Pith. sign in

REVIEW 5 major objections 5 minor 52 references

Multi-Agent Systems for Robotic Autonomy with LLMs

T0 review · 5 major / 5 minor · reviewed 2026-08-15 · deepseek-v4-flash

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

Pith's one-line read Three chained LLM agents can convert a natural-language task description into a feasible simulated robot design and executable reinforcement-learning code, with DeepSeek-R1 as the most reliable backbone.

desk verdict A coherent three-agent LLM pipeline with an informative ablation, but the 'feasible robots' claim rests only on executing generated Gym code, not on geometric or hardware validation. read the letter →

arxiv 2505.05762 v1 pith:KIMVHYBK submitted 2025-05-09 cs.RO cs.AI

Junhong Chen , Ziqi Yang , Haoyuan G Xu , Dandan Zhang , George Mylonas This is my paper · ORCID
classification cs.ROcs.AI
keywords multi-agentsystemslargelanguagemodelsroboticautonomyrobotdesignreinforcementlearningpathgenerationDeepSeek-R1simulation
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

This paper argues that a complete early-stage robot development loop—task analysis, mechanism design, and controller training—can be automated by three chained LLM agents. A Task Analyst converts a natural-language prompt into a coordinate-framed task report, a Robot Designer selects base locations and arm lengths, and a Reinforcement Learning Designer writes the environment, training, and evaluation code that runs in the OpenAI Gym simulator. Across ten scenarios spanning industrial and medical settings and four language models, DeepSeek-R1 produced code that executed and converged in most cases, while the other models often failed at code execution or task alignment. The paper concludes that, given appropriately detailed inputs, LLM-based multi-agent systems can design feasible simulated robots with control strategies.

What carries the argument

The load-bearing mechanism is the three-agent linear workflow with report-to-report handoff. The Task Analyst extracts targets and base positions and establishes the coordinate frame; the Robot Designer applies economic and safety reasoning to choose arm lengths and assign subtasks; and the Reinforcement Learning Designer, described as the most crucial agent, selects an RL algorithm and writes the environment, training, and evaluation code. A code extractor pulls the code files out of the reports so they can be executed in the OpenAI Gym simulation environment. The paper attributes DeepSeek-R1's advantage within this machinery to its self-correction during reasoning, which lets it verify and refine its decisions before finalizing the design.

What would settle it

Take any of the ten published task scenarios, feed the exact input to the pipeline using DeepSeek-R1, and require the generated code to execute without any manual edits and the trained policy to reach every target point in a physics-based simulator that includes joint limits and obstacles. If the code needs hand-fixes or the controller fails to converge under those conditions, then the paper's feasibility claim is only supported at the level of script execution in a 2D environment.

Watch

Extended reading notes

Core claim

The paper's central discovery is that a linear chain of three LLM agents can carry a robotic task from a natural-language description to executed reinforcement-learning code without human intervention at intermediate steps, as long as the prompt specifies the environment, robot base options, target positions, and arm-length options. Each agent consumes the previous agent's report and emits a structured report of its own, and the RL Designer's three generated code files—environment definition, training script, and evaluation script—are extracted and executed. Under the paper's five metrics, DeepSeek-R1 scored highest on all of them, with means of 4.60 for task completion progress, 4.45 for code execution feasibility, 3.85 for model alignment, 3.88 for robot design adaptability, and 4.15 for report maturity, and it was the only model that sometimes chose SAC over PPO. The paper interprets this as showing that the framework generalizes across task domains, and that the main limiting factor is the reasoning ability of the underlying LLM.

Load-bearing premise

The claim that the system 'can design feasible robots with control strategies' rests on generated code running successfully in a 2D OpenAI Gym simulation, with no physical hardware, obstacles, joint limits, or dynamics tested, and no report of how many manual code fixes were required.

Editorial extensions

If this is right

  • If the central claim holds, a user with no robotics programming experience can go from a natural-language task description to a simulated robot design and a working reinforcement-learning controller without writing code by hand.
  • The framework's reliability is bounded by the reasoning strength of the backbone LLM: DeepSeek-R1 reached convergent controllers in most scenarios, while weaker models produced non-executable or non-convergent code on many tasks.
  • The ablation results imply that the Task Analyst's high-level summary is load-bearing: removing it destabilizes all downstream generation, even for DeepSeek-R1, and removing the RL Designer leaves the system with no control logic at all.
  • Input specification matters as much as model choice: both very short and very long task descriptions degraded performance relative to the normal-length description, so prompt design is part of the system's effectiveness.
  • The modular report handoff means individual agents or backbone models can be swapped, making the framework a usable testbed for comparing LLMs on engineering reasoning and code generation.

Reading between the lines

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

  • An implication the paper leaves implicit is that 'feasible' currently means executable in a 2D simulation; running the same generated designs in a rigid-body physics engine with joint torque limits and obstacles would be a materially stricter test that the paper has not performed.
  • The Robot Designer's choices could be validated independently of training: for each scenario, one can compute analytically whether every target point lies inside the reachable workspace of the chosen base and arm-length combination, which the current human-scored adaptability metric does not guarantee.
  • The observation that only DeepSeek-R1 varied between SAC and PPO suggests a testable extension: tie RL algorithm selection to explicit task features such as continuous action spaces, and measure whether that rule-based choice beats what the LLM selects from text alone.
  • The framework could be extended with a verification agent that executes the generated code, checks convergence, and flags failures before the final report is produced; the present pipeline does not include such a feedback loop.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

5 major / 5 minor

Summary. The paper presents a modular multi-agent framework in which a Task Analyst, Robot Designer, and Reinforcement Learning Designer are chained to convert a natural-language task specification (with base positions and arm-length options) into a task analysis report, a robot design report, and executable RL code for a 2D OpenAI Gym environment. Two auxiliary extractor agents package code and reports. The authors evaluate GPT-4o-mini, DeepSeek-V3, GPT-4o, and DeepSeek-R1 on ten industrial/medical scenarios using five self-defined 0-5 metrics (TCP, CEF, MA, RDA, RM) and report that DeepSeek-R1 outperforms the others, with a mean MA of 3.85 versus 1.00-1.55 for the other models. They also report ablations that remove one or two core agents and vary task-description length. The paper claims the system can design feasible robots with control strategies when adequate task inputs are provided.

Significance. The framework is clearly described, and the comparison of four LLMs plus the ablation design are useful elements for the LLM-for-robotics community. If the central claim is re-scoped to 'generating executable RL code in a 2D simulator for some models,' the study provides a reasonable proof-of-concept and a comparison point for future work. The main weaknesses are that feasibility is not checked at the mechanism level, the metrics are not fully operationalized, no code or logs are released, and no single-agent baseline is included. These issues block acceptance of the broad feasibility claim but are addressable in a revision.

major comments (5)
  1. [§3.1, §3.3, §4.4] The abstract and conclusion state that the system 'can design feasible robots with control strategies,' but the evaluation never verifies the designed mechanism's feasibility. The Robot Designer's output (base location and arm lengths) is passed directly to the RL Designer, and the only execution evidence is that the generated env.py, train.py, and eval.py run in OpenAI Gym. A code run can succeed even when the chosen arm cannot reach the assigned targets, because the reward function is also generated by the model. I ask for an explicit, automatic reachability check per scenario (e.g., comparing each assigned target's distance from the selected base with the workspace of the selected arm) and per-case pass/fail counts, or, alternatively, the claims should be reworded to 'code-execution feasibility in 2D simulation.' This is load-bearing because the phrase 'feasible robot' is the central claim.
  2. [§4.4, Tables 3-5] The metrics MA and CEF are presented as objective but are not operationally defined. 'Meets task requirements' and 'without convergence' have no stated thresholds, and CEF counts code that runs after 'minor fixes' as feasible without reporting how many fixes were required. RDA and RM are explicitly subjective, but the paper does not report inter-rater agreement or a rubric; the average of four researchers' scores is given without variance among raters. Without these details and without the underlying generated reports and code, the scores in Tables 3-5 cannot be independently checked or reproduced.
  3. [Abstract, §5.1, Table 4] The aggregate claim of generalizability is not supported by the data. Table 4 shows mean MA = 1.00, 1.10, and 1.55 for GPT-4o-mini, DeepSeek-V3, and GPT-4o, respectively, versus 3.85 for DeepSeek-R1. The abstract's unqualified statement that the system can design feasible robots holds, on the evidence presented, only for DeepSeek-R1 in the simulated setting. The conclusion partially acknowledges this, but the abstract and Section 5 should be aligned, and the claims should be scoped by model and by simulation-only validation.
  4. [§4.3.2, §2] The paper does not compare the multi-agent pipeline with a single-agent baseline. The ablation study removes agents from the proposed pipeline but does not include a condition where one LLM receives the entire task description and directly produces the robot design and RL code. Without such a baseline, the reported benefit of the three-agent decomposition cannot be separated from the benefit of a more detailed prompt or a stronger model. I recommend adding this condition or softening the framing from 'multi-agent systems' to 'multi-stage prompting.'
  5. [§3.2, §4] Reproducibility is a concern: no code, prompt templates, generated reports, or training logs are released, and the prompts in Section 3.2 are only summarized as bullet-point responsibilities. The authors should release at least the exact prompt templates, one complete run per model, and the scoring rubrics (or include them as supplementary material) so that Tables 3-5 can be audited.
minor comments (5)
  1. [Table 1] The caption of Table 1 states that Scenarios 1-5 are industrial and 6-10 are medical, but the rows are the reverse: rows 1-5 (rehabilitation, surgical instruments, elderly feeding, physical therapy, prosthetic limb) are medical, and rows 6-10 (assembly, warehouse, welding, pick-and-place, palletizing) are industrial.
  2. [Tables 3-4] Tables 3 and 4 appear to be duplicated in the submitted text; please remove the duplicate blocks and ensure each table caption is unique.
  3. [§2.1] There are grammatical errors in Section 2, for example 'image reading receptively' should be 'image reading respectively,' and the sentence 'robotic systems now walking into a future from a step over single-function demonstrations' is unclear and should be rewritten.
  4. [§6] In Section 6, 'DeepSeek-R1 keeps a higher quality output with quite a few mistakes' appears to mean 'with few mistakes'; the current wording contradicts 'higher quality output.'
  5. [§5.3] The paper mentions example output figures from code execution but no such figure is included in the provided text; adding at least one learning curve, motor control plot, and end-effector trajectory would strengthen the presentation.

Circularity Check

0 steps flagged · score 0.0 of 10

No circular derivation: the central claims are empirical evaluations of an LLM pipeline, not quantities defined in terms of themselves.

full rationale

The paper does not contain a mathematical derivation chain. The claimed contribution is a three-agent LLM pipeline that converts task descriptions into a design report, RL code, and simulated trajectories, and the evidence is an evaluation campaign using scoring rubrics (TCP, CEF, MA, RDA, RM). None of these metrics is defined as a function of the output it is used to validate; they are human- or execution-based ratings of whether code runs and whether results align with task requirements. The closest potential concern is the abstract's word 'feasible robots,' which is supported only by code running in OpenAI Gym (§3.3) and by subjective author-scored metrics such as RDA (§4.4). That is a limitation in the strength of the empirical evidence, not a circular step: the conclusion is not assumed in the inputs, and the metrics are not fitted parameters renamed as predictions. Self-citations such as [15], [18], [21], [22], [51], and [52] are used for background on RL path generation and prior related systems; none is load-bearing in the sense of supplying a uniqueness theorem, an ansatz, or a definition that forces the reported outcomes. No equation in the paper reduces to its own input, and no fitted quantity is later called a prediction. The finding is therefore 'no significant circularity,' with the caveat that feasibility is demonstrated only at the level of simulated kinematics and script execution rather than physical or geometric verification.

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

The paper introduces no fitted numerical parameters and no new physical entities. Its claims rest on domain assumptions about LLM competence, geometric feasibility, simulation fidelity, and subjective scoring, all of which are stated or implicit in the method section.

assumptions (5)
  • domain assumption LLMs can reliably extract structured engineering information from task descriptions and generate executable RL code with minor fixes.
    Invoked throughout Section 3; CEF scores below 5 for most models show code often needed fixes, so this assumption is only partially met and is load-bearing.
  • domain assumption A planar arm's feasibility can be reduced to 2D geometric reachability of target points from the selected base with selected link lengths.
    The Robot Designer (Section 3.2.2) selects arms so that each robot can reach all assigned target points; obstacles, joint limits, and dynamics are ignored.
  • domain assumption OpenAI Gym simulation is a valid proxy for real robot task performance.
    Section 3.3 uses OpenAI Gym as the standard benchmark; conclusions about feasible robots are drawn entirely from simulation without hardware validation.
  • domain assumption Subjective scores for RDA and RM provided by four researchers are reliable and free of bias.
    Section 4.4 says four researchers independently score these metrics and the average is used; no inter-rater reliability or external audit is reported.
  • domain assumption The Code Extractor faithfully reconstructs the LLM's intended env.py, train.py, and eval.py, and any manual fixes do not change the design.
    Section 3.2.4 describes the extractor but does not report the frequency or size of manual fixes needed to make the code run.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Multi-Agent Systems for Robotic Autonomy with LLMs." pith.science (2026). https://pith.science/paper/KIMVHYBK

@misc{pith2026250505762,
  author       = {Pith},
  title        = {Pith review of: Multi-Agent Systems for Robotic Autonomy with LLMs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/KIMVHYBK}},
  note         = {Machine review of arXiv:2505.05762}
}
read the original abstract

Since the advent of Large Language Models (LLMs), various research based on such models have maintained significant academic attention and impact, especially in AI and robotics. In this paper, we propose a multi-agent framework with LLMs to construct an integrated system for robotic task analysis, mechanical design, and path generation. The framework includes three core agents: Task Analyst, Robot Designer, and Reinforcement Learning Designer. Outputs are formatted as multimodal results, such as code files or technical reports, for stronger understandability and usability. To evaluate generalizability comparatively, we conducted experiments with models from both GPT and DeepSeek. Results demonstrate that the proposed system can design feasible robots with control strategies when appropriate task inputs are provided, exhibiting substantial potential for enhancing the efficiency and accessibility of robotic system development in research and industrial applications.

Figures

Figures reproduced from arXiv: 2505.05762 by the authors.

Figure 1
Figure 1. The overview of the proposed multi-agent robot system [PITH_FULL_IMAGE:figures/full_fig_p003_1.png] view at source ↗
Figure 2
Figure 2. The detailed design of LLMs-based Multi-Agent Systems in the Framework [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

52 extracted references · 42 canonical work pages

  1. [1]

    Improving language understanding by gen- erative pre-training

    Alec Radford, Karthik Narasimhan, Tim Salimans, Ilya Sutskever, et al. Improving language understanding by gen- erative pre-training. 2018. 1

    2018

  2. [2]

    Llama: Open and efficient foundation language models

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

    arXiv 2023

  3. [3]

    Con- stitutional ai: Harmlessness from ai feedback.arXiv preprint arXiv:2212.08073, 2022

    Yuntao Bai, Saurav Kadavath, Sandipan Kundu, Amanda Askell, Jackson Kernion, Andy Jones, Anna Chen, Anna Goldie, Azalia Mirhoseini, Cameron McKinnon, et al. Con- stitutional ai: Harmlessness from ai feedback.arXiv preprint arXiv:2212.08073, 2022. 1

    arXiv 2022

  4. [4]

    Gpt-driven gestures: Leveraging large language mod- els to generate expressive robot motion for enhanced human- robot interaction.IEEE Robotics and Automation Letters,

    Liam Roy, Elizabeth A Croft, Alex Ramirez, and Dana Kuli´c. Gpt-driven gestures: Leveraging large language mod- els to generate expressive robot motion for enhanced human- robot interaction.IEEE Robotics and Automation Letters,

    work page

  5. [5]

    Natural multimodal fusion-based human–robot in- teraction: Application with voice and deictic posture via large language model.IEEE Robotics & Automation Maga- zine, 2025

    Yuzhi Lai, Shenghai Yuan, Youssef Nassar, Mingyu Fan, At- maraaj Gopal, Arihiro Yorita, Naoyuki Kubota, and Matthias R¨atsch. Natural multimodal fusion-based human–robot in- teraction: Application with voice and deictic posture via large language model.IEEE Robotics & Automation Maga- zine, 2025

    work page 2025

  6. [6]

    Llm for generating simulation inputs to evaluate path planning algo- rithms

    Chenyang Wang, Jonathan Diller, and Qi Han. Llm for generating simulation inputs to evaluate path planning algo- rithms. In2024 International Conference on Machine Learn- ing and Applications (ICMLA), pages 176–181. IEEE, 2024. 1

    work page 2024

  7. [7]

    Sensingagent: Advancing vehicular sens- ing systems for spatiotemporal cognitive intelligence.IEEE Transactions on Intelligent Vehicles, 2024

    Yuhang Liu, Yutong Wang, Yuhang Li, Chaoyue Dai, and Fei-Yue Wang. Sensingagent: Advancing vehicular sens- ing systems for spatiotemporal cognitive intelligence.IEEE Transactions on Intelligent Vehicles, 2024. 1

    work page 2024

  8. [8]

    Tingting Yang, Ping Feng, Qixin Guo, Jindi Zhang, Jia- hong Ning, Xinghan Wang, and Zhongyang Mao. Autohma- llm: Efficient task coordination and execution in heteroge- neous multi-agent systems using hybrid large language mod- els.IEEE Transactions on Cognitive Communications and Networking, 2025

    work page 2025

Show all 52 references
  1. [9]

    A dual-agent collaboration framework based on llms for nursing robots to perform bi- manual coordination tasks.IEEE Robotics and Automation Letters, 2025

    Zhendong Zhao, Xiaotian Yue, Jiexin Xie, Chuanhong Fang, Zhenzhou Shao, and Shijie Guo. A dual-agent collaboration framework based on llms for nursing robots to perform bi- manual coordination tasks.IEEE Robotics and Automation Letters, 2025. 1

    2025

  2. [10]

    Human-level control through deep reinforcement learn- ing.nature, 518(7540):529–533, 2015

    V olodymyr Mnih, Koray Kavukcuoglu, David Silver, An- drei A Rusu, Joel Veness, Marc G Bellemare, Alex Graves, Martin Riedmiller, Andreas K Fidjeland, Georg Ostrovski, et al. Human-level control through deep reinforcement learn- ing.nature, 518(7540):529–533, 2015. 1

    2015

  3. [11]

    Deep learning, reinforcement learning, and world models.Neural Networks, 152:267–275, 2022

    Yutaka Matsuo, Yann LeCun, Maneesh Sahani, Doina Pre- cup, David Silver, Masashi Sugiyama, Eiji Uchibe, and Jun Morimoto. Deep learning, reinforcement learning, and world models.Neural Networks, 152:267–275, 2022. 1

    2022

  4. [12]

    Huihui Sun, Hui Jiang, Long Zhang, Changlin Wu, and Sen Qian. Multi-robot hierarchical safe reinforcement learn- ing autonomous decision-making strategy based on uni- formly ultimate boundedness constraints.Scientific Reports, 15(1):5990, 2025. 1

    2025

  5. [13]

    Autonomous environment-adaptive microrobot swarm navigation enabled by deep learning- based real-time distribution planning.Nature Machine In- telligence, 4(5):480–493, 2022

    Lidong Yang, Jialin Jiang, Xiaojie Gao, Qinglong Wang, Qi Dou, and Li Zhang. Autonomous environment-adaptive microrobot swarm navigation enabled by deep learning- based real-time distribution planning.Nature Machine In- telligence, 4(5):480–493, 2022. 1

    2022

  6. [14]

    Robotic motion planning based on deep reinforce- ment learning and artificial neural networks.IEEE Transac- tions on Automation Science and Engineering, 2024

    Huashan Liu, Xiangjian Li, Menghua Dong, Yuqing Gu, and Bo Shen. Robotic motion planning based on deep reinforce- ment learning and artificial neural networks.IEEE Transac- tions on Automation Science and Engineering, 2024. 1

    2024

  7. [15]

    Path generation with rein- forcement learning for surgical robot control

    Junhong Chen, Zeyu Wang, Ruiqi Zhu, Ruiyang Zhang, Weibang Bai, and Benny Lo. Path generation with rein- forcement learning for surgical robot control. In2022 IEEE- EMBS International Conference on Biomedical and Health Informatics (BHI), pages 1–4. IEEE, 2022. 1, 2

    2022

  8. [16]

    Llm- augmented symbolic rl with landmark-based task decompo- sition

    Alireza Kheirandish, Duo Xu, and Faramarz Fekri. Llm- augmented symbolic rl with landmark-based task decompo- sition. InICASSP 2025-2025 IEEE International Conference on Acoustics, Speech and Signal Processing (ICASSP), pages 1–5. IEEE, 2025. 1

    2025

  9. [17]

    Guiding pretraining in reinforcement learning with large language models

    Yuqing Du, Olivia Watkins, Zihan Wang, C ´edric Colas, Trevor Darrell, Pieter Abbeel, Abhishek Gupta, and Jacob Andreas. Guiding pretraining in reinforcement learning with large language models. InInternational Conference on Ma- chine Learning, pages 8657–8677. PMLR, 2023

    2023

  10. [18]

    Tactile perception: a biomimetic whisker-based method for clinical gastrointesti- nal diseases screening.npj Robotics, 1(1):3, 2023

    Zeyu Wang, Frank P-W Lo, Yunran Huang, Junhong Chen, James Calo, Wei Chen, and Benny Lo. Tactile perception: a biomimetic whisker-based method for clinical gastrointesti- nal diseases screening.npj Robotics, 1(1):3, 2023. 1

    2023

  11. [19]

    Language-guided pattern formation for swarm robotics with multi-agent reinforcement learning

    Hsu-Shen Liu, So Kuroki, Tadashi Kozuno, Wei-Fang Sun, and Chun-Yi Lee. Language-guided pattern formation for swarm robotics with multi-agent reinforcement learning. In2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 8998–9005. IEEE, 2024. 1

    2024

  12. [20]

    Talker: A task-activated language model based knowledge-extension reasoning system.IEEE Robotics and Automation Letters, 2024

    Jiabin Lou, Rongye Shi, Yuxin Lin, Qunbo Wang, and Wen- jun Wu. Talker: A task-activated language model based knowledge-extension reasoning system.IEEE Robotics and Automation Letters, 2024. 1

    2024

  13. [21]

    An ai-driven bionic whisker system assisting for clinical gas- trointestinal disease screening

    Zeyu Wang, Frank P-W Lo, Junhong Chen, James Calo, Benny PL Lo, Alex J Thompson, and Eric M Yeatman. An ai-driven bionic whisker system assisting for clinical gas- trointestinal disease screening. In2024 International Joint Conference on Neural Networks (IJCNN), pages 1–8. IEEE,

  14. [22]

    An intelligent robotic endoscope control system based on fusing natural language processing and vision models

    Beili Dong, Junhong Chen, Zeyu Wang, Kaizhong Deng, Yiping Li, Benny Lo, and George Mylonas. An intelligent robotic endoscope control system based on fusing natural language processing and vision models. In2024 IEEE In- ternational Conference on Robotics and Automation (ICRA),...

    2024

  15. [23]

    Natural language controlled real-time object recognition framework for household robot

    Ruchi Bagwe, Rashika Natharani, Kiran George, and Anand Panangadan. Natural language controlled real-time object recognition framework for household robot. In2021 IEEE 11th Annual Computing and Communication Workshop and Conference (CCWC), pages 1215–1220, 2021

    2021

  16. [24]

    PhD thesis, Brac University, 2024

    Golam Dastagir, Khaled Mushahed Hossain, Md Mahfujul Haque, and Raiyan Bin Gaffar.NLP-based AI robot for in- telligent interaction. PhD thesis, Brac University, 2024. 2

    2024

  17. [25]

    Tod4ir: A humanised task-oriented dialogue system for industrial robots.IEEE Access, 10:91631–91649,

    Chen Li, Xiaochun Zhang, Dimitrios Chrysostomou, and Hongji Yang. Tod4ir: A humanised task-oriented dialogue system for industrial robots.IEEE Access, 10:91631–91649,

  18. [26]

    Gpt-4v(ision) for robotics: Multimodal task planning from human demonstration.IEEE Robotics and Automation Letters, 9(11):10567–10574, 2024

    Naoki Wake, Atsushi Kanehira, Kazuhiro Sasabuchi, Jun Takamatsu, and Katsushi Ikeuchi. Gpt-4v(ision) for robotics: Multimodal task planning from human demonstration.IEEE Robotics and Automation Letters, 9(11):10567–10574, 2024. 2

    2024

  19. [27]

    Shyam Sundar Kannan, Vishnunandan L. N. Venkatesh, and Byung-Cheol Min. Smart-llm: Smart multi-agent robot task planning using large language models. In2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 12140–12147, 2024. 2

    2024

  20. [28]

    A review of multi-agent mobile robot systems applications.International Journal of Elec- trical and Computer Engineering, 12(4):3517–3529, 2022

    Ammar Abdul Ameer Rasheed, Mohammed Najm Abdullah, and Ahmed Sabah Al-Araji. A review of multi-agent mobile robot systems applications.International Journal of Elec- trical and Computer Engineering, 12(4):3517–3529, 2022. 2

    2022

  21. [29]

    Palm-e: An embodied multimodal language model

    Danny Driess, Fei Xia, Mehdi SM Sajjadi, Corey Lynch, Aakanksha Chowdhery, Ayzaan Wahid, Jonathan Tompson, Quan Vuong, Tianhe Yu, Wenlong Huang, et al. Palm-e: An embodied multimodal language model. 2023. 2

    2023

  22. [30]

    Are we close to realizing self-programming robots that overcome the unexpected? In2025 IEEE/SICE International Symposium on System Integration (SII), pages 368–374

    Jair A Bottega, Takashi Tsubouchi, Xinyue Ruan, and Ak- ihisa Ohya. Are we close to realizing self-programming robots that overcome the unexpected? In2025 IEEE/SICE International Symposium on System Integration (SII), pages 368–374. IEEE, 2025. 2

    2025

  23. [31]

    Auto- matic milp model construction for multi-robot task alloca- tion and scheduling based on large language models.arXiv preprint arXiv:2503.13813, 2025

    Mingming Peng, Zhendong Chen, Jie Yang, Jin Huang, Zhengqi Shi, Qihao Liu, Xinyu Li, and Liang Gao. Auto- matic milp model construction for multi-robot task alloca- tion and scheduling based on large language models.arXiv preprint arXiv:2503.13813, 2025. 2

    2025 arXiv

  24. [32]

    Twostep: Multi-agent task planning using classical planners and large language models.arXiv preprint arXiv:2403.17246, 2024

    Ishika Singh, David Traum, and Jesse Thomason. Twostep: Multi-agent task planning using classical planners and large language models.arXiv preprint arXiv:2403.17246, 2024. 2

    2024 arXiv

  25. [33]

    Autotamp: Autoregressive task and motion planning with llms as translators and check- ers

    Yongchao Chen, Jacob Arkin, Charles Dawson, Yang Zhang, Nicholas Roy, and Chuchu Fan. Autotamp: Autoregressive task and motion planning with llms as translators and check- ers. In2024 IEEE International Conference on Robotics and Automation (ICRA), pages 6695–6702, 2024

    2024

  26. [34]

    Isr-llm: Iterative self-refined large language model for long-horizon sequential task planning

    Zhehua Zhou, Jiayang Song, Kunpeng Yao, Zhan Shu, and Lei Ma. Isr-llm: Iterative self-refined large language model for long-horizon sequential task planning. In2024 IEEE In- ternational Conference on Robotics and Automation (ICRA), pages 2081–2088, 2024

    2024

  27. [35]

    Roco: Dialec- tic multi-robot collaboration with large language models

    Zhao Mandi, Shreeya Jain, and Shuran Song. Roco: Dialec- tic multi-robot collaboration with large language models. In 2024 IEEE International Conference on Robotics and Au- tomation (ICRA), pages 286–299. IEEE, 2024. 2

    2024

  28. [36]

    Yongchao Chen, Jacob Arkin, Yang Zhang, Nicholas Roy, and Chuchu Fan. Scalable multi-robot collaboration with large language models: Centralized or decentralized sys- tems? In2024 IEEE International Conference on Robotics and Automation (ICRA), pages 4311–4317. IEEE, 2024

    2024

  29. [37]

    Under- standing large-language model (llm)-powered human-robot interaction

    Callie Y Kim, Christine P Lee, and Bilge Mutlu. Under- standing large-language model (llm)-powered human-robot interaction. InProceedings of the 2024 ACM/IEEE inter- national conference on human-robot interaction, pages 371– 380, 2024. 2

    2024

  30. [38]

    Large language models for multi-robot systems: A survey.arXiv preprint arXiv:2502.03814, 2025

    Peihan Li, Zijian An, Shams Abrar, and Lifeng Zhou. Large language models for multi-robot systems: A survey.arXiv preprint arXiv:2502.03814, 2025. 2

    2025 arXiv

  31. [39]

    Double-dqn based path smoothing and tracking control method for robotic vehi- cle navigation.Computers and Electronics in Agriculture, 166:104985, 2019

    Wenyu Zhang, Jingyao Gai, Zhigang Zhang, Lie Tang, Qingxi Liao, and Youchun Ding. Double-dqn based path smoothing and tracking control method for robotic vehi- cle navigation.Computers and Electronics in Agriculture, 166:104985, 2019. 2

    2019

  32. [40]

    Deep reinforcement learning for decentralized multi-robot control: A dqn approach to robust- ness and information integration

    Bin Wu and C Steve Suh. Deep reinforcement learning for decentralized multi-robot control: A dqn approach to robust- ness and information integration. InASME International Mechanical Engineering Congress and Exposition, volume 88636, page V005T07A035. American Society of Mecha...

    2024

  33. [41]

    A3c based motion learning for an autonomous mobile robot in crowds

    Yoko Sasaki, Syusuke Matsuo, Asako Kanezaki, and Hiroshi Takemura. A3c based motion learning for an autonomous mobile robot in crowds. In2019 IEEE International Confer- ence on Systems, Man and Cybernetics (SMC), pages 1036– 1042, 2019. 2

    2019

  34. [42]

    Deep reinforcement learning with enhanced ppo for safe mobile robot navigation.arXiv preprint arXiv:2405.16266, 2024

    Hamid Taheri, Seyed Rasoul Hosseini, and Mohammad Ali Nekoui. Deep reinforcement learning with enhanced ppo for safe mobile robot navigation.arXiv preprint arXiv:2405.16266, 2024. 2

    2024 arXiv

  35. [43]

    Obstacle avoidance control method for robotic assembly process based on lagrange ppo

    Weixin Quan, Wenbo Zhu, Qinghua Lu, Lufeng Luo, Kai Wang, and Meng Liu. Obstacle avoidance control method for robotic assembly process based on lagrange ppo. InInterna- tional Conference on Cognitive Systems and Signal Process- ing, pages 16–26. Springer, 2023

    2023

  36. [44]

    Towards hardware accelerated reinforcement learning for application-specific robotic control

    Shengjia Shao, Jason Tsai, Michal Mysior, Wayne Luk, Thomas Chau, Alexander Warren, and Ben Jeppesen. Towards hardware accelerated reinforcement learning for application-specific robotic control. In2018 IEEE 29th In- ternational Conference on Application-specific Systems, Ar- ...

    2018

  37. [45]

    Rac-sac: An improved actor-critic algorithm for continuous multi-task manipulation on robot arm control

    Phuc Dang Thi, Chinh Nguyen Truong, and Hieu Dau Sy. Rac-sac: An improved actor-critic algorithm for continuous multi-task manipulation on robot arm control. InProceed- ings of the 12th International Symposium on Information and Communication Technology, pages 824–830, 2023. 2

    2023

  38. [46]

    Yolo-marl: You only llm once for multi-agent rein- forcement learning.arXiv preprint arXiv:2410.03997, 2024

    Yuan Zhuang, Yi Shen, Zhili Zhang, Yuxiao Chen, and Fei Miao. Yolo-marl: You only llm once for multi-agent rein- forcement learning.arXiv preprint arXiv:2410.03997, 2024. 2

    2024 arXiv

  39. [47]

    Mpc-based admittance control for robotic manipulators

    Arne Wahrburg and Kim Listmann. Mpc-based admittance control for robotic manipulators. In2016 IEEE 55th Con- ference on Decision and Control (CDC), pages 7548–7554,

  40. [48]

    Mpc for robot manipulators with integral sliding modes generation.IEEE/ASME Transactions on Mechatronics, 22(3):1299–1307, 2017

    Gian Paolo Incremona, Antonella Ferrara, and Lalo Magni. Mpc for robot manipulators with integral sliding modes generation.IEEE/ASME Transactions on Mechatronics, 22(3):1299–1307, 2017. 2

    2017

  41. [49]

    Imitation learning for autonomous tra- jectory learning of robot arms in space.arXiv preprint arXiv:2008.04007, 2020

    RB Shyam, Zhou Hao, Umberto Montanaro, and Ger- hard Neumann. Imitation learning for autonomous tra- jectory learning of robot arms in space.arXiv preprint arXiv:2008.04007, 2020. 2

    2008 arXiv

  42. [50]

    Pedro Aguiar

    Jos ´e Pedro Carvalho and A. Pedro Aguiar. Deep reinforce- ment learning for zero-shot coverage path planning with mo- bile robots.IEEE/CAA Journal of Automatica Sinica, pages 1–16, 2025. 2

    2025

  43. [51]

    Automated trajec- tory generation for robotic surgical tasks

    Ziqi Yang, Ruiyang Zhang, Junhong Chen, Xuhui Zhou, Yunxiao Ren, Ziyue Tong, and Benny Lo. Automated trajec- tory generation for robotic surgical tasks. In2024 Interna- tional Conference on Advanced Robotics and Mechatronics (ICARM), pages 39–44. IEEE, 2024. 2

    2024

  44. [52]

    A step towards conditional autonomy-robotic appendectomy.IEEE Robotics and Au- tomation Letters, 8(5):2429–2436, 2023

    Ruiyang Zhang, Junhong Chen, Zeyu Wang, Ziqi Yang, Yunxiao Ren, Peilun Shi, James Calo, Kyle Lam, Sanjay Purkayastha, and Benny Lo. A step towards conditional autonomy-robotic appendectomy.IEEE Robotics and Au- tomation Letters, 8(5):2429–2436, 2023. 2

    2023

Pith tools

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