Prompt Informed Reinforcement Learning for Visual Coverage Path Planning

Venkat Margapuri

REVIEW 4 major objections 6 minor 32 references

Prompt Informed Reinforcement Learning for Visual Coverage Path Planning

T0 review · 4 major / 6 minor · reviewed 2026-08-06 · deepseek-v4-flash

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

Pith's one-line read Prompt-Informed Reinforcement Learning (PIRL) claims that GPT-3.5's zero-shot position and camera advice, added to PPO's reward as a soft deviation penalty, materially improves visual coverage path planning for a simulated UAV.

desk verdict A plausible but under-supported application of LLM reward shaping to coverage planning; the key ablation—LLM advice versus generic shaping structure—is missing. read the letter →

arxiv 2507.10284 v1 pith:2AKZ7RH5 submitted 2025-07-14 cs.RO cs.MA

Venkat Margapuri This is my paper

classification cs.ROcs.MA

keywords visualcoveragepathplanningunmannedaerialvehiclereinforcementlearningrewardshapinglargelanguagemodelsproximalpolicyoptimizationzero-shottransferlanguage-guidedexploration

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

PIRL is offered as a way to make a reinforcement-learning agent for visual coverage path planning follow high-level semantic guidance without turning the language model into the controller. The paper claims that GPT-3.5, given a structured text summary of the UAV's position, camera pose, battery, and coverage state, can recommend where to move and how to point the camera, and that folding those recommendations into PPO's reward as a soft deviation penalty materially improves learning. Across OpenAI Gym and the physics-based Webots simulator, PIRL reports up to 14% and 27% higher visual coverage over the best baselines, up to 25% higher battery efficiency in one configuration, and lower redundant viewing, with zero-shot transfer from training to Webots. If true, the result matters because it offers a route from natural-language task knowledge to better exploration behavior without expert demonstrations or per-environment reward re-engineering.

What carries the argument

The load-bearing object is the Prompt-Adaptive Reward Engine (PARE), the module that turns each LLM recommendation into a reward term. PARE computes $f_{\mathrm{LLM}}(s_{t+1}) = \lambda_{\mathrm{cam}} \cdot ( -\|\theta_{t+1} - \theta^{\mathrm{LLM}}_{t+1}\|_1 ) + \lambda_{\mathrm{move}} \cdot ( \alpha \cdot \mathrm{DirAlign}(s_{t+1}) + (1-\alpha) \cdot \mathrm{PosAlign}(s_{t+1}) )$, where $\mathrm{DirAlign}$ is the cosine similarity between the agent's movement vector and the LLM's recommended movement direction, and $\mathrm{PosAlign}$ is $1 - \|\mathbf{p}_{t+1} - \mathbf{p}^{\mathrm{LLM}}_{t+1}\|_2 / d_{\max}$. The auxiliary term is added to PPO's curiosity-driven coverage reward, so the language model never selects actions itself; it only reshapes the reward landscape. The structured prompt that feeds GPT-3.5 concatenates a task description, a numerical environment summary, and a request template, making the recommendation a zero-shot function of the current state.

What would settle it

Replace the LLM recommendations inside PARE with recommendations sampled uniformly from the same discrete position and camera action space, keeping all hyperparameters fixed, and rerun the 30x30x3 OpenAI Gym benchmark; if PIRL still reaches a visual coverage ratio near 0.73, the language model's semantic content is not the cause of the reported improvement.

Watch

Extended reading notes

Core claim

The paper's central claim is that a pretrained LLM can serve as a semantic reward shaper for PPO-based visual coverage path planning and that this hybrid beats both pure RL and pure LLM baselines. In PIRL, each state is serialized into a prompt; GPT-3.5 recommends a next position and camera setting; the Prompt-Adaptive Reward Engine (PARE) converts the deviation between the agent's action and that recommendation into an auxiliary reward, while the base reward continues to incentivize coverage gain and curiosity and to penalize redundancy, collisions, and idle behavior. Over 100 episodes per setting, PIRL reports a visual coverage ratio of 0.73 on the 30x30x3 Gym grid versus 0.64 for the best baseline, 0.79 on the 45x45x3 Webots grid versus 0.62, and the lowest redundancy values in every Webots setting. The authors read these results as showing that language-guided reward modulation generalizes zero-shot from the grid-based Gym to the physics-based Webots simulator.

Load-bearing premise

The whole method depends on GPT-3.5 returning useful, consistent position-and-camera suggestions from a plain-text summary of the grid; if those suggestions are noisy or systematically biased, the auxiliary reward will steer the policy away from good coverage rather than toward it.

Editorial extensions

If this is right

If PIRL's gains are real, natural-language priors can be injected into an existing RL training loop with no expert demonstrations and no per-environment reward design.
Because the agent is trained in Gym and evaluated zero-shot in Webots, the approach suggests that language-shaped rewards can produce policies robust to different simulator dynamics.
Penalizing deviation rather than enforcing recommendations lets the policy override poor LLM advice whenever the base reward disagrees, which is the intended guard against reward hacking.
Coverage gains do not always come with energy savings: in the 60x60x3 Gym setting PIRL's battery efficiency falls below baselines even though its coverage is highest, so deployment would need to tune the LLM-alignment weight or add a battery constraint.

Reading between the lines

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

An outside reader would want an ablation that swaps GPT-3.5 for a deterministic heuristic advisor inside the same PARE wrapper; if the heuristic reproduces the gains, the improvement comes from the reward-shaping structure rather than from the language model's semantic knowledge.
The reported numbers are means over 100 episodes with varied obstacles and starts, but no confidence intervals or LLM-call variance are given, so a natural robustness check is to rerun the small-grid benchmark across multiple seeds and GPT-3.5 temperature settings.
The same recipe could plausibly transfer to other spatial exploration tasks, such as agricultural monitoring or next-best-view planning, where an LLM's common-sense priors about where to look can stand in for expensive expert demonstrations.

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.

Desk Editor's Note

read the letter

The paper's real contribution is narrow but legitimate: it applies GPT-3.5 reward shaping to visual coverage path planning with PPO, and it tests zero-shot transfer from OpenAI Gym to Webots across three grid sizes. That specific comparison is not in the cited literature, and the method description (the PARE module with directional, positional, and camera alignment terms) is clear enough to reproduce in principle. The authors also chose sensible baselines, including PPO with exploratory weight initialization and an LLM-only controller, and their limitations section openly acknowledges the LLM dependency, discrete state space, and sim-to-real gap. Credit where due: this is an honest incremental study, not a fake one.

The soft spots are real and mostly statistical. Tables 2 and 3 report only means over 100 episodes, with no standard deviations, confidence intervals, or significance tests. The abstract's "up to 25% higher battery efficiency" is cherry-picked: in the 60x60x3 Gym environment PIRL gets BE 0.23 versus PPO-SR's 0.48, and the discussion admits the trade-off. More importantly, the central claim—that the LLM's semantic guidance drives the improvement—is not actually tested. The closest baseline, PPO-EWRI, removes the entire fLLM term, but that term includes both the LLM's recommendations and the shaping structure (cosine alignment, distance penalty, camera L1 penalty). Without a control that replaces GPT-3.5 with a simple heuristic advisor (move toward nearest uncovered cell, for example) while keeping the same reward formula, you cannot attribute the gains to the LLM's content rather than to the generic alignment structure. The stochasticity of GPT-3.5 is also unaddressed: no temperature setting, no seeds, no variance across calls.

Who is this for? Researchers working on LLM-informed RL for robot exploration will want to know about it, mainly as a data point and a cautionary example. It deserves a serious referee, because the questions it raises are answerable and the method is plausible. A solid revision would need the missing ablation, error bars, and full prompts/code.

Recommendation: engage with it, but push for the control experiment and statistical rigor before believing the headline numbers.

Referee Report

4 major / 6 minor

Summary. The paper proposes Prompt-Informed Reinforcement Learning (PIRL), a method that uses GPT-3.5 as a zero-shot reward shaper for a PPO-based visual coverage path planning agent. The agent is trained in a 15x15x3 OpenAI Gym environment and evaluated zero-shot in larger Gym environments and in the Webots physics simulator. The authors compare PIRL against PPO with static rewards, PPO with exploratory weight initialization, an LLM-only controller, and imitation learning, reporting VCR, BE, and RVC metrics from 100 episodes per configuration. The central claim is that LLM-guided reward shaping yields higher visual coverage, better battery efficiency, and lower redundancy than the baselines.

Significance. If the central claim were fully supported, the contribution would be of interest: using an LLM as a soft, semantic reward shaper rather than as an action generator is a plausible way to inject task priors into RL, and the zero-shot evaluation in Webots addresses an important generalization question. The paper also defines the evaluation metrics cleanly and includes a limitations section. However, the experimental design does not isolate the LLM's semantic content from the generic shaping structure, the results are reported without variance or significance measures, and the abstract overstates the battery-efficiency and overall outperformance claims relative to the tables. Because these issues are load-bearing for the stated contribution, the current evidence is insufficient to accept the paper's main conclusions.

major comments (4)

[Section 3, Section 4.4, Tables 2 and 3] The central attribution of PIRL's gains to GPT-3.5's semantic guidance is not isolated. PIRL differs from the closest baseline, PPO-EWRI, by the addition of the auxiliary reward fLLM(st+1) = λcam·fcam(st+1) + λmove·fmove(st+1), which includes directional alignment, positional alignment, and a camera L1 penalty. The reported improvements over PPO-EWRI could therefore be produced by the shaping structure alone, independent of whether the recommended targets come from an LLM, a simple heuristic, or random in-range targets. The paper provides no control experiment with a non-LLM advisor, nor an analysis of recommendation accuracy. To support the claim that 'semantic feedback' from the LLM is responsible, the authors should add ablations with, for example, a nearest-uncovered-cell heuristic and random targets, keeping the reward formula and all hyperparameters fixed.
[Section 4.4, Section 4.5, Tables 2 and 3] All quantitative claims rest on means over 100 episodes without standard deviations, confidence intervals, or significance tests. The phrase 'significantly enhances' in Section 4.5 is therefore unsupported. Additionally, GPT-3.5 is stochastic, but the paper gives no seed control, temperature setting, API version, or analysis of variance across repeated LLM calls. The authors should report dispersion measures, run multiple training seeds, and perform significance tests, and they should describe how LLM stochasticity is handled.
[Abstract, Section 4.4, Section 4.5] The abstract's claim of 'up to 25% higher battery efficiency' and the broad claim that PIRL outperforms baselines are contradicted by the paper's own tables. In OpenAI Gym 45x45x3, PIRL's BE is 0.58 versus PPO-EWRI's 0.62; in Gym 60x60x3, PIRL's BE is 0.23 versus PPO-SR's 0.48; in Webots 30x30x3, PIRL's BE is 0.42 versus PPO-SR's 0.47; and in Webots 60x60x3, PIRL's BE is 0.35 versus LLM-only's 0.36. While Section 4.5 acknowledges that BE results are 'nuanced,' the abstract and conclusion present the method as uniformly superior. The claims should be carefully scoped to the specific environments and metrics where PIRL is actually best.
[Section 3, Equation for rt+1 and PARE, Table 1] The reward definition is ambiguous about how the LLM-alignment coefficients compose. The total reward includes the term λLLM·fLLM(st+1), and fLLM is then defined as λcam·fcam(st+1) + λmove·fmove(st+1). Table 1 lists a range for λLLM but not for λcam or λmove, so it is unclear whether λLLM multiplies the whole expression or whether λcam and λmove are subsumed into the listed range. This ambiguity affects reproducibility and should be resolved with explicit coefficient values or ranges for each parameter.

minor comments (6)

[Section 3] The indicator functions in the reward equation use the symbol ⊮, which appears to be a typographical variation of the standard 1; using $\mathbb{1}$ would improve clarity.
[Section 4.4, Tables 2 and 3] The table headers contain formatting errors, such as '60x60x3VCR' run together with no space, which should be corrected in a revision.
[Section 4.2] The paper states that the EWRI ranges 'ensure that the relative priorities among RF coefficients are preserved,' but no formal argument or sensitivity analysis is provided; this should be softened or supported.
[Section 4.4] The Webots evaluation is described as 'sim-to-real-like,' but the paper also notes that Webots does not interpret image streams and that position-based coverage approximation is used; the generalization claim should be qualified accordingly.
[Section 4.2] Training details such as network architecture, learning rate, number of episodes, discount factor, and the exact GPT-3.5 prompt template are not provided; these are needed for reproducibility.
[Throughout] There are several typographical issues, including 'UAVs' appearing inconsistently as 'UA Vs' and 'GPT-3.5' being referenced without a model version or date in the experimental section; a careful proofread is recommended.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: PIRL's reward shaping is evaluated against independent coverage, battery, and redundancy metrics.

full rationale

The paper contains no derivation chain that reduces to its own inputs. PIRL augments a PPO reward with fLLM(st+1) = lambda_cam * f_cam(st+1) + lambda_move * f_move(st+1), where f_cam and f_move penalize deviation from GPT-3.5 recommendations (Section 3). The evaluation metrics VCR, BE, and RVC are defined externally (Section 4.3) in terms of unique ground cells, battery level, and redundant views; none is algebraically equivalent to LLM-alignment. The baselines (PPO-SR, PPO-EWRI, LLM-only, IL) are compared empirically, and no fitted parameter is later relabeled as a prediction. The paper's limitations (Section 4.6) concern LLM dependence, discrete state spaces, and real-world deployment gaps, none of which concedes circularity. There is no load-bearing self-citation and no imported uniqueness theorem. The central empirical claim is therefore self-contained with respect to external benchmarks, and no circular step can be exhibited from the paper's equations or citations.

Assumptions & free parameters 3 free parameters · 5 assumptions · 2 invented entities

The central claim rests on hand-chosen reward coefficient ranges, an unvalidated discrete view-cone coverage model, and the assumption that GPT-3.5 text-based recommendations are useful and stable. PARE and PIRL are algorithmic entities whose effectiveness is shown only in the paper's own experiments. No code, data, or formal proof is provided.

free parameters (3)

Reward coefficients λ (EWRI draws) = λ_c U(0.5,1.5), λ_r U(-1.0,-0.2), λ_b U(-0.5,-0.1), λ_cam U(0.2,0.6), λ_cur U(0.2,0.5), λ_collision U(-1.5,-0.8)…
Hand-chosen ranges in Table 1. The paper states these preserve objective priorities, but the central comparison depends on these choices and they are not derived.
α balance in fmove = not reported
α ∈ [0,1] balances directional versus positional alignment in PARE (Section 3). The specific value is never given, yet the shaped reward depends on it.
Training hyperparameters (PPO, network, episodes) = not reported
Learning rate, architecture, episode count, and GAE/PPO parameters are absent. These are required to reproduce training and to judge whether the baselines were fairly tuned.

assumptions (5)

standard math PPO clipped surrogate objective and Generalized Advantage Estimation reliably optimize the policy in this MDP.
Taken from [22]; Section 3 uses L_CLIP and GAE without modification.
domain assumption The discrete view-cone coverage model using position and camera parameters is a valid representation of visual coverage.
Section 3 defines ViewCone and the coverage map update. No validation against real camera imagery or continuous geometry is provided.
domain assumption A text summary of the current state is sufficient for GPT-3.5 to output useful position and camera recommendations.
Section 4.2 feeds a structured prompt to GPT-3.5. If the LLM gives poor recommendations, PARE penalizes deviation from them and actively hurts learning.
domain assumption Webots position-based coverage approximation matches the OpenAI Gym coverage evaluation.
Section 4.4 states Webots uses a position-based coverage approximation consistent with Gym. The 'sim-to-real-like' claim depends on this equivalence.
ad hoc to paper EWRI coefficient ranges ensure no reward hacking and preserve objective priorities.
Stated in Section 4.2 without evidence. The λ_LLM range is set lower than the λ_c upper bound to keep coverage dominant, but this is a manual design choice.

invented entities (2)

Prompt-Adaptive Reward Engine (PARE)
purpose: Computes the auxiliary reward fLLM that penalizes deviation from LLM recommendations, enabling LLM-guided reward shaping for PPO.
Software module introduced in this paper. Its behavior is only evaluated inside the paper's own simulations; no external benchmark or independent verification is provided.
PIRL (Prompt-Informed Reinforcement Learning)
purpose: The overall training framework that couples GPT-3.5 reward shaping with curiosity-driven PPO for visual coverage path planning.
Algorithmic contribution of the paper. Its utility is demonstrated only in the reported simulations, with no independent reproduction or analysis.

how reviews work

0 comments

Cite this review

Pith. "Pith review of Prompt Informed Reinforcement Learning for Visual Coverage Path Planning." pith.science (2026). https://pith.science/paper/2AKZ7RH5

@misc{pith2026250710284,
  author       = {Pith},
  title        = {Pith review of: Prompt Informed Reinforcement Learning for Visual Coverage Path Planning},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/2AKZ7RH5}},
  note         = {Machine review of arXiv:2507.10284}
}

read the original abstract

Visual coverage path planning with unmanned aerial vehicles (UAVs) requires agents to strategically coordinate UAV motion and camera control to maximize coverage, minimize redundancy, and maintain battery efficiency. Traditional reinforcement learning (RL) methods rely on environment-specific reward formulations that lack semantic adaptability. This study proposes Prompt-Informed Reinforcement Learning (PIRL), a novel approach that integrates the zero-shot reasoning ability and in-context learning capability of large language models with curiosity-driven RL. PIRL leverages semantic feedback from an LLM, GPT-3.5, to dynamically shape the reward function of the Proximal Policy Optimization (PPO) RL policy guiding the agent in position and camera adjustments for optimal visual coverage. The PIRL agent is trained using OpenAI Gym and evaluated in various environments. Furthermore, the sim-to-real-like ability and zero-shot generalization of the agent are tested by operating the agent in Webots simulator which introduces realistic physical dynamics. Results show that PIRL outperforms multiple learning-based baselines such as PPO with static rewards, PPO with exploratory weight initialization, imitation learning, and an LLM-only controller. Across different environments, PIRL outperforms the best-performing baseline by achieving up to 14% higher visual coverage in OpenAI Gym and 27% higher in Webots, up to 25% higher battery efficiency, and up to 18\% lower redundancy, depending on the environment. The results highlight the effectiveness of LLM-guided reward shaping in complex spatial exploration tasks and suggest a promising direction for integrating natural language priors into RL for robotics.

Figures

Figures reproduced from arXiv: 2507.10284 by the authors.

**Figure 2.** Environment with Three Obstacles and UAV in Webots The trained PPO agent is inferred in two different simulation environments, OpenAI Gym, and Webots. While OpenAI Gym offers a grid-based, logic-level simulation, Webots allows for physics-based, sensorlevel robotic simulation. Therefore, evaluating the model in Webots allows for sim-to-real-like evaluation by determining whether the learned policy can tolerate an … view at source ↗

Discussion (0). Sign in to comment.

Reference graph

Works this paper leans on

32 extracted references · 22 canonical work pages

[1]

Tabnet: Attentive interpretable tabular learning

Sercan Ö Arik and Tomas Pfister. Tabnet: Attentive interpretable tabular learning. In Proceedings of the AAAI conference on artificial intelligence, volume 35, pages 6679– 6687, 2021

work page 2021
[2]

Uav-based delivery systems: A systematic review, current trends, and research challenges

Francesco Betti Sorbelli. Uav-based delivery systems: A systematic review, current trends, and research challenges. Journal on Autonomous Transportation Systems, 1(3): 1–40, 2024

work page 2024
[3]

Openai gym

Greg Brockman, Vicki Cheung, Ludwig Pettersson, Jonas Schneider, John Schulman, Jie Tang, and Wojciech Zaremba. Openai gym. arXiv preprint arXiv:1606.01540 , 2016

arXiv 2016
[4]

Hierarchical model-based imitation learning for planning in autonomous driving

Eli Bronstein, Mark Palatucci, Dominik Notz, Brandyn White, Alex Kuefler, Yiren Lu, Supratik Paul, Payam Nikdel, Paul Mougin, Hongge Chen, et al. Hierarchical model-based imitation learning for planning in autonomous driving. In2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 8652–8659. IEEE, 2022

work page 2022
[5]

Deep reinforcement learning for zero-shot coverage path planning with mobile robots

José Pedro Carvalho and A Pedro Aguiar. Deep reinforcement learning for zero-shot coverage path planning with mobile robots. IEEE/CAA Journal of Automatica Sinica, 2025

work page 2025
[6]

Transformer-based imitative reinforcement learning for multirobot path planning

Lin Chen, Yaonan Wang, Zhiqiang Miao, Yang Mo, Mingtao Feng, Zhen Zhou, and Hesheng Wang. Transformer-based imitative reinforcement learning for multirobot path planning. IEEE Transactions on Industrial Informatics , 19(10):10233–10243, 2023

work page 2023
[7]

A proximal policy optimization with curiosity algorithm for virtual drone navigation

Rupayan Das, Angshuman Khan, and Gunjan Paul. A proximal policy optimization with curiosity algorithm for virtual drone navigation. Engineering Research Express, 6 (1):015057, 2024

work page 2024
[8]

BERT: Pre- training of deep bidirectional transformers for language understanding

Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: Pre- training of deep bidirectional transformers for language understanding. In Jill Burstein, Christy Doran, and Thamar Solorio, editors, Proceedings of the 2019 Conference of AUTHOR: VENKA T MARGAPURI 11 the North American Chapter of the Association for Computational Linguistics: Hu...

work page 2019

Show all 32 references

[9]

A survey on in-context learning

Qingxiu Dong, Lei Li, Damai Dai, Ce Zheng, Jingyuan Ma, Rui Li, Heming Xia, Jingjing Xu, Zhiyong Wu, Tianyu Liu, et al. A survey on in-context learning. arXiv preprint arXiv:2301.00234, 2022

2022 arXiv
[10]

The role of unmanned aerial vehicles in military communications: application scenarios, current trends, and beyond

Michael Gargalakos. The role of unmanned aerial vehicles in military communications: application scenarios, current trends, and beyond. The Journal of Defense Modeling and Simulation, 21(3):313–321, 2024

2024
[11]

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

Daya Guo, Dejian Yang, Haowei Zhang, Junxiao Song, Ruoyu Zhang, Runxin Xu, Qi- hao 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

2025 arXiv
[12]

Language instructed reinforcement learning for human-ai coordination

Hengyuan Hu and Dorsa Sadigh. Language instructed reinforcement learning for human-ai coordination. In International Conference on Machine Learning , pages 13584–13598. PMLR, 2023

2023
[13]

Language mod- els as zero-shot planners: Extracting actionable knowledge for embodied agents

Wenlong Huang, Pieter Abbeel, Deepak Pathak, and Igor Mordatch. Language mod- els as zero-shot planners: Extracting actionable knowledge for embodied agents. In International conference on machine learning, pages 9118–9147. PMLR, 2022

2022
[14]

General-purpose in-context learning by meta-learning transformers

Louis Kirsch, James Harrison, Jascha Sohl-Dickstein, and Luke Metz. General-purpose in-context learning by meta-learning transformers. arXiv preprint arXiv:2212.04458, 2022

2022 arXiv
[15]

Large language models are zero-shot reasoners

Takeshi Kojima, Shixiang Shane Gu, Machel Reid, Yutaka Matsuo, and Yusuke Iwa- sawa. Large language models are zero-shot reasoners. Advances in neural information processing systems, 35:22199–22213, 2022

2022
[16]

Reward design with language models

Minae Kwon, Sang Michael Xie, Kalesha Bullard, and Dorsa Sadigh. Reward design with language models. arXiv preprint arXiv:2303.00001, 2023

2023 arXiv
[17]

Self-imitation learning by plan- ning

Sha Luo, Hamidreza Kasaei, and Lambert Schomaker. Self-imitation learning by plan- ning. In 2021 IEEE International Conference on Robotics and Automation (ICRA) , pages 4823–4829. IEEE, 2021

2021
[18]

Inform: Mitigating reward hacking in rlhf via information-theoretic reward modeling

Yuchun Miao, Sen Zhang, Liang Ding, Rong Bao, Lefei Zhang, and Dacheng Tao. Inform: Mitigating reward hacking in rlhf via information-theoretic reward modeling. In The Thirty-eighth Annual Conference on Neural Information Processing Systems , 2024

2024
[19]

Colin Raffel, Noam Shazeer, Adam Roberts, Katherine Lee, Sharan Narang, Michael Matena, Yanqi Zhou, Wei Li, and Peter J. Liu. Exploring the limits of transfer learning with a unified text-to-text transformer.J. Mach. Learn. Res., 21(1), January 2020. ISSN 1532-4435

2020
[20]

A vision based deep reinforcement learning algorithm for uav obstacle avoidance

Jeremy Roghair, Amir Niaraki, Kyungtae Ko, and Ali Jannesari. A vision based deep reinforcement learning algorithm for uav obstacle avoidance. In Proceedings of SAI Intelligent Systems Conference, pages 115–128. Springer, 2021. 12 AUTHOR: VENKA T MARGAPURI

2021
[21]

Vpp: visibility-based path planning heuristic for monitoring large regions of complex terrain using a uav onboard camera

Andres J Sanchez-Fernandez, Luis F Romero, Gerardo Bandera, and Siham Tabik. Vpp: visibility-based path planning heuristic for monitoring large regions of complex terrain using a uav onboard camera. IEEE Journal of Selected Topics in Applied Earth Observations and Remote Sensi...

2021
[22]

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

2017 arXiv
[23]

Enhancing conversational ai with llms for customer support automa- tion

Farooq Shareef. Enhancing conversational ai with llms for customer support automa- tion. In 2024 2nd International Conference on Self Sustainable Artificial Intelligence Systems (ICSSAS), pages 239–244. IEEE, 2024

2024
[24]

Correcting robot plans with natural language feedback

Pratyusha Sharma, Balakumar Sundaralingam, Valts Blukis, Chris Paxton, Tucker Her- mans, Antonio Torralba, Jacob Andreas, and Dieter Fox. Correcting robot plans with natural language feedback. arXiv preprint arXiv:2204.05186, 2022

2022 arXiv
[25]

Hierarchical interpretable imitation learning for end-to-end autonomous driving

Siyu Teng, Long Chen, Yunfeng Ai, Yuanye Zhou, Zhe Xuanyuan, and Xuemin Hu. Hierarchical interpretable imitation learning for end-to-end autonomous driving. IEEE Transactions on Intelligent Vehicles, 8(1):673–683, 2022

2022
[26]

Unmanned aerial vehicle for precision agriculture: A review

Francesco Toscano, Costanza Fiorentino, Nicola Capece, Ugo Erra, Danilo Travascia, Antonio Scopa, Marios Drosos, and Paola D’Antonio. Unmanned aerial vehicle for precision agriculture: A review. IEEE access, 2024

2024
[27]

A review on code generation with llms: Application and evaluation

Jianxun Wang and Yixiang Chen. A review on code generation with llms: Application and evaluation. In 2023 IEEE International Conference on Medical Artificial Intelli- gence (MedAI), pages 284–289. IEEE, 2023

2023
[28]

Reinforcement learning enhanced llms: A survey

Shuhe Wang, Shengyu Zhang, Jie Zhang, Runyi Hu, Xiaoya Li, Tianwei Zhang, Jiwei Li, Fei Wu, Guoyin Wang, and Eduard Hovy. Reinforcement learning enhanced llms: A survey. arXiv preprint arXiv:2412.10400, 2024

2024 arXiv
[29]

Coverage path planning for kiwifruit picking robots based on deep reinforcement learn- ing

Yinchu Wang, Zhi He, Dandan Cao, Li Ma, Kai Li, Liangsheng Jia, and Yongjie Cui. Coverage path planning for kiwifruit picking robots based on deep reinforcement learn- ing. Computers and Electronics in Agriculture, 205:107593, 2023

2023
[30]

Agv path planning using curiosity-driven deep reinforcement learning

Huilin Yin, Yinjia Lin, Jun Yan, Qian Meng, Karin Festl, Lukas Schichler, and Daniel Watzenig. Agv path planning using curiosity-driven deep reinforcement learning. In 2023 IEEE 19th International Conference on Automation Science and Engineering (CASE), pages 1–6. IEEE, 2023

2023
[31]

Language to rewards for robotic skill synthesis

Wenhao Yu, Nimrod Gileadi, Chuyuan Fu, Sean Kirmani, Kuang-Huei Lee, Montse Gonzalez Arenas, Hao-Tien Lewis Chiang, Tom Erez, Leonard Hasenclever, Jan Humplik, et al. Language to rewards for robotic skill synthesis. arXiv preprint arXiv:2306.08647, 2023

2023 arXiv
[32]

Deep reinforcement learning for next-best-view planning in agricultural applications

Xiangyu Zeng, Tobias Zaenker, and Maren Bennewitz. Deep reinforcement learning for next-best-view planning in agricultural applications. In 2022 International Conference on Robotics and Automation (ICRA), pages 2323–2329. IEEE, 2022

2022

Pith tools

Reviewed August 6, 2026 · model on record in the stance chip above.

[1] [1]

Tabnet: Attentive interpretable tabular learning

Sercan Ö Arik and Tomas Pfister. Tabnet: Attentive interpretable tabular learning. In Proceedings of the AAAI conference on artificial intelligence, volume 35, pages 6679– 6687, 2021

work page 2021

[2] [2]

Uav-based delivery systems: A systematic review, current trends, and research challenges

Francesco Betti Sorbelli. Uav-based delivery systems: A systematic review, current trends, and research challenges. Journal on Autonomous Transportation Systems, 1(3): 1–40, 2024

work page 2024

[3] [3]

Openai gym

Greg Brockman, Vicki Cheung, Ludwig Pettersson, Jonas Schneider, John Schulman, Jie Tang, and Wojciech Zaremba. Openai gym. arXiv preprint arXiv:1606.01540 , 2016

arXiv 2016

[4] [4]

Hierarchical model-based imitation learning for planning in autonomous driving

Eli Bronstein, Mark Palatucci, Dominik Notz, Brandyn White, Alex Kuefler, Yiren Lu, Supratik Paul, Payam Nikdel, Paul Mougin, Hongge Chen, et al. Hierarchical model-based imitation learning for planning in autonomous driving. In2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 8652–8659. IEEE, 2022

work page 2022

[5] [5]

Deep reinforcement learning for zero-shot coverage path planning with mobile robots

José Pedro Carvalho and A Pedro Aguiar. Deep reinforcement learning for zero-shot coverage path planning with mobile robots. IEEE/CAA Journal of Automatica Sinica, 2025

work page 2025

[6] [6]

Transformer-based imitative reinforcement learning for multirobot path planning

Lin Chen, Yaonan Wang, Zhiqiang Miao, Yang Mo, Mingtao Feng, Zhen Zhou, and Hesheng Wang. Transformer-based imitative reinforcement learning for multirobot path planning. IEEE Transactions on Industrial Informatics , 19(10):10233–10243, 2023

work page 2023

[7] [7]

A proximal policy optimization with curiosity algorithm for virtual drone navigation

Rupayan Das, Angshuman Khan, and Gunjan Paul. A proximal policy optimization with curiosity algorithm for virtual drone navigation. Engineering Research Express, 6 (1):015057, 2024

work page 2024

[8] [8]

BERT: Pre- training of deep bidirectional transformers for language understanding

Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. BERT: Pre- training of deep bidirectional transformers for language understanding. In Jill Burstein, Christy Doran, and Thamar Solorio, editors, Proceedings of the 2019 Conference of AUTHOR: VENKA T MARGAPURI 11 the North American Chapter of the Association for Computational Linguistics: Hu...

work page 2019