Pith. sign in

REVIEW 2 major objections 1 minor 42 references

Iterative RLHF refines an LLM to generate more expressive and fluid co-speech gestures on the Pepper robot.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.3

2026-06-26 20:47 UTC pith:EIC4GJSG

load-bearing objection The paper shows a workable LLM-plus-RLHF pipeline for Pepper gestures but supplies almost no quantitative evidence that the human feedback step actually improves anything. the 2 major comments →

arxiv 2606.18747 v1 pith:EIC4GJSG submitted 2026-06-17 cs.RO cs.AI

Generating Natural and Expressive Robot Gestures through Iterative Reinforcement Learning with Human Feedback using LLMs

Chris Lee , Flora Salim , Benjamin Tag , Francisco Cruz This is my paper
classification cs.RO cs.AI
keywords robot gesturesco-speech gesturesRLHFlarge language modelshuman-robot interactionPepper robotgesture generation
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper integrates ChatGPT with the Pepper humanoid to produce co-speech gestures from natural language input, but finds the initial outputs stiff and unnatural. It then applies an iterative reinforcement learning process that collects human ratings of the gestures and uses those ratings to update the generation. A sympathetic reader would care because social robots need movements that match conversational intent and physical constraints if they are to support long-term human interaction. The central mechanism is the closed loop of generation, user evaluation, and model adjustment that occurs without requiring hand-crafted animations.

Core claim

Integrating ChatGPT into Pepper enables flexible runtime gesture synthesis aligned with speech, yet the baseline motions are perceived as rigid. An iterative RLHF procedure that finetunes the model on successive rounds of user ratings produces gestures rated as more expressive, relevant, and fluid than the initial LLM output.

What carries the argument

The iterative RLHF loop that collects human preference ratings on generated gestures and uses them to update the LLM's gesture-generation behavior.

Load-bearing premise

Human ratings collected during the user studies supply an unbiased and consistent signal of naturalness that can steer the model without introducing new artifacts.

What would settle it

A blind rating study in which participants prefer the baseline LLM gestures over the RLHF-tuned ones at a statistically significant rate would falsify the claimed improvement.

Watch this falsifier. Get emailed when new claim-graph text bears on it.

Share X Bluesky LinkedIn Reddit HN

If this is right

  • Gesture generation can adapt at runtime to new conversational contexts without expert-authored libraries.
  • The same feedback loop can incorporate physical constraints of the robot platform directly into the model updates.
  • Long-term acceptance of social robots improves when movements are perceived as more natural and relevant.
  • The approach reduces dependence on static animation sets that cannot scale to diverse environments.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same iterative feedback structure could be applied to other robot platforms whose joint limits differ from Pepper.
  • Repeated user studies over time might allow the system to develop user-specific gesture styles.
  • If the rating interface can be made lightweight, the loop could run continuously during ordinary conversations rather than in dedicated studies.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

2 major / 1 minor

Summary. The manuscript describes integrating ChatGPT with the Pepper robot to generate co-speech gestures from natural language, starting from a baseline that produces stiff motions and then applying an iterative RLHF process that collects human evaluations across successive generations to refine the LLM's output, with the central claim that this yields more expressive, relevant, and fluid movements.

Significance. If the empirical improvement holds under controlled evaluation, the approach could enable scalable, context-adaptive gesture synthesis for social robots without expert-authored animations, addressing a practical limitation in HRI. The iterative human-in-the-loop design is a reasonable response to the difficulty of capturing perceived naturalness in high-DoF motion.

major comments (2)
  1. [Abstract] Abstract: the claim that 'RLHF improved the LLM's co-speech generative capabilities, producing more expressive, relevant and fluid movements' supplies no quantitative metrics, sample sizes, statistical tests, baseline descriptions, or effect sizes, so the data-to-claim link cannot be assessed.
  2. [User study / evaluation section (exact section number not visible in provided text)] Iterative user study description: no mention is made of blinding, counterbalancing, order randomization, fatigue controls, or inter-rater agreement statistics for the human ratings used as the reward signal; without these, the reported improvement risks being an artifact of the feedback collection process rather than a genuine enhancement of generative capability.
minor comments (1)
  1. [Abstract] Abstract: the transition from baseline LLM to RLHF-enhanced system could be stated more explicitly when presenting the results.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the detailed and constructive review. The comments identify important areas where the presentation of results and methodological details can be strengthened. We address each major comment below and will revise the manuscript accordingly.

read point-by-point responses

  1. Referee: [Abstract] Abstract: the claim that 'RLHF improved the LLM's co-speech generative capabilities, producing more expressive, relevant and fluid movements' supplies no quantitative metrics, sample sizes, statistical tests, baseline descriptions, or effect sizes, so the data-to-claim link cannot be assessed.

    Authors: We agree that the abstract would be strengthened by including quantitative support for the claim. The full manuscript reports a user study comparing baseline LLM-generated gestures against those refined through iterative RLHF, with human ratings collected on expressiveness, relevance, and fluidity. In the revision we will expand the abstract to report key metrics (mean rating improvements, participant count, number of evaluated gestures), explicitly name the baseline, and reference the statistical comparisons performed. revision: yes

  2. Referee: [User study / evaluation section (exact section number not visible in provided text)] Iterative user study description: no mention is made of blinding, counterbalancing, order randomization, fatigue controls, or inter-rater agreement statistics for the human ratings used as the reward signal; without these, the reported improvement risks being an artifact of the feedback collection process rather than a genuine enhancement of generative capability.

    Authors: We acknowledge that the current description of the iterative user study omits several standard controls. The study collected ratings from participants who were not told whether a gesture came from the baseline or RLHF model. In revision we will add an explicit subsection detailing the evaluation protocol: randomized presentation order across conditions, counterbalancing of gesture sets, session-length limits to reduce fatigue, and any available inter-rater agreement statistics (e.g., Krippendorff’s alpha or ICC). These additions will clarify that the reward signal was collected under controlled conditions. revision: yes

Circularity Check

0 steps flagged

No circularity: empirical results from iterative user study do not reduce to self-defined inputs

full rationale

The paper describes an LLM-based gesture generator for a robot, followed by an iterative RLHF process that collects human ratings in a user study and uses them to refine outputs. The central claim—that RLHF yields more expressive, relevant, and fluid movements—is presented as an empirical outcome measured in separate evaluations, not as a quantity defined by or fitted to the feedback signal itself. No equations, parameters, or derivations appear in the provided text. No self-citations are invoked as load-bearing uniqueness theorems or ansatzes. The method is self-contained against external benchmarks (human ratings collected during the study), satisfying the criteria for an honest non-finding of circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

No mathematical model or new theoretical constructs; the approach rests on standard LLM prompting and RLHF applied to an existing robot platform.

pith-pipeline@v0.9.1-grok · 5772 in / 970 out tokens · 24431 ms · 2026-06-26T20:47:29.143582+00:00 · methodology

0 comments
Cite this review

Pith. "Pith review of Generating Natural and Expressive Robot Gestures through Iterative Reinforcement Learning with Human Feedback using LLMs." pith.science (2026). https://pith.science/paper/EIC4GJSG

@misc{pith2026260618747,
  author       = {Pith},
  title        = {Pith review of: Generating Natural and Expressive Robot Gestures through Iterative Reinforcement Learning with Human Feedback using LLMs},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/EIC4GJSG}},
  note         = {Machine review of arXiv:2606.18747}
}
Share X Bluesky LinkedIn Reddit HN
read the original abstract

Expressive gestures are essential for natural and effective communication, complementing speech when verbal cues alone are insufficient (e.g., pointing). For social robots such as the humanoid Pepper, producing natural and expressive movements is critical for improving human-robot interaction (HRI) and long-term acceptance. However, generating gestures remains challenging due to reliance on expert-authored animations, resulting in rigid behaviors that are impractical for dynamic and diverse environments. Alternatively, machine learning approaches often struggle to capture perceived naturalness, becoming increasingly challenging with more degrees of freedom. Consequently, producing expressive robot gestures requires a system that can adapt to the environment while adhering to social norms and physical constraints. Recent advances in large language models (LLMs) enable dynamic code generation, offering new opportunities for runtime gesture synthesis from natural language. In this paper, we integrate ChatGPT into the humanoid robot Pepper to generate co-speech gestures aligned with conversational output. While this baseline enables flexible gesture generation, the resulting motions are often perceived as stiff and unnatural. To address this limitation, we introduce an iterative reinforcement learning with human feedback (RLHF) system that finetunes gesture generation based on user evaluations, leveraging an iterative user study to compare Pepper's generated gestures. Our results show that RLHF improved the LLM's co-speech generative capabilities, producing more expressive, relevant and fluid movements.

Figures

Figures reproduced from arXiv: 2606.18747 by Benjamin Tag, Chris Lee, Flora Salim, Francisco Cruz.

Figure 1
Figure 1. Figure 1: Overview of expressive co-speech gesture generation system with iterative reinforcement learning with human feedback. [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Pipeline from function library of primitive robotic motions to final execution. [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Perceived average expressiveness, relevance, and fluidity user ratings across categories per iteration. Although Iteration 1 shows a significant [PITH_FULL_IMAGE:figures/full_fig_p006_3.png] view at source ↗
Figure 4
Figure 4. Figure 4: Smoothed Error Rate per iteration during DPO finetuning. [PITH_FULL_IMAGE:figures/full_fig_p006_4.png] view at source ↗
Figure 5
Figure 5. Figure 5: Perceived average expressiveness, relevance, and fluidity user ratings across categories under ablation testing. [PITH_FULL_IMAGE:figures/full_fig_p007_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: Average primitive movement functions called per co-speech gesture [PITH_FULL_IMAGE:figures/full_fig_p007_6.png] view at source ↗

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

42 extracted references · 1 canonical work pages

  1. [1]

    Large language models for human–robot interaction: A review,

    C. Zhang, J. Chen, J. Li, Y . Peng, and Z. Mao, “Large language models for human–robot interaction: A review,”Biomimetic Intelligence and Robotics, vol. 3, no. 4, p. 100131, 2023

    2023

  2. [2]

    Generative Artificial Intelligence in Robotic Manipulation: A Survey,

    K. Zhang, P. Yun, J. Cen, J. Cai, D. Zhu, H. Yuan, C. Zhao, and et al, “Generative Artificial Intelligence in Robotic Manipulation: A Survey,” 2025

    2025

  3. [3]

    Large Language Models for Robotics: A Survey,

    F. Zeng, W. Gan, Y . Wang, N. Liu, and P. S. Yu, “Large Language Models for Robotics: A Survey,” 2023

    2023

  4. [4]

    Large language models for robotics: Opportunities, challenges, and perspectives,

    J. Wang, E. Shi, H. Hu, C. Ma, Y . Liu, X. Wang, Y . Yao, and et al, “Large language models for robotics: Opportunities, challenges, and perspectives,”Journal of Automation and Intelligence (2025), p. S2949855424000613, 2024

    2025

  5. [5]

    Speech gesture generation from the trimodal context of text, audio, and speaker identity,

    Y . Yoon, B. Cha, J.-H. Lee, M. Jang, J. Lee, J. Kim, and G. Lee, “Speech gesture generation from the trimodal context of text, audio, and speaker identity,”ACM Trans. Graph., vol. 39, no. 6, pp. 222:1– 222:16, 2020

    2020

  6. [6]

    Modeling and evaluating beat gestures for social robots,

    U. Zabala, I. Rodriguez, J. M. Mart ´ınez-Otzeta, and E. Lazkano, “Modeling and evaluating beat gestures for social robots,”Multimed Tools Appl, vol. 81, no. 3, pp. 3421–3438, 2022

    2022

  7. [7]

    Measurement In- struments for the Anthropomorphism, Animacy, Likeability, Perceived Intelligence, and Perceived Safety of Robots,

    C. Bartneck, D. Kuli ´c, E. Croft, and S. Zoghbi, “Measurement In- struments for the Anthropomorphism, Animacy, Likeability, Perceived Intelligence, and Perceived Safety of Robots,”Int J of Soc Robotics, vol. 1, no. 1, pp. 71–81, 2009

    2009

  8. [8]

    A primer for conducting experiments in human–robot interaction,

    G. Hoffman and X. Zhao, “A primer for conducting experiments in human–robot interaction,”ACM Transactions of Human-Robot Interaction, vol. 10, no. 1, 2021

    2021

  9. [9]

    Incremental learning of humanoid robot behavior from natural interaction and large language models,

    L. B ¨armann, R. Kartmann, F. Peller-Konrad, J. Niehues, A. Waibel, and T. Asfour, “Incremental learning of humanoid robot behavior from natural interaction and large language models,”Front. Robot. AI, vol. 11, 2024

    2024

  10. [10]

    Reinforcement Learning Approaches in Social Robotics,

    N. Akalin and A. Loutfi, “Reinforcement Learning Approaches in Social Robotics,”Sensors, vol. 21, no. 4, p. 1292, 2021

    2021

  11. [11]

    A Survey of Reinforcement Learning from Human Feedback,

    T. Kaufmann, P. Weng, V . Bengs, and E. H ¨ullermeier, “A Survey of Reinforcement Learning from Human Feedback,” 2025

    2025

  12. [12]

    Training language models to follow instructions with human feedback,

    L. Ouyang, J. Wu, X. Jiang, D. Almeida, C. L. Wainwright, P. Mishkin, et al., “Training language models to follow instructions with human feedback,” inProceedings of the 36th International Conference on Neural Information Processing Systems, NeurIPS ’22, (Red Hook, NY , USA), pp. 27730–27744, Curran Associates Inc., 2022

    2022

  13. [13]

    Generative Expressive Robot Behaviors using Large Language Models,

    K. Mahadevan, J. Chien, N. Brown, Z. Xu, C. Parada, F. Xia, A. Zeng, and et al, “Generative Expressive Robot Behaviors using Large Language Models,” inProceedings of the 2024 ACM/IEEE International Conference on Human-Robot Interaction, pp. 482–491, 2024

    2024

  14. [14]

    ChatGPT for Robotics: Design Principles and Model Abilities,

    S. Vemprala, R. Bonatti, A. Bucker, and A. Kapoor, “ChatGPT for Robotics: Design Principles and Model Abilities,” 2023

    2023

  15. [15]

    Eval- uating LLMs for Code Generation in HRI: A Comparative Study of ChatGPT, Gemini, and Claude,

    A. Sobo, A. Mubarak, A. Baimagambetov, and N. Polatidis, “Eval- uating LLMs for Code Generation in HRI: A Comparative Study of ChatGPT, Gemini, and Claude,”Applied Artificial Intelligence, vol. 39, no. 1, p. 2439610, 2025

    2025

  16. [16]

    The Impact of VR and 2D Interfaces on Human Feedback in Preference- Based Robot Learning,

    J. De Heuvel, D. Marta, S. Holk, I. Leite, and M. Bennewitz, “The Impact of VR and 2D Interfaces on Human Feedback in Preference- Based Robot Learning,” in2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 19024–19030, 2025

    2025

  17. [17]

    A Mass-Produced Sociable Humanoid Robot: Pepper: The First Machine of Its Kind,

    A. K. Pandey and R. Gelin, “A Mass-Produced Sociable Humanoid Robot: Pepper: The First Machine of Its Kind,”IEEE Robot. Automat. Mag., vol. 25, no. 3, pp. 40–48, 2018

    2018

  18. [18]

    Mehrabian,Nonverbal Communication

    A. Mehrabian,Nonverbal Communication. New York: Routledge, 2017

    2017

  19. [19]

    Effects of nonverbal communication on efficiency and robustness in human- robot teamwork,

    C. Breazeal, C. Kidd, A. Thomaz, G. Hoffman, and M. Berlin, “Effects of nonverbal communication on efficiency and robustness in human- robot teamwork,” in2005 IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 708–713, 2005

    2005

  20. [20]

    A human-centered approach to robot gesture based communication within collaborative working processes,

    T. Ende, S. Haddadin, S. Parusel, T. W ¨usthoff, M. Hassenzahl, and A. Albu-Sch ¨affer, “A human-centered approach to robot gesture based communication within collaborative working processes,” in2011 IEEE/RSJ International Conference on Intelligent Robots and Systems, pp. 3367–3374, 2011

    2011

  21. [21]

    Would you let a humanoid play storytelling with your child? A usability study on LLM-powered nar- rative Humanoid-Robot Interaction,

    M. Lombardi, C. Calabrese, D. Ghiglino, C. Foglino, D. De Tommaso, G. Da Lisca, L. Natale, and et al, “Would you let a humanoid play storytelling with your child? A usability study on LLM-powered nar- rative Humanoid-Robot Interaction,” in2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 20066– 20073, 2025

    2025

  22. [22]

    Synchronized gesture and speech production for humanoid robots,

    V . Ng-Thow-Hing, P. Luo, and S. Okita, “Synchronized gesture and speech production for humanoid robots,” in2010 IEEE/RSJ Interna- tional Conference on Intelligent Robots and Systems, pp. 4617–4624, 2010

    2010

  23. [23]

    Deep Reinforcement Learning with Interactive Feedback in a Hu- man–Robot Environment,

    I. Moreira, J. Rivas, F. Cruz, R. Dazeley, A. Ayala, and B. Fernandes, “Deep Reinforcement Learning with Interactive Feedback in a Hu- man–Robot Environment,”Applied Sciences, vol. 10, no. 16, p. 5574, 2020

    2020

  24. [24]

    Autonomous Robotic Reinforcement Learning with Asyn- chronous Human Feedback,

    M. Balsells, M. T. Villasevil, Z. Wang, S. Desai, P. Agrawal, and A. Gupta, “Autonomous Robotic Reinforcement Learning with Asyn- chronous Human Feedback,” inProceedings of The 7th Conference on Robot Learning, pp. 774–799, PMLR, 2023

    2023

  25. [25]

    Interactively shaping agents via human reinforcement: the TAMER framework,

    W. B. Knox and P. Stone, “Interactively shaping agents via human reinforcement: the TAMER framework,” inProceedings of the fifth international conference on Knowledge capture, K-CAP ’09, (New York, NY , USA), pp. 9–16, Association for Computing Machinery, 2009

    2009

  26. [26]

    Guiding a reinforcement learner with natural language advice: Initial results in RoboCup soccer,

    G. Kuhlmann, P. Stone, R. Mooney, and J. Shavlik, “Guiding a reinforcement learner with natural language advice: Initial results in RoboCup soccer,”AAAI Workshop - Technical Report, 2004

    2004

  27. [27]

    Deep reinforcement learning from human preferences,

    P. F. Christiano, J. Leike, T. B. Brown, M. Martic, S. Legg, and D. Amodei, “Deep reinforcement learning from human preferences,” inProceedings of the 31st International Conference on Neural In- formation Processing Systems, NeurIPS’17, (Red Hook, NY , USA), pp. 4302–4310, Curran Associates Inc., 2017

    2017

  28. [28]

    Affective Behavior Learning for Social Robot Haru with Implicit Evaluative Feedback,

    H. Wang, J. Lin, Z. Ma, Y . Vasylkiv, H. Brock, K. Nakamura, R. Gomez, and et al, “Affective Behavior Learning for Social Robot Haru with Implicit Evaluative Feedback,” in2022 IEEE/RSJ Interna- tional Conference on Intelligent Robots and Systems (IROS), pp. 3881– 3888, 2022

    2022

  29. [29]

    Shaping Progressive Net of Reinforcement Learning for Policy Transfer with Human Evaluative Feedback,

    R. Juan, J. Huang, R. Gomez, K. Nakamura, Q. Sha, B. He, and G. Li, “Shaping Progressive Net of Reinforcement Learning for Policy Transfer with Human Evaluative Feedback,” in2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 1281–1288, 2021

    2021

  30. [30]

    Data-Efficient Alignment of Large Language Models with Human Feedback Through Natural Language,

    D. Jin, S. Mehri, D. Hazarika, A. Padmakumar, S. Lee, Y . Liu, and M. Namazifar, “Data-Efficient Alignment of Large Language Models with Human Feedback Through Natural Language,” 2023

    2023

  31. [31]

    Survey on Large Language Model-Enhanced Reinforcement Learning: Concept, Taxonomy, and Methods,

    Y . Cao, H. Zhao, Y . Cheng, T. Shu, Y . Chen, G. Liu, G. Liang, and et al, “Survey on Large Language Model-Enhanced Reinforcement Learning: Concept, Taxonomy, and Methods,”IEEE Transactions on Neural Networks and Learning Systems, vol. 36, no. 6, pp. 9737–9757, 2025

    2025

  32. [32]

    Reinforcement Learning Enhanced LLMs: A Survey,

    S. Wang, S. Zhang, J. Zhang, R. Hu, X. Li, T. Zhang, J. Li, and et al, “Reinforcement Learning Enhanced LLMs: A Survey,”arXiv preprint arXiv:2412.10400, 2024

    work page arXiv 2024

  33. [33]

    Self-rewarding language models,

    W. Yuan, R. Y . Pang, K. Cho, X. Li, S. Sukhbaatar, J. Xu, and J. Weston, “Self-rewarding language models,” inProceedings of the 41st International Conference on Machine Learning, vol. 235 of ICML’24, (Vienna, Austria), pp. 57905–57923, JMLR.org, 2024

    2024

  34. [34]

    i-Sim2Real: Reinforce- ment Learning of Robotic Policies in Tight Human-Robot Interaction Loops,

    S. W. Abeyruwan, L. Graesser, D. B. D’Ambrosio, A. Singh, A. Shankar, A. Bewley, D. Jain, and et al, “i-Sim2Real: Reinforce- ment Learning of Robotic Policies in Tight Human-Robot Interaction Loops,” inProceedings of The 6th Conference on Robot Learning, pp. 212–224, PMLR, 2023

    2023

  35. [35]

    Skill Preferences: Learning to Extract and Execute Robotic Skills from Human Feedback,

    X. Wang, K. Lee, K. Hakhamaneshi, P. Abbeel, and M. Laskin, “Skill Preferences: Learning to Extract and Execute Robotic Skills from Human Feedback,” inProceedings of the 5th Conference on Robot Learning, pp. 1259–1268, PMLR, 2022

    2022

  36. [36]

    Learning to summarize from human feed- back,

    N. Stiennon, L. Ouyang, J. Wu, D. M. Ziegler, R. Lowe, C. V oss, A. Radford, and et al, “Learning to summarize from human feed- back,” inProceedings of the 34th International Conference on Neural Information Processing Systems, NeurIPS ’20, (Red Hook, NY , USA), pp. 3008–3021, Curran Associates Inc., 2020

    2020

  37. [37]

    Primitive Skill-Based Robot Learning from Human Evaluative Feedback,

    A. Hiranaka, M. Hwang, S. Lee, C. Wang, L. Fei-Fei, J. Wu, and R. Zhang, “Primitive Skill-Based Robot Learning from Human Evaluative Feedback,” in2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 7817–7824, 2023

    2023

  38. [38]

    Interactive Reinforcement Learning from Natural Language Feedback,

    I. Tarakli, S. Vinanzi, and A. D. Nuovo, “Interactive Reinforcement Learning from Natural Language Feedback,” in2024 IEEE/RSJ In- ternational Conference on Intelligent Robots and Systems (IROS), pp. 11478–11484, 2024

    2024

  39. [39]

    EMOTION: Expressive Motion Sequence Generation for Humanoid Robots With In-Context Learning,

    P. Huang, Y . Hu, N. Nechyporenko, D. Kim, W. Talbott, and J. Zhang, “EMOTION: Expressive Motion Sequence Generation for Humanoid Robots With In-Context Learning,”IEEE Robotics and Automation Letters, vol. 10, no. 8, pp. 7699–7706, 2025

    2025

  40. [40]

    Co-Speech Gesture Synthesis using Discrete Gesture Token Learning,

    S. Lu, Y . Yoon, and A. Feng, “Co-Speech Gesture Synthesis using Discrete Gesture Token Learning,” in2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 9808–9815, 2023

    2023

  41. [41]

    Gesture2Vec: Clustering Ges- tures using Representation Learning Methods for Co-speech Gesture Generation,

    P. J. Yazdian, M. Chen, and A. Lim, “Gesture2Vec: Clustering Ges- tures using Representation Learning Methods for Co-speech Gesture Generation,” in2022 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pp. 3100–3107, 2022

    2022

  42. [42]

    GPT-Driven Gestures: Leveraging Large Language Models to Generate Expressive Robot Motion for Enhanced Human-Robot Interaction,

    L. Roy, E. A. Croft, A. Ramirez, and D. Kuli ´c, “GPT-Driven Gestures: Leveraging Large Language Models to Generate Expressive Robot Motion for Enhanced Human-Robot Interaction,”IEEE Robotics and Automation Letters, vol. 10, no. 5, pp. 4172–4179, 2025

    2025