Table 1; no sensitivity analysis.

assumptions (5)

ad hoc to paper Eq. (13) with Θ_a(x)=x and Θ_b(x)=1/(1+e^{-x}) is a valid QoE metric.
The paper invokes the Weber-Fechner law but does not derive Eq. (13) from it; the metric is not validated against human perception.
domain assumption LLAVA's text output preserves sufficient information for vehicular decision-making.
Only BERT cosine similarity and attention maps are shown; no downstream task accuracy is measured.
ad hoc to paper Text messages can be compared with a squared L2 norm in fine-tuning loss (Eq. 18).
The messages are text sequences; the squared norm in Eq. (18) is undefined for text. This is a technical error.
domain assumption GAE-PPO converges reliably for the non-convex combinatorial problem (14).
PPO and GAE convergence is taken from the RL literature; no proof is given for this problem class.
domain assumption Channel models and SINR equations (5)-(6) describe the target vehicular environment.
Inherited from prior work [13,31,32] and not validated against field data.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Embodied AI-Enhanced Vehicular Networks: An Integrated Large Language Models and Reinforcement Learning Method." pith.science (2026). https://pith.science/paper/T5SAK3KZ

@misc{pith2026250101141,
  author       = {Pith},
  title        = {Pith review of: Embodied AI-Enhanced Vehicular Networks: An Integrated Large Language Models and Reinforcement Learning Method},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/T5SAK3KZ}},
  note         = {Machine review of arXiv:2501.01141}
}

read the original abstract

This paper investigates adaptive transmission strategies in embodied AI-enhanced vehicular networks by integrating large language models (LLMs) for semantic information extraction and deep reinforcement learning (DRL) for decision-making. The proposed framework aims to optimize both data transmission efficiency and decision accuracy by formulating an optimization problem that incorporates the Weber-Fechner law, serving as a metric for balancing bandwidth utilization and quality of experience (QoE). Specifically, we employ the large language and vision assistant (LLAVA) model to extract critical semantic information from raw image data captured by embodied AI agents (i.e., vehicles), reducing transmission data size by approximately more than 90\% while retaining essential content for vehicular communication and decision-making. In the dynamic vehicular environment, we employ a generalized advantage estimation-based proximal policy optimization (GAE-PPO) method to stabilize decision-making under uncertainty. Simulation results show that attention maps from LLAVA highlight the model's focus on relevant image regions, enhancing semantic representation accuracy. Additionally, our proposed transmission strategy improves QoE by up to 36\% compared to DDPG and accelerates convergence by reducing required steps by up to 47\% compared to pure PPO. Further analysis indicates that adapting semantic symbol length provides an effective trade-off between transmission quality and bandwidth, achieving up to a 61.4\% improvement in QoE when scaling from 4 to 8 vehicles.

Figures

Figures reproduced from arXiv: 2501.01141 by the authors.

**Figure 2.** System model illustrates a cellular-based vehicula [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

**Figure 3.** The architecture of the LLAVA model for semantic [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗

Figures from the paper (6 more)

**Figure 4.** Figure 4: The workflow of the GAE-PPO method for optimizing tran [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]

**Figure 5.** Figure 5: The illustration of semantic information extractio [PITH_FULL_IMAGE:figures/full_fig_p011_5.png]

**Figure 6.** Figure 6: Comparison of vision-LLM models in terms of paramete [PITH_FULL_IMAGE:figures/full_fig_p012_6.png]

**Figure 7.** Figure 7: Convergence behavior with different methods. [PITH_FULL_IMAGE:figures/full_fig_p012_7.png]

**Figure 9.** Figure 9: Comparison of GAE-PPO and pure PPO in fulfilling commu [PITH_FULL_IMAGE:figures/full_fig_p013_9.png]

**Figure 10.** Figure 10: The SSIM versus different words sentence length [PITH_FULL_IMAGE:figures/full_fig_p013_10.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 3 Pith papers

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

Efficient and Trustworthy Block Propagation for Blockchain-enabled Mobile Embodied AI Networks: A Graph Resfusion Approach
cs.AI 2025-01 conditional novelty 5.0 of 10

A graph Resfusion model that starts from a feasible route and denoises it toward a shorter route achieves lower block propagation time and higher trust scores than greedy, genetic, GAT, and Difusco baselines in simula...
Toward Edge General Intelligence with Agentic AI and Agentification: Concepts, Technologies, and Future Directions
cs.NI 2025-08 conditional novelty 4.0 of 10

A survey that organizes agentic AI for 6G edge networks into four pillars, compactness, efficiency, knowledge and reasoning, and migration, and illustrates them with prior case studies.
TinyMA-IEI-PPO: Exploration Incentive-Driven Multi-Agent DRL with Self-Adaptive Pruning for Vehicular Embodied AI Agent Twins Migration
cs.MA 2025-04 reject novelty 4.0 of 10

A pruned multi-agent PPO algorithm with exploration incentives is designed to approximate the Stackelberg equilibrium of a vehicular AI twin migration game.

Reference graph

Works this paper leans on

51 extracted references · 41 canonical work pages · cited by 3 Pith papers

[1]

A hybrid deep sensor anomaly detection for autonomous vehicles in 6G-V2X environment,

S. B. Prathiba, G. Raja, S. Anbalagan, et al. , “A hybrid deep sensor anomaly detection for autonomous vehicles in 6G-V2X environment,” IEEE T rans. Network Sci. Eng. , vol. 10, no. 3, pp. 1246–1255, 2023

work page 2023
[2]

Clouds on the road: A software-deﬁned fog computing framework for intelligent re- source management in vehicular Ad-Hoc networks,

A. Nahar, K. K. Mondal, D. Das, et al. , “Clouds on the road: A software-deﬁned fog computing framework for intelligent re- source management in vehicular Ad-Hoc networks,” IEEE T rans. Mob. Comput., vol. 23, no. 12, pp. 12 778–12 792, 2024. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2015 14

work page 2024
[3]

Privacy-preserving and secure distributed data sharing scheme for V ANETs,

L. Wang, H. Zhong, J. Cui, et al. , “Privacy-preserving and secure distributed data sharing scheme for V ANETs,” IEEE T rans. Mob. Comput., 2024

work page 2024
[4]

Fedagl: A communication-efﬁcient federated vehicular network,

S. Liu, Y . Li, P . Guan, et al. , “Fedagl: A communication-efﬁcient federated vehicular network,” IEEE T rans. Intell. Veh., 2024

work page 2024
[5]

A reliable physical layer key generation scheme based on RSS and LSTM network in V ANET,

Z. Wang, Y . Liu, J. Wang, et al. , “A reliable physical layer key generation scheme based on RSS and LSTM network in V ANET,” IEEE Internet Things J.l , vol. 11, no. 1, pp. 692–707, 2023

work page 2023
[6]

Comprehensive survey on machine learning in vehicular network: Technology , applic ations and challenges,

F. Tang, B. Mao, N. Kato, et al. , “Comprehensive survey on machine learning in vehicular network: Technology , applic ations and challenges,” IEEE Commun. Survey T uts. , vol. 23, no. 3, pp. 2027–2057, 2021

work page 2027
[7]

A survey of embodied AI: From simulators to research tasks,

J. Duan, S. Y u, H. L. Tan, et al. , “A survey of embodied AI: From simulators to research tasks,” IEEE T rans. Emerging T op. Comput. Intell., vol. 6, no. 2, pp. 230–244, 2022

work page 2022
[8]

Autonomous vehic les and embedded artiﬁcial intelligence: The challenges of fra ming machine driving decisions,

M. Cunneen, M. Mullins, and F. Murphy , “Autonomous vehic les and embedded artiﬁcial intelligence: The challenges of fra ming machine driving decisions,” Applied Artiﬁcial Intelligence , vol. 33, no. 8, pp. 706–731, 2019

work page 2019

Show all 51 references

[9]

Generative AI agents with large language model for satellite networks via a mixture of exper ts transmission,

R. Zhang, H. Du, Y . Liu, et al. , “Generative AI agents with large language model for satellite networks via a mixture of exper ts transmission,” IEEE J. Sel. Areas Commun. , pp. 1–1, 2024

2024
[10]

Omnidrive: A holistic LLM-agent framework for autonomous driving with 3d perception, reaso ning and planning,

S. Wang, Z. Y u, X. Jiang, et al. , “Omnidrive: A holistic LLM-agent framework for autonomous driving with 3d perception, reaso ning and planning,” arXiv preprint arXiv:2405.01533 , 2024

2024 arXiv
[11]

Energy efﬁciency maximization in RIS-assisted SWIPT networks with RSMA: A PPO-based ap- proach,

R. Zhang, K. Xiong, Y . Lu, et al. , “Energy efﬁciency maximization in RIS-assisted SWIPT networks with RSMA: A PPO-based ap- proach,” IEEE J. Sel. Areas Commun. , vol. 41, no. 5, pp. 1413–1430, 2023

2023
[12]

Generative AI-enabled vehicular networks: Fundamentals, framework, and case study ,

R. Zhang, K. Xiong, H. Du, et al., “Generative AI-enabled vehicular networks: Fundamentals, framework, and case study ,” IEEE Netw., vol. 38, no. 4, pp. 259–267, 2024

2024
[13]

Semantic-aware spectrum sharing in internet of vehicles based on deep reinforcement learning,

Z. Shao, Q. Wu, P . Fan, et al., “Semantic-aware spectrum sharing in internet of vehicles based on deep reinforcement learning, ” arXiv preprint arXiv:2406.07213, 2024

2024 arXiv
[14]

Interactive ai with retrieval- augmented generation for next generation networking,

R. Zhang, H. Du, Y . Liu, et al. , “Interactive ai with retrieval- augmented generation for next generation networking,” IEEE Netw., vol. 38, no. 6, pp. 414–424, 2024

2024
[15]

Spectrum sharing in vehicul ar networks based on multi-agent reinforcement learning,

L. Liang, H. Y e, and G. Y . Li, “Spectrum sharing in vehicul ar networks based on multi-agent reinforcement learning,” IEEE J. Sel. Areas Commun. , vol. 37, no. 10, pp. 2282–2292, 2019

2019
[16]

Graph-based resource sharing in vehicular communication,

L. Liang, S. Xie, G. Y . Li, et al. , “Graph-based resource sharing in vehicular communication,” IEEE T rans. Wireless Commun., vol. 17, no. 7, pp. 4579–4592, 2018

2018
[17]

Cooperative deep reinforcement learning enabled power allocation for packet duplication U RLLC in multi-connectivity vehicular networks,

J. Xue, K. Y u, T. Zhang, et al. , “Cooperative deep reinforcement learning enabled power allocation for packet duplication U RLLC in multi-connectivity vehicular networks,” IEEE T rans. Mob. Com- put., 2024

2024
[18]

Distributed deep reinforce- ment learning based gradient quantization for federated le arning enabled vehicle edge computing,

C. Zhang, W. Zhang, Q. Wu, et al. , “Distributed deep reinforce- ment learning based gradient quantization for federated le arning enabled vehicle edge computing,” IEEE Internet Things J.l , pp. 1–1, 2024

2024
[19]

Secure transmission scheme based on joint radar and communication in mobile vehicular networ ks,

Y . Y ao, F. Shu, Z. Li, et al. , “Secure transmission scheme based on joint radar and communication in mobile vehicular networ ks,” IEEE T rans. Intell. T ransp. Syst. , vol. 24, no. 9, pp. 10 027–10 037, 2023

2023
[20]

Habitat: A platform for embodied ai research,

M. Savva, A. Kadian, O. Maksymets, et al., “Habitat: A platform for embodied ai research,” in Proc. IEEE ICCV , 2019, pp. 9339–9347

2019
[21]

LLM-planner: Few- shot grounded planning for embodied agents with large langu age models,

C. H. Song, J. Wu, C. Washington, et al. , “LLM-planner: Few- shot grounded planning for embodied agents with large langu age models,” in Proc. IEEE ICCV , 2023, pp. 2998–3009

2023
[22]

Towards efﬁcient LLM grounding for embodied multi-agent collaboration,

Y . Zhang, S. Y ang, C. Bai, et al. , “Towards efﬁcient LLM grounding for embodied multi-agent collaboration,” arXiv preprint arXiv:2405.14314, 2024

2024
[23]

Ros-LLM: A ROS framework for embodied AI with task feedback and structured reasoning ,

C. E. Mower, Y . Wan, H. Y u, et al. , “Ros-LLM: A ROS framework for embodied AI with task feedback and structured reasoning ,” arXiv preprint arXiv:2406.19741 , 2024

2024 arXiv
[24]

BadRobot: Jailbreaking llm-based embodied AI in the physical world,

H. Zhang, C. Zhu, X. Wang, et al. , “BadRobot: Jailbreaking llm-based embodied AI in the physical world,” arXiv preprint arXiv:2407.20242, 2024

2024 arXiv
[25]

Contrastive learning-based se- mantic communications,

S. Tang, Q. Y ang, L. Fan, et al. , “Contrastive learning-based se- mantic communications,” IEEE T rans. Commun. , vol. 72, no. 10, pp. 6328–6343, 2024

2024
[26]

PEAC: Unsupervised pre- training for cross-embodiment reinforcement learning,

C. Ying, Z. Hao, X. Zhou, et al. , “PEAC: Unsupervised pre- training for cross-embodiment reinforcement learning,” arXiv preprint arXiv:2405.14073, 2024

2024 arXiv
[27]

Human-in-the-loop embodied intelligence with interactive simulation environment for surgical robot learning,

Y . Long, W. Wei, T. Huang, et al. , “Human-in-the-loop embodied intelligence with interactive simulation environment for surgical robot learning,” IEEE Robot. Autom. Lett. , vol. 8, no. 8, pp. 4441– 4448, 2023

2023
[28]

Simple embodied language learning as a byproduct of meta-reinforcement learning,

E. Z. Liu, S. Suri, T. Mu, et al., “Simple embodied language learning as a byproduct of meta-reinforcement learning,” in Proc. ICML . PMLR, 2023, pp. 21 997–22 008

2023
[29]

True knowledge comes from prac- tice: Aligning llms with embodied environments via reinforcement learning,

W. Tan, W. Zhang, S. Liu, et al., “True knowledge comes from prac- tice: Aligning llms with embodied environments via reinforcement learning,” arXiv preprint arXiv:2401.14151 , 2024

2024 arXiv
[30]

Lte rssi based vehicular lo- calization system in long tunnel environment,

B. Shin, J. H. Lee, C. Kim, et al. , “Lte rssi based vehicular lo- calization system in long tunnel environment,” IEEE T rans. Ind. Informat., vol. 19, no. 11, pp. 11 102–11 114, 2023

2023
[31]

Joint spectrum sharing a nd V2V/V2I task ofﬂoading for vehicular edge computing networ ks based on coalition formation game,

M. Huang, Z. Shen, and G. Zhang, “Joint spectrum sharing a nd V2V/V2I task ofﬂoading for vehicular edge computing networ ks based on coalition formation game,” IEEE T rans. Intell. T ransp. Syst., 2024

2024
[32]

Online MEC ofﬂoading for V2V networks,

F. Liu, J. Chen, Q. Zhang, et al. , “Online MEC ofﬂoading for V2V networks,” IEEE T rans. Mob. Comput. , vol. 22, no. 10, pp. 6097– 6109, 2023

2023
[33]

Deep learning enabled semantic communication systems,

H. Xie, Z. Qin, G. Y . Li, et al. , “Deep learning enabled semantic communication systems,” IEEE T rans. Signal Process. , vol. 69, pp. 2663–2675, 2021

2021
[34]

Biasing federated learning with a new adversarial graph attention network,

K. Li, J. Zheng, W. Ni, et al., “Biasing federated learning with a new adversarial graph attention network,” IEEE T rans. Mob. Comput. , pp. 1–15, 2024

2024
[35]

A wireless AI-generated content (AIGC) provisioning framework empowered by semant ic communication,

R. Cheng, Y . Sun, D. Niyato, et al. , “A wireless AI-generated content (AIGC) provisioning framework empowered by semant ic communication,” IEEE T rans. Mob. Comput., pp. 1–14, 2024

2024
[36]

Psyqoe: Improving quality-of- experience assessment with psychological effects in video stream- ing,

D. Sheng, Q. Qi, J. Wang, et al. , “Psyqoe: Improving quality-of- experience assessment with psychological effects in video stream- ing,” IEEE T rans. Serv. Comput., pp. 1–14, 2024

2024
[37]

Ra-clip: Retrieval augmented contrastive language-image pre-training,

C.-W. Xie, S. Sun, X. Xiong, et al. , “Ra-clip: Retrieval augmented contrastive language-image pre-training,” in Proc. IEEE CVPR , 2023, pp. 19 265–19 274

2023
[38]

Visual instruction tuning,

H. Liu, C. Li, Q. Wu, et al. , “Visual instruction tuning,” NIPS, vol. 36, 2024

2024
[39]

Trajectory and communication design for cache-enabled UA Vs in cellular networks: A deep reinforcem ent learning approach,

J. Ji, K. Zhu, and L. Cai, “Trajectory and communication design for cache-enabled UA Vs in cellular networks: A deep reinforcem ent learning approach,” IEEE T rans. Mob. Comput., vol. 22, no. 10, pp. 6190–6204, 2022

2022
[40]

Constrained update projection approach to safe policy optimization,

L. Y ang, J. Ji, J. Dai, et al., “Constrained update projection approach to safe policy optimization,” NIPS, vol. 35, pp. 9111–9124, 2022

2022
[41]

Dynamic ofﬂoading for multiuser muti- CAP mec networks: A deep reinforcement learning approach,

C. Li, J. Xia, F. Liu, et al. , “Dynamic ofﬂoading for multiuser muti- CAP mec networks: A deep reinforcement learning approach,” IEEE T rans. Veh. T echnol., vol. 70, no. 3, pp. 2922–2927, 2021

2021
[42]

Attention is all you need,

A. Vaswani, “Attention is all you need,” NIPS, 2017

2017
[43]

Distributed energy-efﬁcient multi- UA V navigation for long-term communication coverage by dee p reinforcement learning,

C. H. Liu, X. Ma, X. Gao, et al., “Distributed energy-efﬁcient multi- UA V navigation for long-term communication coverage by dee p reinforcement learning,” IEEE T rans. Mob. Comput., vol. 19, no. 6, pp. 1274–1285, 2020

2020
[44]

Europarl: A parallel corpus for statistical machine translation,

P . Koehn, “Europarl: A parallel corpus for statistical machine translation,” in Proceedings of machine translation summit x: papers , 2005, pp. 79–86

2005
[45]

Improved baselines with visual instruc- tion tuning,

H. Liu, C. Li, Y . Li, et al. , “Improved baselines with visual instruc- tion tuning,” in Proc. IEEE CVPR , 2024, pp. 26 296–26 306

2024
[46]

Deep reinforcement learning based aoi minimization for NOMA-enabled integrated satellite-terr estrial networks,

X. He, Y . Y ang, J. Lee,et al., “Deep reinforcement learning based aoi minimization for NOMA-enabled integrated satellite-terr estrial networks,” IEEE T rans. Veh. T echnol., pp. 1–6, 2024

2024
[47]

Joint device scheduling and bandwidth allocation for federated learning over wirel ess networks,

T. Zhang, K.-Y . Lam, J. Zhao, et al. , “Joint device scheduling and bandwidth allocation for federated learning over wirel ess networks,” IEEE T rans. Wireless Commun., pp. 1–1, 2023

2023
[48]

Video-llava: Learning united visual representation by alignment before projection,

B. Lin, Y . Y e, B. Zhu, et al. , “Video-llava: Learning united visual representation by alignment before projection,” arXiv preprint arXiv:2311.10122, 2023

2023 arXiv
[49]

Qwen-vl: A frontier large vision-language model with versatile abilities,

J. Bai, S. Bai, S. Y ang, et al. , “Qwen-vl: A frontier large vision-language model with versatile abilities,” arXiv preprint arXiv:2308.12966, 2023

2023 arXiv
[50]

Deepseek-vl: towards real-world vision-language understanding,

H. Lu, W. Liu, B. Zhang, et al. , “Deepseek-vl: towards real-world vision-language understanding,” arXiv preprint arXiv:2403.05525 , 2024

2024 arXiv
[51]

Mobileaibench: Benchmark- ing LLMs and LMMs for on-device use cases,

R. Murthy , L. Y ang, J. Tan, et al. , “Mobileaibench: Benchmark- ing LLMs and LMMs for on-device use cases,” arXiv preprint arXiv:2406.10290, 2024

2024 arXiv

Pith tools

Reviewed August 10, 2026 · model on record in the stance chip above.

[1] [1]

A hybrid deep sensor anomaly detection for autonomous vehicles in 6G-V2X environment,

S. B. Prathiba, G. Raja, S. Anbalagan, et al. , “A hybrid deep sensor anomaly detection for autonomous vehicles in 6G-V2X environment,” IEEE T rans. Network Sci. Eng. , vol. 10, no. 3, pp. 1246–1255, 2023

work page 2023

[2] [2]

Clouds on the road: A software-deﬁned fog computing framework for intelligent re- source management in vehicular Ad-Hoc networks,

A. Nahar, K. K. Mondal, D. Das, et al. , “Clouds on the road: A software-deﬁned fog computing framework for intelligent re- source management in vehicular Ad-Hoc networks,” IEEE T rans. Mob. Comput., vol. 23, no. 12, pp. 12 778–12 792, 2024. JOURNAL OF LATEX CLASS FILES, VOL. 14, NO. 8, AUGUST 2015 14

work page 2024

[3] [3]

Privacy-preserving and secure distributed data sharing scheme for V ANETs,

L. Wang, H. Zhong, J. Cui, et al. , “Privacy-preserving and secure distributed data sharing scheme for V ANETs,” IEEE T rans. Mob. Comput., 2024

work page 2024

[4] [4]

Fedagl: A communication-efﬁcient federated vehicular network,

S. Liu, Y . Li, P . Guan, et al. , “Fedagl: A communication-efﬁcient federated vehicular network,” IEEE T rans. Intell. Veh., 2024

work page 2024

[5] [5]

A reliable physical layer key generation scheme based on RSS and LSTM network in V ANET,

Z. Wang, Y . Liu, J. Wang, et al. , “A reliable physical layer key generation scheme based on RSS and LSTM network in V ANET,” IEEE Internet Things J.l , vol. 11, no. 1, pp. 692–707, 2023

work page 2023

[6] [6]

Comprehensive survey on machine learning in vehicular network: Technology , applic ations and challenges,

F. Tang, B. Mao, N. Kato, et al. , “Comprehensive survey on machine learning in vehicular network: Technology , applic ations and challenges,” IEEE Commun. Survey T uts. , vol. 23, no. 3, pp. 2027–2057, 2021

work page 2027

[7] [7]

A survey of embodied AI: From simulators to research tasks,

J. Duan, S. Y u, H. L. Tan, et al. , “A survey of embodied AI: From simulators to research tasks,” IEEE T rans. Emerging T op. Comput. Intell., vol. 6, no. 2, pp. 230–244, 2022

work page 2022

[8] [8]

Autonomous vehic les and embedded artiﬁcial intelligence: The challenges of fra ming machine driving decisions,

M. Cunneen, M. Mullins, and F. Murphy , “Autonomous vehic les and embedded artiﬁcial intelligence: The challenges of fra ming machine driving decisions,” Applied Artiﬁcial Intelligence , vol. 33, no. 8, pp. 706–731, 2019

work page 2019