Pith. sign in

REVIEW 4 major objections 6 minor 60 references

HCRMP: A LLM-Hinted Contextual Reinforcement Learning Framework for Autonomous Driving

T0 review · 4 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

Pith's one-line read The paper argues that autonomous driving should keep the language model advisory, not authoritative, and reports up to 80.3% task success with an 11.4% collision-rate reduction in safety-critical driving.

desk verdict A useful reframing of LLM-RL integration for driving, but the evidence is too thin and the hallucination-robustness claim doesn't follow from the design. read the letter →

arxiv 2505.15793 v2 pith:CZ4B7CND submitted 2025-05-21 cs.RO cs.LG

classification cs.ROcs.LG
keywords LLM-hintedreinforcementlearningautonomousdrivingmotionplanningLLMhallucinationmitigationmulti-criticPPOretrieval-augmentedsemanticgroundingstateaugmentationlow-frequencyhigh-frequencyRL
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 argues that existing LLM-RL driving systems fail because the language model's outputs directly drive the policy or define its rewards, so hallucinations become the policy's mistakes. It proposes an LLM-hinted paradigm instead: the language model supplies semantic descriptions of the scene and adaptive weights across safety, efficiency, and comfort critics, while the reinforcement learning agent retains control of the learned policy. In the proposed HCRMP architecture, these hints enter only as state augmentation and as modulation of the multi-critic advantage estimate. Evaluations in an urban driving simulator report a task success rate up to 80.3% across traffic densities and an 11.4% reduction in collision rate in safety-critical scenarios. If correct, LLM common sense can improve RL driving without letting the model's errors dictate actions.

What carries the argument

The load-bearing object is the integrated advantage estimate $\hat A^{\mathrm{int}}(s,a) = \sum_i \lambda_i A_i^{\pi_\theta}(s,a)$ from Eq. 2, where each $\lambda_i$ is an LLM-generated, knowledge-base-anchored weight between critics for safety, efficiency, and comfort. That estimate feeds the clipped PPO objective in Eq. 3, so the LLM influences the policy only through the multiplier of each critic's advantage. The Augmented Semantic Representation module extends the raw state $s^{\mathrm{raw}}_t$ with semantic embeddings $s^{\mathrm{llm}}_t$, and the Semantic Cache module substitutes a nearest-neighbor historical weight vector whenever the LLM's low-frequency output misses its time window.

What would settle it

Replace the LLM's semantic hints in the reported simulator runs with deliberately wrong hints, such as reversed scenario classifications and swapped critic weights, and measure success and collision rates; if performance collapses, the claim that RL absorbs hallucinations is not supported, whereas if it stays near the reported levels, the buffering is real.

Watch

Extended reading notes

Core claim

The central claim is that maintaining relative independence between the LLM and the RL agent is the key to using LLM knowledge in autonomous driving: the LLM should hint, not decide. HCRMP encodes that separation with three modules: the Augmented Semantic Representation module extends the RL state with multi-level scene abstractions; the Contextual Stability Anchor module retrieves traffic-regulatory passages from a knowledge base and uses them to stabilize the LLM's per-critic weights; and the Semantic Cache module matches historical conditions when the LLM's low-frequency outputs are delayed. The paper reports that this arrangement raises task success rate to 80.3% and cuts collision rate by 11.4% relative to baselines in safety-critical conditions.

Load-bearing premise

The load-bearing premise is that the language model's semantic hints are informative enough on average, and wrong often enough only in ways the RL update can treat as noise, so the scaled advantage in the PPO objective still points toward better driving.

Editorial extensions

If this is right

  • If the 80.3% success figure holds, weakly coupled LLM hints can improve RL driving across low, medium, and high traffic densities without retraining per condition.
  • The 11.4% collision-rate reduction implies that grounding LLM-generated critic weights in a static knowledge base stabilizes multi-objective policy updates in safety-critical situations.
  • The asynchronous Semantic Cache shows a concrete way to reconcile LLM latency with real-time control: cache and retrieve past semantic states when fresh hints are unavailable.
  • The multi-critic PPO formulation suggests a general recipe for LLM-RL integration in other control domains: use LLM outputs as state features and objective multipliers, not as direct action labels or rewards.

Reading between the lines

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

  • A direct stress test the paper leaves undone: inject deliberately wrong or inverted semantic hints and measure whether success and collision rates hold; that would quantify how much of the buffering comes from the architecture rather than from the specific LLM's accuracy.
  • Because the reported non-hallucination rate of roughly 57.95% was measured on a single unnamed model, a natural extension is to vary LLM quality and map the performance degradation curve; the paper's own claim predicts a much flatter curve than LLM-dominated methods would show.
  • The normalized critic weights mean even a mistaken LLM can shift the trade-off among safety, efficiency, and comfort; adding a learned trust or calibration layer on top of the knowledge base might be the next step beyond this paper.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The paper proposes HCRMP, an LLM-hinted reinforcement-learning motion-planning framework for autonomous driving in the CARLA simulator. The architecture augments the RL state with LLM-generated scenario- and object-level semantic hints (ASR module), modulates a multi-critic PPO advantage estimate with LLM-generated per-attribute weights that are anchored by retrieval-augmented generation from a knowledge base (CSA module), and decouples the low-frequency LLM from high-frequency RL control via a semantic cache (SCM). The central claim is that maintaining 'relative independence' between the LLM and the RL agent makes the system robust to LLM hallucinations, with a reported task success rate of up to 80.3% and an 11.4% collision-rate reduction in safety-critical conditions. Experiments compare HCRMP to Vanilla PPO, E2ECLA, AutoReward, and VLM-RL, and include an ablation in a single trilemma scenario.

Significance. If the hallucination-robustness claim were supported, the weakly coupled design of using LLM hints for state augmentation and critic-weight modulation rather than for direct action or reward generation would be a worthwhile contribution to the LLM-RL autonomous-driving literature. The paper also offers a clear taxonomy of LLM-dominated RL methods and a plausible asynchronous execution scheme. However, the current evidence does not establish the headline claim: there are no hallucination-injection experiments, no explicit mechanism for the RL agent to detect or discard unreliable LLM weights, and no statistical confidence measures for the point estimates in the result tables. With additional experiments and architectural clarification, the approach could be of interest to the autonomous-driving and LLM-RL communities.

major comments (4)
  1. [Section 3.3, Eqs. (2)-(3)] The claimed ability of the RL agent to 'counteract potential erroneous semantic indications through policy learning' is not realized in the architecture as written. The LLM-generated weights λ_i scale each critic advantage directly in the GAE estimate, and the PPO clipping objective in Eq. (3) maximizes the probability ratio weighted by this advantage. A systematically wrong λ_i, for instance a near-zero safety weight in a hazardous situation, changes the sign and magnitude of the advantage for safe actions, so the policy gradient actively pushes the agent away from safe behavior. The actor network receives no uncertainty estimate or reliability gate for λ_i; the RAG Top-3 retrieval in the CSA module reduces the frequency of erroneous weights but does not eliminate this structural coupling. Thus, the headline robustness claim requires either an architectural mechanism that detects or ignores unreliable weights, or a direct hallucination-injection experiment demonstrating that corrupted λ_i do not degrade policy performance. Neither is provided; the paper's own motivating statistic that a SOTA LLM has only 57.95% non-hallucination rate actually implies that a large fraction of λ_i can be biased.
  2. [Section 4.2, Tables 1-2] All reported metrics are single point estimates with no error bars, number of seeds, or episode counts. The key comparative claim of an 11.4% collision-rate reduction appears to be the difference between HCRMP and VLM-RL averaged over the six safety-critical rows (32.3% versus 43.7%), but the text never states this comparison explicitly, and without variance or significance testing the difference could be within random variation. Additionally, the HCRMP Trilemma Low row reports SR=75 and CR=28, which sum to 103 and cannot both be true under the paper's definitions; this internal inconsistency needs correction or explanation.
  3. [Section 4.1-4.2 and Figure 1] The LLM used inside HCRMP is never named. The paper evaluates Gemini-2.5-Pro in Figure 1 to motivate the hallucination problem, but it does not state whether the HCRMP experiments use the same model, a different model, which prompt templates are used, or how the 57.95% non-hallucination rate was measured (dataset, question set, scoring rubric). Without this information, the extent to which HCRMP actually encounters hallucination rates similar to those claimed cannot be assessed, and the experiments are not reproducible.
  4. [Section 4.3, Table 3 and Figure 3] The ablation is run only on the medium-density trilemma scenario and reports single-seed point estimates. The reward-curve comparison in Figure 3 has no axis labels or confidence bands, so the claim that CSA 'stabilizes' training rests on qualitative visual inspection. Furthermore, the naming of ablation variants is confusing: 'HCRMP w/ ASR' in Table 3 appears to denote the full model, but the text contrasts it with 'HCRMP without ASR' without clarifying whether CSA is present in both. The table should be relabeled and the exact configuration of each row stated.
minor comments (6)
  1. [Abstract and Section 5] The reported 'task success rate of up to 80.3%' is not traceable in Tables 1 and 2; the highest tabulated success rate is 99% (overtaking, low density) and the maximum in safety-critical conditions is 75%. Please clarify how the 80.3% figure is computed.
  2. [Section 3.2] The 'semantic compensation mechanism' for low-density traffic is mentioned but never described; its inputs, outputs, and effect on the state vector should be specified.
  3. [Figure 3] The axes in Figure 3 are unlabeled; the caption should define what 'Reward' and 'Step' refer to and which training or evaluation episodes are plotted.
  4. [Section 1] The sentence 'LLM poss strong semantic understanding' contains a typo, and the phrase 'intrinsic modulation' undercuts the 'relative independence' terminology used elsewhere in the same paragraph; the intended degree of coupling should be defined precisely.
  5. [Section 4.2] The sentence attributing 'efficiency and comfort' to ASR and 'extending the state space' to CSA appears to swap the roles of the two modules as defined in Section 3; please correct this attribution.
  6. [Section 3.4] The Semantic Cache Module is not directly evaluated; the paper should report how often the cache fallback occurs and how the nearest-neighbor matching performance affects success and collision rates.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: HCRMP's empirical results are self-contained against external baselines, and the LLM-hint routing concern is a robustness issue rather than a circular reduction.

full rationale

The paper contains no derivation chain in which an output is defined in terms of the very quantity it claims to predict. HCRMP's reported success and collision rates are empirical results from CARLA experiments against external baselines (Vanilla PPO, E2ECLA, AutoReward, VLM-RL), not quantities fitted from the same data and then renamed as predictions. Equations (2) and (3) do route LLM-generated weights lambda_i directly into the advantage estimate and clipping objective, which weakens the paper's claimed 'relative independence' and hallucination-buffering mechanism; however, this is an architectural robustness or correctness limitation, not a circular step, because the paper never formally derives a predicted metric from those equations. The 57.95% non-hallucination rate is cited as motivation rather than used as a fitted parameter. No load-bearing self-citations appear: none of the cited prior works overlap with the author list, and no uniqueness theorem or ansatz is imported from the authors' own previous work. The central claim is therefore self-contained with respect to circularity, even though its empirical support may be questioned on other grounds.

Assumptions & free parameters 3 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the assumption that LLM hints are sufficiently reliable and that the RL policy can absorb their errors, plus the proxy validity of CARLA as a driving environment. No numerical constants are fitted to data; the lambda_i weights are generated per state by an LLM whose identity is not disclosed.

free parameters (3)
  • Semantic vector dimensions (4 scenario-level + 9 object-level)
    Hand-coded encoding scheme in Section 3.2; no fitting procedure is described.
  • Top-3 retrieval selection = 3
    Chosen by hand in Section 3.3 for the RAG-based semantic anchor; no sensitivity analysis is provided.
  • LLM-generated critic weight hints lambda_i = state-dependent values in [0,1]
    These are outputs of the LLM in Eq. 2, not fitted by the authors, but they are also not derived from any principled optimization; the paper provides no calibration procedure.
assumptions (5)
  • domain assumption The MDP formulation with a state space combining raw features and LLM-augmented semantic features is a valid model of the driving task.
    Introduced in Section 3.1; the paper assumes this state representation contains enough information for optimal control.
  • standard math PPO with a multi-critic advantage combination is a sound policy optimization method for this problem.
    PPO is a standard algorithm [53]; the multi-critic extension in Eq. 2 and Eq. 3 is assumed to converge to a useful policy.
  • ad hoc to paper LLM semantic hints are informative on average, and the RL agent can learn to disregard unreliable hints.
    This is the central premise of the LLM-Hinted paradigm in Section 1 and Section 3.3; the paper does not test it with deliberate hallucination injections.
  • domain assumption CARLA Town 2 scenarios are representative of safety-critical autonomous driving conditions.
    All experiments are run in Town 2 of CARLA as stated in Section 4.1.1; generalization to real roads is assumed, not demonstrated.
  • domain assumption The reported 57.95% non-hallucination rate for Gemini-2.5-Pro is an accurate measurement of LLM reliability.
    Mentioned in Section 1 with reference [38], but the evaluation protocol, prompt set, and scoring criteria are not described in this paper.

how reviews work

0 comments
Cite this review

Pith. "Pith review of HCRMP: A LLM-Hinted Contextual Reinforcement Learning Framework for Autonomous Driving." pith.science (2026). https://pith.science/paper/CZ4B7CND

@misc{pith2026250515793,
  author       = {Pith},
  title        = {Pith review of: HCRMP: A LLM-Hinted Contextual Reinforcement Learning Framework for Autonomous Driving},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/CZ4B7CND}},
  note         = {Machine review of arXiv:2505.15793}
}
read the original abstract

Integrating Large Language Models (LLMs) with Reinforcement Learning (RL) can enhance autonomous driving (AD) performance in complex scenarios. However, current LLM-Dominated RL methods over-rely on LLM outputs, which are prone to hallucinations. Evaluations show that state-of-the-art LLM indicates a non-hallucination rate of only approximately 57.95% when assessed on essential driving-related tasks. Thus, in these methods, hallucinations from the LLM can directly jeopardize the performance of driving policies. This paper argues that maintaining relative independence between the LLM and the RL is vital for solving the hallucinations problem. Consequently, this paper is devoted to propose a novel LLM-Hinted RL paradigm. The LLM is used to generate semantic hints for state augmentation and policy optimization to assist RL agent in motion planning, while the RL agent counteracts potential erroneous semantic indications through policy learning to achieve excellent driving performance. Based on this paradigm, we propose the HCRMP (LLM-Hinted Contextual Reinforcement Learning Motion Planner) architecture, which is designed that includes Augmented Semantic Representation Module to extend state space. Contextual Stability Anchor Module enhances the reliability of multi-critic weight hints by utilizing information from the knowledge base. Semantic Cache Module is employed to seamlessly integrate LLM low-frequency guidance with RL high-frequency control. Extensive experiments in CARLA validate HCRMP's strong overall driving performance. HCRMP achieves a task success rate of up to 80.3% under diverse driving conditions with different traffic densities. Under safety-critical driving conditions, HCRMP significantly reduces the collision rate by 11.4%, which effectively improves the driving performance in complex scenarios.

Figures

Figures reproduced from arXiv: 2505.15793 by the authors.

Figure 1
Figure 1. LLM performance evaluation and hallucination impact on LLM-RL methods. Figure (a) shows the SOTA [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. The framework of our proposed HCRMP. LLM acts in the Augmented Semantic Representation module [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. HCRMP variants rewards: dynamic trends and statistical distributions. Figure (a) visualizes that the dynamic [PITH_FULL_IMAGE:figures/full_fig_p008_3.png] view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

60 extracted references · 23 canonical work pages

  1. [1]

    Reinforcement learning algorithms: A brief survey

    Ashish Kumar Shakya, Gopinatha Pillai, and Sohom Chakrabarty. Reinforcement learning algorithms: A brief survey. Expert Systems with Applications, 231:120495, 2023

  2. [2]

    Deep reinforcement learning: A survey

    Xu Wang, Sen Wang, Xingxing Liang, Dawei Zhao, Jincai Huang, Xin Xu, Bin Dai, and Qiguang Miao. Deep reinforcement learning: A survey. IEEE Transactions on Neural Networks and Learning Systems, 35(4):5064– 5078, 2022

  3. [3]

    A review of safe reinforcement learning: Methods, theory and applications

    Shangding Gu, Long Yang, Yali Du, Guang Chen, Florian Walter, Jun Wang, and Alois Knoll. A review of safe reinforcement learning: Methods, theory and applications. arXiv preprint arXiv:2205.10330, 2022

  4. [4]

    Milestones in autonomous driving and intelligent vehicles: Survey of surveys

    Long Chen, Yuchen Li, Chao Huang, Bai Li, Yang Xing, Daxin Tian, Li Li, Zhongxu Hu, Xiaoxiang Na, Zixuan Li, et al. Milestones in autonomous driving and intelligent vehicles: Survey of surveys. IEEE Transactions on Intelligent Vehicles, 8(2):1046–1056, 2022

  5. [5]

    Event-triggered model predictive control with deep reinforcement learning for autonomous driving

    Fengying Dang, Dong Chen, Jun Chen, and Zhaojian Li. Event-triggered model predictive control with deep reinforcement learning for autonomous driving. IEEE transactions on intelligent vehicles, 9(1):459–468, 2023

  6. [6]

    Deep reinforcement learning with nmpc assistance nash switching for urban autonomous driving

    Sina Alighanbari and Nasser L Azad. Deep reinforcement learning with nmpc assistance nash switching for urban autonomous driving. IEEE Transactions on Intelligent Vehicles, 8(3):2604–2615, 2022

  7. [7]

    Language models are few-shot learners

    Tom Brown, Benjamin Mann, Nick Ryder, Melanie Subbiah, Jared D Kaplan, Prafulla Dhariwal, Arvind Nee- lakantan, Pranav Shyam, Girish Sastry, Amanda Askell, et al. Language models are few-shot learners. Advances in neural information processing systems, 33:1877–1901, 2020

  8. [8]

    Llama: Open and efficient foundation language models

    Hugo Touvron, Thibaut Lavril, Gautier Izacard, Xavier Martinet, Marie-Anne Lachaux, Timothée Lacroix, Baptiste Rozière, Naman Goyal, Eric Hambro, Faisal Azhar, et al. Llama: Open and efficient foundation language models. arXiv preprint arXiv:2302.13971, 2023

Show all 60 references
  1. [9]

    Explain yourself! leveraging language models for commonsense reasoning

    Nazneen Fatema Rajani, Bryan McCann, Caiming Xiong, and Richard Socher. Explain yourself! leveraging language models for commonsense reasoning. arXiv preprint arXiv:1906.02361, 2019

  2. [10]

    Can large language models explain themselves? a study of llm-generated self-explanations

    Shiyuan Huang, Siddarth Mamidanna, Shreedhar Jangam, Yilun Zhou, and Leilani H Gilpin. Can large language models explain themselves? a study of llm-generated self-explanations. arXiv preprint arXiv:2310.11207, 2023

  3. [11]

    The llama 3 herd of models

    Aaron Grattafiori, Abhimanyu Dubey, Abhinav Jauhri, Abhinav Pandey, Abhishek Kadian, Ahmad Al-Dahle, Aiesha Letman, Akhil Mathur, Alan Schelten, Alex Vaughan, et al. The llama 3 herd of models. arXiv preprint arXiv:2407.21783, 2024

  4. [12]

    Deepseek llm: Scaling open-source language models with longtermism

    Xiao Bi, Deli Chen, Guanting Chen, Shanhuang Chen, Damai Dai, Chengqi Deng, Honghui Ding, Kai Dong, Qiushi Du, Zhe Fu, et al. Deepseek llm: Scaling open-source language models with longtermism. arXiv preprint arXiv:2401.02954, 2024. 9 HCRMP: A LLM-Hinted Contextual Reinforceme...

  5. [13]

    Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning

    Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qihao Zhu, Shirong Ma, Peiyi Wang, Xiao Bi, et al. Deepseek-r1: Incentivizing reasoning capability in llms via reinforcement learning. arXiv preprint arXiv:2501.12948, 2025

  6. [14]

    Limsim++: A closed-loop platform for deploying multimodal llms in autonomous driving

    Daocheng Fu, Wenjie Lei, Licheng Wen, Pinlong Cai, Song Mao, Min Dou, Botian Shi, and Yu Qiao. Limsim++: A closed-loop platform for deploying multimodal llms in autonomous driving. In 2024 IEEE Intelligent Vehicles Symposium (IV), pages 1084–1090. IEEE, 2024

  7. [15]

    Lampilot: An open benchmark dataset for autonomous driving with language model programs

    Yunsheng Ma, Can Cui, Xu Cao, Wenqian Ye, Peiran Liu, Juanwu Lu, Amr Abdelraouf, Rohit Gupta, Kyungtae Han, Aniket Bera, et al. Lampilot: An open benchmark dataset for autonomous driving with language model programs. In Proceedings of the IEEE/CVF Conference on Computer Vision...

  8. [16]

    Drivearena: A closed-loop generative simulation platform for autonomous driving

    Xuemeng Yang, Licheng Wen, Yukai Ma, Jianbiao Mei, Xin Li, Tiantian Wei, Wenjie Lei, Daocheng Fu, Pinlong Cai, Min Dou, et al. Drivearena: A closed-loop generative simulation platform for autonomous driving. arXiv preprint arXiv:2408.00415, 2024

  9. [17]

    Driving with llms: Fusing object-level vector modality for explainable autonomous driving

    Long Chen, Oleg Sinavski, Jan Hünermann, Alice Karnsund, Andrew James Willmott, Danny Birch, Daniel Maund, and Jamie Shotton. Driving with llms: Fusing object-level vector modality for explainable autonomous driving. In 2024 IEEE International Conference on Robotics and Automa...

  10. [18]

    Llm4drive: A survey of large language models for autonomous driving

    Zhenjie Yang, Xiaosong Jia, Hongyang Li, and Junchi Yan. Llm4drive: A survey of large language models for autonomous driving. arXiv preprint arXiv:2311.01043, 2023

  11. [19]

    Survey on large language model-enhanced reinforcement learning: Concept, taxonomy, and methods

    Yuji Cao, Huan Zhao, Yuheng Cheng, Ting Shu, Yue Chen, Guolong Liu, Gaoqi Liang, Junhua Zhao, Jinyue Yan, and Yun Li. Survey on large language model-enhanced reinforcement learning: Concept, taxonomy, and methods. IEEE Transactions on Neural Networks and Learning Systems, 2024

  12. [20]

    The evolving landscape of llm-and vlm-integrated reinforcement learning

    Sheila Schoepp, Masoud Jafaripour, Yingyue Cao, Tianpei Yang, Fatemeh Abdollahi, Shadan Golestan, Zahin Sufiyan, Osmar R Zaiane, and Matthew E Taylor. The evolving landscape of llm-and vlm-integrated reinforcement learning. arXiv preprint arXiv:2502.15214, 2025

  13. [21]

    Robust rl with llm-driven data synthesis and policy adaptation for autonomous driving

    Sihao Wu, Jiaxu Liu, Xiangyu Yin, Guangliang Cheng, Xingyu Zhao, Meng Fang, Xinping Yi, and Xiaowei Huang. Robust rl with llm-driven data synthesis and policy adaptation for autonomous driving. arXiv preprint arXiv:2410.12568, 2024

  14. [22]

    Highwayllm: Decision-making and navigation in highway driving with rl-informed language model

    Mustafa Yildirim, Barkin Dagda, and Saber Fallah. Highwayllm: Decision-making and navigation in highway driving with rl-informed language model. arXiv preprint arXiv:2405.13547, 2024

  15. [23]

    Alphadrive: Unleashing the power of vlms in autonomous driving via reinforcement learning and reasoning

    Bo Jiang, Shaoyu Chen, Qian Zhang, Wenyu Liu, and Xinggang Wang. Alphadrive: Unleashing the power of vlms in autonomous driving via reinforcement learning and reasoning. arXiv preprint arXiv:2503.07608, 2025

  16. [24]

    Optimizing autonomous driving for safety: A human- centric approach with llm-enhanced rlhf

    Yuan Sun, Navid Salami Pargoo, Peter Jin, and Jorge Ortiz. Optimizing autonomous driving for safety: A human- centric approach with llm-enhanced rlhf. In Companion of the 2024 on ACM International Joint Conference on Pervasive and Ubiquitous Computing, pages 76–80, 2024

  17. [25]

    Vlm-rl: A unified vision language models and reinforcement learning framework for safe autonomous driving

    Zilin Huang, Zihao Sheng, Yansong Qu, Junwei You, and Sikai Chen. Vlm-rl: A unified vision language models and reinforcement learning framework for safe autonomous driving. arXiv preprint arXiv:2412.15544, 2024

  18. [26]

    Curricuvlm: Towards safe autonomous driving via personalized safety-critical curriculum learning with vision-language models

    Zihao Sheng, Zilin Huang, Yansong Qu, Yue Leng, Sruthi Bhavanam, and Sikai Chen. Curricuvlm: Towards safe autonomous driving via personalized safety-critical curriculum learning with vision-language models. arXiv preprint arXiv:2502.15119, 2025

  19. [27]

    Large language model guided deep reinforcement learning for decision making in autonomous driving

    Hao Pang, Zhenpo Wang, and Guoqiang Li. Large language model guided deep reinforcement learning for decision making in autonomous driving. arXiv preprint arXiv:2412.18511, 2024

  20. [28]

    Learningflow: Automated policy learning workflow for urban driving with large language models

    Zengqi Peng, Yubin Wang, Xu Han, Lei Zheng, and Jun Ma. Learningflow: Automated policy learning workflow for urban driving with large language models. arXiv preprint arXiv:2501.05057, 2025

  21. [29]

    Autoreward: Closed-loop reward design with large language models for autonomous driving

    Xu Han, Qiannan Yang, Xianda Chen, Zhenghan Cai, Xiaowen Chu, and Meixin Zhu. Autoreward: Closed-loop reward design with large language models for autonomous driving. IEEE Transactions on Intelligent Vehicles, 2024

  22. [30]

    Clip-rldrive: Human-aligned autonomous driving via clip-based reward shaping in reinforcement learning

    Erfan Doroudian and Hamid Taghavifar. Clip-rldrive: Human-aligned autonomous driving via clip-based reward shaping in reinforcement learning. arXiv preprint arXiv:2412.16201, 2024

  23. [31]

    Lord: Large models based opposite reward design for autonomous driving

    Xin Ye, Feng Tao, Abhirup Mallik, Burhaneddin Yaman, and Liu Ren. Lord: Large models based opposite reward design for autonomous driving. In 2025 IEEE/CVF Winter Conference on Applications of Computer Vision (WACV), pages 5072–5081. IEEE, 2025. 10 HCRMP: A LLM-Hinted Contextua...

  24. [32]

    Revolve: Reward evolution with large language models for autonomous driving

    Rishi Hazra, Alkis Sygkounas, Andreas Persson, Amy Loutfi, and Pedro Zuidberg Dos Martires. Revolve: Reward evolution with large language models for autonomous driving. arXiv e-prints, pages arXiv–2406, 2024

  25. [33]

    Llms know more than they show: On the intrinsic representation of llm hallucinations

    Hadas Orgad, Michael Toker, Zorik Gekhman, Roi Reichart, Idan Szpektor, Hadas Kotek, and Yonatan Be- linkov. Llms know more than they show: On the intrinsic representation of llm hallucinations. arXiv preprint arXiv:2410.02707, 2024

  26. [34]

    Crosscheckgpt: Universal hallucination ranking for multimodal foundation models

    Guangzhi Sun, Potsawee Manakul, Adian Liusie, Kunat Pipatanakul, Chao Zhang, Phil Woodland, and Mark Gales. Crosscheckgpt: Universal hallucination ranking for multimodal foundation models. arXiv preprint arXiv:2405.13684, 2024

  27. [35]

    Llm hallucinations in practical code generation: Phenomena, mechanism, and mitigation

    Ziyao Zhang, Yanlin Wang, Chong Wang, Jiachi Chen, and Zibin Zheng. Llm hallucinations in practical code generation: Phenomena, mechanism, and mitigation. arXiv preprint arXiv:2409.20550, 2024

  28. [36]

    Exploring and evaluating hallucinations in llm-powered code generation

    Fang Liu, Yang Liu, Lin Shi, Houkun Huang, Ruifeng Wang, Zhen Yang, Li Zhang, Zhongqi Li, and Yuchi Ma. Exploring and evaluating hallucinations in llm-powered code generation. arXiv preprint arXiv:2404.00971, 2024

  29. [37]

    Llm-check: Investigating detection of hallucinations in large language models

    Gaurang Sriramanan, Siddhant Bharti, Vinu Sankar Sadasivan, Shoumik Saha, Priyatham Kattakinda, and Soheil Feizi. Llm-check: Investigating detection of hallucinations in large language models. Advances in Neural Information Processing Systems, 37:34188–34216, 2024

  30. [38]

    Sti-bench: Are mllms ready for precise spatial-temporal world understanding? arXiv preprint arXiv:2503.23765, 2025

    Yun Li, Yiming Zhang, Tao Lin, XiangRui Liu, Wenxiao Cai, Zheng Liu, and Bo Zhao. Sti-bench: Are mllms ready for precise spatial-temporal world understanding? arXiv preprint arXiv:2503.23765, 2025

  31. [39]

    Rladapter: Bridging large language models to reinforcement learning in open worlds

    Wanpeng Zhang and Zongqing Lu. Rladapter: Bridging large language models to reinforcement learning in open worlds. 2023

  32. [40]

    Pre-trained language models for interactive decision-making

    Shuang Li, Xavier Puig, Chris Paxton, Yilun Du, Clinton Wang, Linxi Fan, Tao Chen, De-An Huang, Ekin Akyürek, Anima Anandkumar, et al. Pre-trained language models for interactive decision-making. Advances in Neural Information Processing Systems, 35:31199–31212, 2022

  33. [41]

    Unleashing the power of pre-trained language models for offline reinforcement learning

    Ruizhe Shi, Yuyao Liu, Yanjie Ze, Simon S Du, and Huazhe Xu. Unleashing the power of pre-trained language models for offline reinforcement learning. arXiv preprint arXiv:2310.20587, 2023

  34. [42]

    Ask more, know better: Reinforce-learned prompt questions for decision making with large language models

    Xue Yan, Yan Song, Xinyu Cui, Filippos Christianos, Haifeng Zhang, David Henry Mguni, and Jun Wang. Ask more, know better: Reinforce-learned prompt questions for decision making with large language models. arXiv preprint arXiv:2310.18127, 2023

  35. [43]

    Efficient reinforcement learning with large language model priors

    Xue Yan, Yan Song, Xidong Feng, Mengyue Yang, Haifeng Zhang, Haitham Bou Ammar, and Jun Wang. Efficient reinforcement learning with large language model priors. arXiv preprint arXiv:2410.07927, 2024

  36. [44]

    Grounding large language models in interactive environments with online reinforcement learning

    Thomas Carta, Clément Romac, Thomas Wolf, Sylvain Lamprier, Olivier Sigaud, and Pierre-Yves Oudeyer. Grounding large language models in interactive environments with online reinforcement learning. In International Conference on Machine Learning, pages 3676–3713. PMLR, 2023

  37. [45]

    Fine-tuning large vision-language models as decision-making agents via reinforcement learning

    Simon Zhai, Hao Bai, Zipeng Lin, Jiayi Pan, Peter Tong, Yifei Zhou, Alane Suhr, Saining Xie, Yann LeCun, Yi Ma, et al. Fine-tuning large vision-language models as decision-making agents via reinforcement learning. Advances in neural information processing systems, 37:110935–11...

  38. [46]

    Teaching large language models to reason with reinforcement learning

    Alex Havrilla, Yuqing Du, Sharath Chandra Raparthy, Christoforos Nalmpantis, Jane Dwivedi-Yu, Maksym Zhuravinskyi, Eric Hambro, Sainbayar Sukhbaatar, and Roberta Raileanu. Teaching large language models to reason with reinforcement learning. arXiv preprint arXiv:2403.04642, 2024

  39. [47]

    Latent reward: Llm-empowered credit assignment in episodic reinforcement learning

    Yun Qu, Yuhang Jiang, Boyuan Wang, Yixiu Mao, Cheems Wang, Chang Liu, and Xiangyang Ji. Latent reward: Llm-empowered credit assignment in episodic reinforcement learning. In Proceedings of the AAAI Conference on Artificial Intelligence, volume 39, pages 20095–20103, 2025

  40. [48]

    Text2reward: Reward shaping with language models for reinforcement learning

    Tianbao Xie, Siheng Zhao, Chen Henry Wu, Yitao Liu, Qian Luo, Victor Zhong, Yanchao Yang, and Tao Yu. Text2reward: Reward shaping with language models for reinforcement learning. arXiv preprint arXiv:2309.11489, 2023

  41. [49]

    Lmpriors: Pre-trained language models as task-specific priors

    Kristy Choi, Chris Cundy, Sanjari Srivastava, and Stefano Ermon. Lmpriors: Pre-trained language models as task-specific priors. arXiv preprint arXiv:2210.12530, 2022

  42. [50]

    Guiding pretraining in reinforcement learning with large language models

    Yuqing Du, Olivia Watkins, Zihan Wang, Cédric Colas, Trevor Darrell, Pieter Abbeel, Abhishek Gupta, and Jacob Andreas. Guiding pretraining in reinforcement learning with large language models. In International Conference on Machine Learning, pages 8657–8677. PMLR, 2023

  43. [51]

    Eureka: Human-level reward design via coding large language models

    Yecheng Jason Ma, William Liang, Guanzhi Wang, De-An Huang, Osbert Bastani, Dinesh Jayaraman, Yuke Zhu, Linxi Fan, and Anima Anandkumar. Eureka: Human-level reward design via coding large language models. arXiv preprint arXiv:2310.12931, 2023. 11 HCRMP: A LLM-Hinted Contextual...

  44. [52]

    Shufflenet v2: Practical guidelines for efficient cnn architecture design

    Ningning Ma, Xiangyu Zhang, Hai-Tao Zheng, and Jian Sun. Shufflenet v2: Practical guidelines for efficient cnn architecture design. In Proceedings of the European conference on computer vision (ECCV), pages 116–131, 2018

  45. [53]

    Proximal policy optimization algorithms

    John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347, 2017

  46. [54]

    Retrieval-augmented generation for knowledge- intensive nlp tasks

    Patrick Lewis, Ethan Perez, Aleksandra Piktus, Fabio Petroni, Vladimir Karpukhin, Naman Goyal, Heinrich Küttler, Mike Lewis, Wen-tau Yih, Tim Rocktäschel, et al. Retrieval-augmented generation for knowledge- intensive nlp tasks. Advances in neural information processing system...

  47. [55]

    Driving with regulation: Interpretable decision-making for autonomous vehicles with retrieval-augmented reasoning via llm

    Tianhui Cai, Yifan Liu, Zewei Zhou, Haoxuan Ma, Seth Z Zhao, Zhiwen Wu, and Jiaqi Ma. Driving with regulation: Interpretable decision-making for autonomous vehicles with retrieval-augmented reasoning via llm. arXiv preprint arXiv:2410.04759, 2024

  48. [56]

    Rag-driver: Generalisable driving explanations with retrieval-augmented in-context learning in multi-modal large language model

    Jianhao Yuan, Shuyang Sun, Daniel Omeiza, Bo Zhao, Paul Newman, Lars Kunze, and Matthew Gadd. Rag-driver: Generalisable driving explanations with retrieval-augmented in-context learning in multi-modal large language model. arXiv preprint arXiv:2402.10828, 2024

  49. [57]

    Billion-scale similarity search with gpus

    Jeff Johnson, Matthijs Douze, and Hervé Jégou. Billion-scale similarity search with gpus. IEEE Transactions on Big Data, 7(3):535–547, 2019

  50. [58]

    Carla: An open urban driving simulator

    Alexey Dosovitskiy, German Ros, Felipe Codevilla, Antonio Lopez, and Vladlen Koltun. Carla: An open urban driving simulator. In Conference on robot learning, pages 1–16. PMLR, 2017

  51. [59]

    An end-to-end curriculum learning approach for autonomous driving scenarios

    Luca Anzalone, Paola Barra, Silvio Barra, Aniello Castiglione, and Michele Nappi. An end-to-end curriculum learning approach for autonomous driving scenarios. IEEE Transactions on Intelligent Transportation Systems, 23(10):19817–19826, 2022

  52. [60]

    Standards for passenger comfort in automated vehicles: Acceleration and jerk

    Ksander N De Winkel, Tugrul Irmak, Riender Happee, and Barys Shyrokau. Standards for passenger comfort in automated vehicles: Acceleration and jerk. Applied Ergonomics, 106:103881, 2023. 12

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.