pith. machine review for the scientific record. sign in

arxiv: 2604.18331 · v1 · submitted 2026-04-20 · 💻 cs.RO

Recognition: unknown

Will People Enjoy a Robot Trainer? A Case Study with Snoopie the Pacerbot

Maximilian Du , Jennifer Grannen , Shuran Song , Dorsa Sadigh
Authors on Pith no claims yet

Pith reviewed 2026-05-10 03:59 UTC · model grok-4.3

classification 💻 cs.RO
keywords robot trainerquadruped robotinterval traininguser studypace adherencephysical embodimentwearable comparison
0
0 comments X

The pith

A quadruped robot trainer improves runners' pace adherence by 60 percent and earns strong user preference over wearable devices.

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

The paper tests whether a robot quadruped can function as an effective personal trainer for interval running, a task that requires precise timing and sustained consistency over long sessions. It runs a user study that pits the physical robot against a common wearable device to measure differences in performance and user experience. Results indicate markedly higher adherence to target paces and steadier speeds when the robot is present. Participants also rated the robot higher on ease of use, enjoyment, and helpfulness. A reader would care because the work points to physical robots as a practical way to support better exercise habits without relying solely on screens or straps.

Core claim

SNOOPIE, an autonomous robot quadruped pacer, delivers interval training exercises matched to each user's abilities and produces 60.6 percent better adherence to the pace schedule along with 45.9 percent greater consistency in running speeds than a wearable trainer. Users expressed clear preference for the robot on measures of ease of use, interaction enjoyability, and overall helpfulness.

What carries the argument

SNOOPIE, the autonomous quadruped pacer that physically runs with the user and supplies real-time guidance and feedback through its embodied presence.

Load-bearing premise

The gains in adherence, consistency, and preference stem from the robot's physical body rather than from its newness or from choices made in how the study was run.

What would settle it

A controlled follow-up study in which participants first become familiar with the robot through repeated sessions and then compare it again to the wearable device; if the advantages shrink or vanish, the embodiment claim weakens.

Figures

Figures reproduced from arXiv: 2604.18331 by Dorsa Sadigh, Jennifer Grannen, Maximilian Du, Shuran Song.

Figure 1
Figure 1. Figure 1: Benefits of a Robot Trainer. Much like a real personal trainer, SNOOPIE the Pacerbot leverages its embodied nature to guide runners at set paces that are customized to the runners’ physical abilities. The quadruped’s physicality leads to more accurate pacing exercises and greater perceived enjoyment compared to existing wearable technology. We hypothesize that the unique combination of physical embodiment … view at source ↗
Figure 2
Figure 2. Figure 2: Snoopie the Pacerbot. The pace-training robot is a Unitree Go2 quadruped fitted with two fisheye cameras for user interface and autonomous navigation (A). The pace training interaction has two phases (B): an initial speed calibration and the robot-led interval training exercise (§III-C). During speed calibration (C), the robot adjusts its speed to find a comfortable user pace. During interval training (D),… view at source ↗
Figure 3
Figure 3. Figure 3: User Study Results. We track the participants’ speeds during the pacing exercises to compute pacing performance and report the average and standard error across all three methods. On average, pacing with SNOOPIE led to lower error on target pace (A) and lower overall speed variance (B). An example of a participant’s data is shown on the right (C). Notably, the user speed transitions during interval changes… view at source ↗
Figure 4
Figure 4. Figure 4: Extended Interaction Session Results. Across five sessions, the participant spent 112.7 minutes with SNOOPIE and ran a total of 11.0 kilometers (A). An excerpt of one session is displayed in (B), which shows his changing speeds as the robot brings him through the interval exercise. Throughout his extended interaction, his session-to-session qualitative ratings stayed relatively constant (C), a promising in… view at source ↗
read the original abstract

The physicality of exercise makes the role of athletic trainers unique. Their physical presence allows them to guide a student through a motion, demonstrate an exercise, and give intuitive feedback. Robot quadrupeds are also embodied agents with robust agility and athleticism. In our work, we investigate whether a robot quadruped can serve as an effective and enjoyable personal trainer device. We focus on a case study of interval training for runners: a repetitive, long-horizon task where precision and consistency are important. To meet this challenge, we propose SNOOPIE, an autonomous robot quadruped pacer capable of running interval training exercises tailored to challenge a user's personal abilities. We conduct a set of user experiments that compare the robot trainer to a wearable trainer device--the Apple Watch--to investigate the benefits of a physical embodiment in exercise-based interactions. We demonstrate 60.6% better adherence to a pace schedule and were 45.9% more consistent across their running speeds with the quadruped trainer. Subjective results also showed that participants strongly preferred training with the robot over wearable devices across many qualitative axes, including its ease of use (+56.7%), enjoyability of the interaction (+60.6%), and helpfulness (+39.1%). Additional videos and visualizations can be found on our website: https://sites.google.com/view/snoopie

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

1 major / 1 minor

Summary. The manuscript introduces SNOOPIE, an autonomous quadruped robot pacer for interval running training. It reports a user study comparing the robot trainer to an Apple Watch wearable device, claiming 60.6% better adherence to a pace schedule, 45.9% higher consistency in running speeds, and strong subjective preferences for the robot on ease of use (+56.7%), enjoyability (+60.6%), and helpfulness (+39.1%).

Significance. If the results hold after addressing experimental controls, this case study provides concrete evidence that embodied robotic trainers can improve adherence and user experience in repetitive physical tasks compared to wearables. It contributes to human-robot interaction research by combining quantitative performance metrics with qualitative preference data and includes supporting videos on a project website.

major comments (1)
  1. [User Experiments] User Experiments section: The comparison to the Apple Watch confounds physical embodiment with feedback modality. SNOOPIE provides continuous visual pacing and real-time following by running alongside the user, while the Apple Watch supplies only discrete notifications. No control condition (e.g., a non-embodied continuous visual pacer) isolates the embodiment effect, so the reported 60.6% adherence and 45.9% consistency gains cannot be securely attributed to the robot's physical presence rather than the specific trainer behavior. This directly undermines the central claim about benefits of embodiment.
minor comments (1)
  1. [Abstract] Abstract: The reported percentages would be more interpretable if participant count, statistical tests, and study protocol details were included.

Simulated Author's Rebuttal

1 responses · 0 unresolved

We thank the referee for their constructive comments on our manuscript. We provide a point-by-point response to the major comment below and outline the revisions we will make to address the concerns raised.

read point-by-point responses

  1. Referee: The comparison to the Apple Watch confounds physical embodiment with feedback modality. SNOOPIE provides continuous visual pacing and real-time following by running alongside the user, while the Apple Watch supplies only discrete notifications. No control condition (e.g., a non-embodied continuous visual pacer) isolates the embodiment effect, so the reported 60.6% adherence and 45.9% consistency gains cannot be securely attributed to the robot's physical presence rather than the specific trainer behavior. This directly undermines the central claim about benefits of embodiment.

    Authors: We acknowledge that the experimental design compares SNOOPIE's continuous visual pacing and following behavior to the Apple Watch's discrete notifications, which does confound the effects of physical embodiment with the specific feedback modality. Our study is presented as a case study evaluating a robot trainer against a common wearable device used for pacing in running training. While the robot's embodiment enables the continuous, adaptive following that contributes to the observed improvements in adherence and consistency, we agree that we cannot definitively attribute the gains solely to embodiment without additional control conditions. In the revised manuscript, we will clarify the scope of our claims to focus on the practical benefits of the embodied robot trainer in comparison to the wearable baseline, rather than claiming a pure isolation of embodiment effects. We will add a discussion of this limitation and suggest that future work could include controls such as a non-embodied continuous visual pacer to further isolate the contribution of physical presence. This revision will ensure our conclusions are appropriately qualified. revision: partial

Circularity Check

0 steps flagged

No circularity: direct empirical study with no derivations or models

full rationale

The paper is a case study reporting user experiment results on adherence, consistency, and subjective preference when comparing a quadruped robot trainer to an Apple Watch. It contains no equations, models, fitted parameters, predictions, or first-principles derivations that could reduce to inputs by construction. Claims rest on direct measurements from participants rather than any self-referential chain, self-citation load-bearing argument, or renamed known result. The skeptic concern about study confounds is a validity issue, not circularity.

Axiom & Free-Parameter Ledger

0 free parameters · 0 axioms · 0 invented entities

This is an empirical human-robot interaction study with no mathematical model, derivations, or theoretical constructs. No free parameters, axioms, or invented entities are introduced.

pith-pipeline@v0.9.0 · 5549 in / 1054 out tokens · 39369 ms · 2026-05-10T03:59:25.745289+00:00 · methodology

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

37 extracted references · 1 canonical work pages

  1. [1]

    Assessment of the acceptability and feasibility of using mobile robotic systems for patient evaluation,

    P. R. Chai, F. Z. Dadabhoy, H. Huanget al., “Assessment of the acceptability and feasibility of using mobile robotic systems for patient evaluation,”JAMA Network Open, vol. 4, no. 3, p. e210667, 2021

    2021

  2. [2]

    Co-designing robot dogs with and for neurodivergent individuals: Opportunities and challenges,

    H.-K. Kong, D. Xie, A. Chandra, R. Lowy, A. F. Maignan, S. Ha, C. H. Park, and J. G. Kim, “Co-designing robot dogs with and for neurodivergent individuals: Opportunities and challenges,” inThe 26th International ACM SIGACCESS Conference on Computers and Accessibility. ACM, pp. 1–15

  3. [3]

    Towards robotic companions: Understanding handler-guide dog interactions for informed guide dog robot design,

    H. Hwang, H.-T. Jung, N. A. Giudice, J. Biswas, S. I. Lee, and D. Kim, “Towards robotic companions: Understanding handler-guide dog interactions for informed guide dog robot design,” inProceedings of the 2024 CHI Conference on Human Factors in Computing Systems, ser. CHI ’24. Association for Computing Machinery, pp. 1–20

    2024

  4. [4]

    Hardware companions? what online aibo discussion forums reveal about the human-robotic relationship,

    B. Friedman, P. H. Kahn, and J. Hagman, “Hardware companions? what online aibo discussion forums reveal about the human-robotic relationship,” inProceedings of the SIGCHI Conference on Human Factors in Computing Systems, ser. CHI ’03. New York, NY , USA: Association for Computing Machinery, 2003, p. 273–280

    2003

  5. [5]

    [Online]

    How to get better at pacing (and why it’s worth the effort). [Online]. Available: https://stories.strava.com/articles/ how-to-get-better-at-run-pacing-why-its-worth-the-effort

  6. [6]

    Efficient model learning from joint-action demonstrations for human-robot collaborative tasks,

    S. Nikolaidiset al., “Efficient model learning from joint-action demonstrations for human-robot collaborative tasks,”ACM/IEEE HRI, 2014

    2014

  7. [7]

    V ocal sandbox: Continual learning and adaptation for situated human- robot collaboration,

    J. Grannen, S. Karamcheti, S. Mirchandani, P. Liang, and D. Sadigh, “V ocal sandbox: Continual learning and adaptation for situated human- robot collaboration,” inConference on Robot Learning, 2024

    2024

  8. [8]

    Provox: Person- alization and proactive planning for situated human-robot collaboration,

    J. Grannen, S. Karamcheti, B. Wulfe, and D. Sadigh, “Provox: Person- alization and proactive planning for situated human-robot collaboration,” IEEE Robotics and Automation Letters (RA-L), 2025

    2025

  9. [9]

    Learning from unscripted deictic gesture and language for human-robot interactions

    C. Matuszek, L. Bo, L. Zettlemoyer, and D. Fox, “Learning from unscripted deictic gesture and language for human-robot interactions.” inProceedings of the AAAI Conf. on Artificial Intelligence, vol. 28, 2014

    2014

  10. [10]

    Gesture-informed robot assistance via foundation models,

    L.-H. Lin, Y . Cui, Y . Hao, F. Xia, and D. Sadigh, “Gesture-informed robot assistance via foundation models,” inConference on Robot Learning, 2023

    2023

  11. [11]

    Dqn-tamer: Human-in-the-loop reinforcement learning with intractable feedback,

    R. Arakawa, S. Kobayashi, Y . Unno, Y . Tsuboi, and S.-i. Maeda, “Dqn-tamer: Human-in-the-loop reinforcement learning with intractable feedback,” inWorkshop on Human-Robot Teaming Beyond Human Operational Speeds and Robot Teammates Operating in Dynamic Unstructured Environments, May 2019, iCRA 2019 Workshop

    2019

  12. [12]

    Robotic system for physical training of older adults,

    O. Avioz-Sarig, S. Olatunji, V . Sarne-Fleischmann, and Y . Edan, “Robotic system for physical training of older adults,”International Journal of Social Robotics, vol. 13, no. 5, pp. 1109–1124, 2021

    2021

  13. [13]

    Technical feasibility of supervision of stretching exercises by a humanoid robot coach for chronic low back pain: The r-cool randomized trial,

    A. Blanchard, S. M. Nguyen, M. Devanne, M. Simonnet, M. Le Goff- Pronost, and O. Rémy-Néris, “Technical feasibility of supervision of stretching exercises by a humanoid robot coach for chronic low back pain: The r-cool randomized trial,”BioMed Research International, p. 5667223, 2022

    2022

  14. [14]

    A socially assistive robot exercise coach for the elderly,

    J. Fasola and M. J. Matari ´c, “A socially assistive robot exercise coach for the elderly,”J. Hum.-Robot Interact., vol. 2, no. 2, p. 3–32, 2013

    2013

  15. [15]

    Couch to 5km robot coach: An autonomous, human- trained socially assistive robot,

    K. Winkle, S. Lemaignan, P. Caleb-Solly, U. Leonards, A. Turton, and P. Bremner, “Couch to 5km robot coach: An autonomous, human- trained socially assistive robot,” inACM/IEEE International Conference on Human-Robot Interaction, Mar. 2020, pp. 520–522

    2020

  16. [16]

    My fitness coach is ai: unraveling the influential mechanisms of ai coach-user-human coach interactions on users’ behavioral intention,

    L. Wang, Q. Xue, and J. Xue, “My fitness coach is ai: unraveling the influential mechanisms of ai coach-user-human coach interactions on users’ behavioral intention,” inInternational Conference on Computer Vision, Robotics, and Automation Engineering, vol. 13249, 2024, p. 44

    2024

  17. [17]

    Artificial intelligence-powered social robots for promoting physical activity in older adults: A systematic review,

    J. Shen, J. Yu, H. Zhang, M. A. Lindsey, and R. An, “Artificial intelligence-powered social robots for promoting physical activity in older adults: A systematic review,”Journal of Sport and Health Science, vol. 14, p. 101045, Dec. 2025

    2025

  18. [18]

    Socially assistive robot for providing recommendations: Comparing a humanoid robot with a mobile application,

    S. Rossi, M. Staffa, and A. Tamburro, “Socially assistive robot for providing recommendations: Comparing a humanoid robot with a mobile application,”International Journal of Social Robotics 10, p. 265–278, 2018

    2018

  19. [19]

    A motivational approach to support healthy habits in long-term child–robot interaction,

    R. Ros, E. Oleari, C. Pozzi, F. Sacchitelli, D. Baranzini, A. Bagherzad- halimi, A. Sanna, and Y . Demiris, “A motivational approach to support healthy habits in long-term child–robot interaction,”International Journal of Social Robotics, 2016

    2016

  20. [20]

    How service robots can improve workplace experience: Camaraderie, customization, and humans-in-the-loop,

    Y .-L. Tsai, C. Wadgaonkar, B. Chun, and H. Knight, “How service robots can improve workplace experience: Camaraderie, customization, and humans-in-the-loop,”International Journal of Social Robotics 14.7, pp. 1605–1624, 2022

    2022

  21. [21]

    Roreas: Robot coach for walking and orientation training in clinical post-stroke reha- bilitation—prototype implementation and evaluation in field trials,

    H. M. Gross, A. Scheidig, K. Debeset al., “Roreas: Robot coach for walking and orientation training in clinical post-stroke reha- bilitation—prototype implementation and evaluation in field trials,” Autonomous Robots, vol. 41, pp. 679–698, 2017

    2017

  22. [22]

    Personalizing exoskeleton assistance while walking in the real world,

    P. Sladeet al., “Personalizing exoskeleton assistance while walking in the real world,”Nature, vol. 610, pp. 277–282, 2022

    2022

  23. [23]

    Human-in-the-loop optimization of exoskeleton assistance during walking,

    J. Zhanget al., “Human-in-the-loop optimization of exoskeleton assistance during walking,”Science, vol. 356, no. 6344, pp. 1280– 1284, 2017

    2017

  24. [24]

    Preference-based learning for exoskeleton gait optimization

    M. Tucker, E. Novoseller, C. Kann, Y . Sui, Y . Yue, J. Burdick, and A. D. Ames, “Preference-based learning for exoskeleton gait optimization.” [Online]. Available: http://arxiv.org/abs/1909.12316

    work page arXiv 1909

  25. [25]

    [Online]

    Stay fit with apple watch. [Online]. Available: https://support.apple. com/guide/watch/stay-fit-with-apple-watch-apd9c3cfe913/watchos

  26. [26]

    [Online]

    Fitbit activity trackers & smartwatches. [Online]. Available: https://store.google.com/category/watches_trackers?hl=en-US

  27. [27]

    [Online]

    Dashboard | strava. [Online]. Available: https://www.strava.com/ dashboard

  28. [28]

    The effect of presence on human-robot interaction,

    W. A. Bainbridge, J. Hart, E. S. Kim, and B. Scassellati, “The effect of presence on human-robot interaction,”The IEEE Int. symposium on robot and human interactive communication, p. 701–706, 2008

    2008

  29. [29]

    Understanding expectations for a robotic guide dog for visually impaired people,

    J. T. Kim, M. Byrd, J. L. Crandell, B. N. Walker, G. Turk, and S. Ha, “Understanding expectations for a robotic guide dog for visually impaired people,” inACM/IEEE International Conference on Human- Robot Interaction, 2025, pp. 262–271

    2025

  30. [30]

    Aibo: Towards the era of digital creatures,

    M. Fujita, “Aibo: Towards the era of digital creatures,” inRobotics Research, J. M. Hollerbach and D. E. Koditschek, Eds. London: Springer London, 2000, pp. 315–320

    2000

  31. [31]

    B. \. K. McKay. The japanese 3x3 interval walking workout. [Online]. Available: https://www.artofmanliness.com/health-fitness/ fitness/the-japanese-3x3-interval-walking-workout/

  32. [32]

    How adaptation, training, and customization contribute to benefits from exoskeleton assistance,

    K. L. Poggensee and S. H. Collins, “How adaptation, training, and customization contribute to benefits from exoskeleton assistance,” Science Robotics, vol. 6, no. 58, p. 1078, 2021

    2021

  33. [33]

    1 Mile Run Times By Age And Ability - Running Level

    “1 Mile Run Times By Age And Ability - Running Level.” [Online]. Available: https://runninglevel.com/running-times/1-mile-times

  34. [34]

    The OpenCV Library,

    G. Bradski, “The OpenCV Library,”Dr. Dobb’s Journal of Software Tools, 2000

    2000

  35. [35]

    Automatic generation and detection of highly reliable fiducial markers under occlusion,

    S. Garrido-Jurado, R. Muñoz-Salinas, F. Madrid-Cuevas, and M. Marín- Jiménez, “Automatic generation and detection of highly reliable fiducial markers under occlusion,”Pattern Recognition, vol. 47, no. 6, pp. 2280– 2292, 2014

    2014

  36. [36]

    What Are the Benefits of Interval Running?

    “What Are the Benefits of Interval Running?” [Online]. Available: https://www.nike.com/a/interval-training-running-benefits

  37. [37]

    The system usability scale: Past, present, and future,

    J. R. Lewis, “The system usability scale: Past, present, and future,” International Journal of Human-Computer Interaction, vol. 34, no. 7, pp. 577–590, 2018

    2018