pith. sign in

arxiv: 2603.22560 · v3 · submitted 2026-03-23 · 💻 cs.RO

Allometric Scaling Laws for Bipedal Robots

Naomi Oke , Aja M. Carter , Ben Gu , Steven Man , Cordelia Pride , Sarah Bergbreiter , Aaron M. Johnson This is my paper

Pith reviewed 2026-05-15 00:02 UTC · model grok-4.3

classification 💻 cs.RO
keywords allometric scalingbipedal robotslegged locomotiondynamic similaritytorque scalingrobot designscaling laws
0
0 comments X

The pith

Bipedal robot mass scales with the square of leg length rather than the cube of isometric scaling.

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

Bipedal robots deviate from both biological allometry and simple geometric scaling when size changes. A review of existing designs shows mass growing proportionally to leg length squared across three orders of magnitude. Controlled simulations of three quasi-passive hip-actuated walkers confirm that walking speed still follows the square-root-of-length rule from dynamic similarity. Minimum torque instead tracks mass times length, not the mass-times-length-squared relation that isometric models predict. Foot geometry scaled linearly with leg length preserves performance in the simulations.

Core claim

Across literature data and scaled simulations, robot mass scales approximately with L squared, walking velocity with L to the one-half, and minimum required torque with m times L, while foot geometry scales with L to the one.

What carries the argument

Empirical power-law fits to literature robot data combined with dynamic simulation of scaled quasi-passive walkers in Drake.

If this is right

  • Robot designers can estimate required mass and actuator torque for a target leg length using the fitted exponents instead of isometric assumptions.
  • Walking speed continues to increase with the square root of leg length even when mass does not follow the cubic rule.
  • Proportional scaling of foot geometry with leg length maintains consistent walking performance in the tested designs.
  • Torque requirements grow more slowly with size than isometric models suggest, altering power and actuator sizing at larger scales.

Where Pith is reading between the lines

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

  • If the L-squared mass scaling holds, very large bipeds may require less structural material per unit length than small ones, changing material and fabrication economics.
  • The torque scaling result suggests that actuator power density demands do not increase as steeply with size as previously assumed for legged machines.
  • Extending the same measurement protocol to other actuation schemes or terrain types would test whether the observed exponents are design-specific or general.

Load-bearing premise

The three simulated walker variants and the collected literature data points are representative of general bipedal robots.

What would settle it

A new survey of real bipedal robots across a wide range of leg lengths that yields a mass-length exponent statistically different from two.

Figures

Figures reproduced from arXiv: 2603.22560 by Aaron M. Johnson, Aja M. Carter, Ben Gu, Cordelia Pride, Naomi Oke, Sarah Bergbreiter, Steven Man.

Figure 1
Figure 1. Figure 1: Two of the robots considered in this paper, the 15cm [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Figure 2: Relationships between body length, leg length, and body mass across multiple existing robot morphologies. [PITH_FULL_IMAGE:figures/full_fig_p003_2.png] view at source ↗
Figure 3
Figure 3. Figure 3: Mugatu (Top) and Zippy (Bottom) robots in the Drake [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figure 5
Figure 5. Figure 5: Simulated minimum required torque of Zippy and [PITH_FULL_IMAGE:figures/full_fig_p005_5.png] view at source ↗
Figure 6
Figure 6. Figure 6: X, Y, Z axis lengths that resulted in stable walking in Mugatu and Zippy. [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
Figure 7
Figure 7. Figure 7: Peak velocity, found by optimizing along the X, Y, Z axis, across scales in Mugatu and Zippy. This is compared [PITH_FULL_IMAGE:figures/full_fig_p007_7.png] view at source ↗
read the original abstract

Scaling the design of robots up or down remains a fundamental challenge. While biological systems follow well-established isometric and allometric scaling laws relating mass, stride frequency, velocity, and torque, it is unclear how these relationships translate to robotic systems. In this paper, we generate similar allometric scaling laws for bipedal robots across three orders of magnitude in leg length. First, we conduct a review of legged robots from the literature and extract empirical relationships between leg length (L), body length, mass, and speed. These data show that robot mass scales more closely to L^2, in contrast to the L^3 scaling predicted by isometric scaling. We then perform controlled simulation studies in Drake using three variants of real quasi-passive, hip-actuated walkers with different foot geometries and control strategies. We evaluate the performance of each design scaled with leg length, L. Across all robots, walking velocity follows the expected L^(1/2) trend from dynamic similarity. Minimum required torque scales more closely with m*L than the isometric model of m*L^2. Foot geometry scaled proportionally with L^1. These results provide new insight into how robot designs allometrically scale to different sizes, and how that scaling is different from isometric or biological scaling laws.

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

2 major / 3 minor

Summary. The paper claims that bipedal robots follow distinct allometric scaling laws: literature data show mass scaling approximately as L^2 (vs. isometric L^3), while Drake simulations of three scaled quasi-passive hip-actuated walker variants confirm walking velocity scaling as L^{1/2} per dynamic similarity, minimum torque scaling as m*L (vs. isometric m*L^2), and foot geometry scaling proportionally with L across three orders of magnitude in leg length.

Significance. If the reported trends hold, the work is significant for providing empirical, non-circular scaling relations for robot design that deviate from both isometric models and biological allometry; the combination of independent literature extraction and controlled simulation runs (with consistent velocity, mass, and torque trends) offers practical guidance for scaling bipedal systems efficiently.

major comments (2)
  1. [§4] §4 (simulation studies): the claim that minimum torque scales more closely with m*L than m*L^2 is load-bearing for the central allometric result, but the exact definition and extraction of 'minimum required torque' from the Drake runs (e.g., peak actuator torque or steady-state value) is not fully specified, making it difficult to assess sensitivity to control parameters or foot geometry variants.
  2. [Literature review] Literature data section: the mass ~ L^2 conclusion rests on extracted points from prior robots, yet no statistical measure (R², p-value, or comparison to L^3 fit) is provided to quantify 'more closely,' and the number of data points or selection criteria are not detailed, weakening support for the deviation from isometric scaling.
minor comments (3)
  1. [Abstract] Abstract and §3: the specific leg-length range (e.g., 0.2 m to 2 m) used for the three-order-of-magnitude scaling should be stated explicitly to allow readers to judge the extrapolation.
  2. [Figures] Figure 2 or scaling plots: axis labels for torque (units and normalization) and mass should match the text definitions to improve clarity and avoid misinterpretation of the m*L relation.
  3. [Discussion] The discussion of dynamic similarity would benefit from an explicit citation to Alexander (1984) or similar foundational work on Froude-number scaling for legged locomotion.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive comments and the recommendation for minor revision. We address each major comment below and have revised the manuscript to improve methodological clarity and quantitative support for the claims.

read point-by-point responses

  1. Referee: [§4] §4 (simulation studies): the claim that minimum torque scales more closely with m*L than m*L^2 is load-bearing for the central allometric result, but the exact definition and extraction of 'minimum required torque' from the Drake runs (e.g., peak actuator torque or steady-state value) is not fully specified, making it difficult to assess sensitivity to control parameters or foot geometry variants.

    Authors: We agree that the precise definition and extraction method for minimum required torque should be stated explicitly. In the Drake simulations this quantity is the peak value of the hip-actuator torque command during the steady-state portion of the gait cycle (after discarding the initial transient). We have added a dedicated paragraph in §4 that describes the post-processing step, including the exact time window used and a short sensitivity check confirming that the reported m*L scaling persists under modest variations in control gains and foot parameters. This clarification does not alter the scaling results but makes the procedure fully reproducible. revision: yes

  2. Referee: [Literature review] Literature data section: the mass ~ L^2 conclusion rests on extracted points from prior robots, yet no statistical measure (R², p-value, or comparison to L^3 fit) is provided to quantify 'more closely,' and the number of data points or selection criteria are not detailed, weakening support for the deviation from isometric scaling.

    Authors: We accept that the original text did not supply statistical quantification or selection details. We have expanded the literature-review section to list all extracted robots together with their sources, to state the inclusion criteria (peer-reviewed or technical reports on bipedal platforms with published leg length and mass), and to report the results of log-log linear regressions. The revised text now includes R² values for the mass-versus-L relationship under both the L² and L³ models, allowing readers to evaluate the relative goodness of fit directly. revision: yes

Circularity Check

0 steps flagged

No significant circularity

full rationale

The paper's central claims derive from two independent sources: (1) empirical extraction of mass, length, and speed data from published legged-robot literature, and (2) controlled Drake simulations of three distinct quasi-passive hip-actuated walker variants scaled across leg lengths. Neither the reported L^2 mass scaling nor the L^{1/2} velocity and m L torque scalings reduce to a fitted parameter that is then renamed as a prediction, nor to any self-citation chain or ansatz smuggled from prior work by the same authors. The dynamic-similarity baseline for velocity is invoked as an external reference, not derived internally. The derivation chain therefore remains self-contained against external benchmarks.

Axiom & Free-Parameter Ledger

2 free parameters · 1 axioms · 0 invented entities

The central claims rest on empirical fitting of scaling exponents from reviewed robot data and simulation outputs under the assumption that dynamic similarity governs velocity; no new entities are postulated and no additional free parameters beyond the observed trends are introduced.

free parameters (2)
  • mass scaling exponent = approximately 2
    Fitted from extracted literature data on robot leg length and mass showing closer fit to power of 2 than 3.
  • torque scaling relation = m*L
    Determined from simulation results comparing minimum torque to m*L versus m*L^2.
axioms (1)
  • domain assumption Dynamic similarity principles govern walking velocity scaling in legged systems
    Invoked to predict and confirm L^(1/2) velocity trend across scaled robots.

pith-pipeline@v0.9.0 · 5537 in / 1380 out tokens · 57247 ms · 2026-05-15T00:02:41.651343+00:00 · methodology

discussion (0)

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

Lean theorems connected to this paper

Citations machine-checked in the Pith Canon. Every link opens the source theorem in the public Lean library.

  • IndisputableMonolith/Cost/FunctionalEquation.lean washburn_uniqueness_aczel unclear
    ?
    unclear

    Relation between the paper passage and the cited Recognition theorem.

    robot mass scales more closely to L^2, in contrast to the L^3 scaling predicted by isometric scaling... Minimum required torque scales more closely with m*L than the isometric model of m*L^2. Walking velocity follows the expected L^(1/2) trend from dynamic similarity.

What do these tags mean?
matches
The paper's claim is directly supported by a theorem in the formal canon.
supports
The theorem supports part of the paper's argument, but the paper may add assumptions or extra steps.
extends
The paper goes beyond the formal theorem; the theorem is a base layer rather than the whole result.
uses
The paper appears to rely on the theorem as machinery.
contradicts
The paper's claim conflicts with a theorem or certificate in the canon.
unclear
Pith found a possible connection, but the passage is too broad, indirect, or ambiguous to say the theorem truly supports the claim.

Reference graph

Works this paper leans on

36 extracted references · 36 canonical work pages

  1. [1]

    R. M. Alexander,Principles of Animal Locomotion. Princeton University Press, 2003

    work page 2003

  2. [2]

    T. A. McMahon,Muscles, Reflexes, and Locomotion. Princeton University Press, 1984

    work page 1984

  3. [3]

    Size and shape in biology: Elastic criteria impose limits on biological proportions, and consequently on metabolic rates

    ——, “Size and shape in biology: Elastic criteria impose limits on biological proportions, and consequently on metabolic rates.”Science, vol. 179, no. 4079, pp. 1201–1204, 1973

    work page 1973

  4. [4]

    Elastic and/or geometric similarity in mammalian design?

    A. C. Economos, “Elastic and/or geometric similarity in mammalian design?”Journal of Theoretical Biology, vol. 103, no. 1, 1983

    work page 1983

  5. [5]

    On the origin of biological similarity,

    ——, “On the origin of biological similarity,”Journal of Theoretical Biology, vol. 94, no. 1, pp. 25–60, 1982

    work page 1982

  6. [6]

    Allometry of the limb bones of mammals from shrews (Sorex) to elephant (Loxodonta),

    R. M. Alexander, A. Jayes, G. Maloiy, and E. Wathuta, “Allometry of the limb bones of mammals from shrews (Sorex) to elephant (Loxodonta),”Journal of Zoology, vol. 189, no. 3, pp. 305–314, 1979

    work page 1979

  7. [7]

    Why animals can outrun robots,

    S. A. Burden, T. Libby, K. Jayaram, S. Sponberg, and J. M. Donelan, “Why animals can outrun robots,”Science Robotics, vol. 9, no. 89, p. eadi9754, 2024

    work page 2024

  8. [8]

    Design of a bio-inspired dynamical vertical climbing robot,

    J. E. Clark, D. I. Goldman, P.-C. Lin, G. Lynch, H. Komsuoglu, T. S. Chen, R. J. Full, and D. E. Koditschek, “Design of a bio-inspired dynamical vertical climbing robot,”Robotics: Science and Systems, no. 876, pp. 9–16, 2007

    work page 2007

  9. [9]

    Vertical dry adhesive climbing with a 100×bodyweight payload,

    E. W. Hawkes, D. L. Christensen, and M. R. Cutkosky, “Vertical dry adhesive climbing with a 100×bodyweight payload,” in2015 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2015, pp. 3762–3769

    work page 2015

  10. [10]

    MiniRHex: A small, open-source, fully programmable walking hexapod,

    M. Barragan, N. Flowers, and A. M. Johnson, “MiniRHex: A small, open-source, fully programmable walking hexapod,” inRobotics: Science and Systems Workshop on “Design and Control of Small Legged Robots”, Pittsburgh, PA, June 2018, workshop abstract

    work page 2018

  11. [11]

    Scaling down an insect-size microrobot, HAMR-VI into HAMR-Jr,

    K. Jayaram, J. Shum, S. Castellanos, E. F. Helbling, and R. J. Wood, “Scaling down an insect-size microrobot, HAMR-VI into HAMR-Jr,” inIEEE International Conference on Robotics and Automation, 2020, pp. 10 305–10 311

    work page 2020

  12. [12]

    The microDelta: Downscaling robot mechanisms enables ultrafast and high-precision movement,

    S. Man, S. Kim, and S. Bergbreiter, “The microDelta: Downscaling robot mechanisms enables ultrafast and high-precision movement,” Science Robotics, vol. 10, no. 108, p. eadx3883, 2025

    work page 2025

  13. [13]

    The simplest walking robot: A bipedal robot with one actuator and two rigid bodies,

    J. Kyle, J. K. Yim, K. Hart, S. Bergbreiter, and A. M. Johnson, “The simplest walking robot: A bipedal robot with one actuator and two rigid bodies,” inIEEE-RAS International Conference on Humanoid Robots, 2023, pp. 1–7

    work page 2023

  14. [14]

    Zippy: The smallest power-autonomous bipedal robot,

    S. Man, S. Narita, J. Macera, N. Oke, A. M. Johnson, and S. Bergbre- iter, “Zippy: The smallest power-autonomous bipedal robot,” inIEEE Intl. Conference on Robotics and Automation, 2025

    work page 2025

  15. [15]

    A dynamic similarity hypothesis for the gaits of quadrupedal mammals,

    R. M. Alexander and A. Jayes, “A dynamic similarity hypothesis for the gaits of quadrupedal mammals,”Journal of Zoology, vol. 201, no. 1, pp. 135–152, 1983

    work page 1983

  16. [16]

    Allometric scaling of body length: Elastic or geometric similarity in mammalian design,

    M. Silva, “Allometric scaling of body length: Elastic or geometric similarity in mammalian design,”Journal of Mammalogy, vol. 79, no. 1, pp. 20–32, 1998

    work page 1998

  17. [17]

    Scale effects between body size and limb design in quadrupedal mammals,

    B. M. Kilbourne and L. C. Hoffman, “Scale effects between body size and limb design in quadrupedal mammals,”PloS one, vol. 8, no. 11, p. e78392, 2013

    work page 2013

  18. [18]

    Scaling of avian bipedal locomotion reveals independent effects of body mass and leg posture on gait,

    M. A. Daley and A. Birn-Jeffery, “Scaling of avian bipedal locomotion reveals independent effects of body mass and leg posture on gait,” Journal of Experimental Biology, vol. 221, no. 10, p. jeb152538, 2018

    work page 2018

  19. [19]

    A general scaling law reveals why the largest animals are not the fastest,

    M. R. Hirt, W. Jetz, B. C. Rall, and U. Brose, “A general scaling law reveals why the largest animals are not the fastest,”Nature Ecology & Evolution, vol. 1, no. 8, pp. 1116–1122, 2017

    work page 2017

  20. [20]

    Dynamic similarity and the peculiar allometry of maximum running speed,

    D. Labonte, P. J. Bishop, T. J. Dick, and C. J. Clemente, “Dynamic similarity and the peculiar allometry of maximum running speed,” Nature Communications, vol. 15, no. 1, p. 2181, 2024

    work page 2024

  21. [21]

    Analyzing modeled torque profiles to understand scale-dependent active muscle responses in the hip joint

    F. R. Young, H. J. Chiel, M. C. Tresch, C. J. Heckman, A. J. Hunt, and R. D. Quinn, “Analyzing modeled torque profiles to understand scale-dependent active muscle responses in the hip joint.”Biomimetics, vol. 7, no. 1, 2022

    work page 2022

  22. [22]

    From cockroaches to tanks: The same power-mass-speed relation describes both biological and artificial ground-mobile systems,

    A. Kott, S. Gart, and J. Pusey, “From cockroaches to tanks: The same power-mass-speed relation describes both biological and artificial ground-mobile systems,”PLoS One, vol. 16, no. 4, p. e0249066, 2021

    work page 2021

  23. [23]

    En- ergetics and mechanics of terrestrial locomotion. IV. total mechanical energy changes as a function of speed and body size in birds and mammals,

    N. C. Heglund, M. A. Fedak, C. R. Taylor, and G. A. Cavagna, “En- ergetics and mechanics of terrestrial locomotion. IV. total mechanical energy changes as a function of speed and body size in birds and mammals,”Journal of Experimental Biology, vol. 97, no. 1, 1982

    work page 1982

  24. [24]

    Scaling laws for mesoscale and microscale systems,

    B. Spletzer, “Scaling laws for mesoscale and microscale systems,” IEEE Transactions on CPMT – Advanced Packaging, 08 1999

    work page 1999

  25. [25]

    Fascination of down scaling- Alice the sugar cube robot,

    G. Caprari, T. Estier, and R. Siegwart, “Fascination of down scaling- Alice the sugar cube robot,” inIEEE International Conference on Robotics and Automation: Workshop on Mobile Micro-Robots, 2000

    work page 2000

  26. [26]

    Rigid vs. elastic actuation: Requirements & performance,

    S. Haddadin, N. Mansfeld, and A. Albu-Sch ¨affer, “Rigid vs. elastic actuation: Requirements & performance,” inIEEE/RSJ International Conference on Intelligent Robots and Systems, 2012, pp. 5097–5104

    work page 2012

  27. [27]

    Sizing equations for electrical machinery,

    V . Honsinger, “Sizing equations for electrical machinery,”IEEE Trans- actions on Energy Conversion, vol. 2, no. 1, pp. 116–121, 1987

    work page 1987

  28. [28]

    The influence of DC electric drives on sizing quadruped robots,

    P. Chatzakos and E. Papadopoulos, “The influence of DC electric drives on sizing quadruped robots,” inIEEE International Conference on Robotics and Automation, 2008, pp. 793–798

    work page 2008

  29. [29]

    Torque scaling laws for interior and exterior rotor permanent magnet machines,

    T. Reichert, T. Nussbaumer, and J. W. Kolar, “Torque scaling laws for interior and exterior rotor permanent magnet machines,” inIEEE International Magnetics Conference, 2009, pp. 1–4

    work page 2009

  30. [30]

    Scaling laws in robotics,

    K. Dermitzakis, J. P. Carbajal, and J. H. Marden, “Scaling laws in robotics,”Procedia Computer Science, vol. 7, pp. 250–252, 2011

    work page 2011

  31. [31]

    Scaling of robotic mechanisms,

    K. J. Waldron and C. Hubert, “Scaling of robotic mechanisms,” in IEEE International Conference on Robotics and Automation, vol. 1, 2000, pp. 40–45

    work page 2000

  32. [32]

    Design principles for energy-efficient legged locomotion and implementation on the mit cheetah robot,

    S. Seok, A. Wang, M. Y . M. Chuah, D. J. Hyun, J. Lee, D. M. Otten, J. H. Lang, and S. Kim, “Design principles for energy-efficient legged locomotion and implementation on the mit cheetah robot,” IEEE/ASME Transactions on Mechatronics, vol. 20, no. 3, pp. 1117– 1129, 2015

    work page 2015

  33. [33]

    Scaling laws for parallel motor-gearbox arrangements,

    E. Saerens, S. Crispel, P. L. Garc ´ıa, V . Ducastel, J. Beckers, J. De Win- ter, R. Furn ´emont, B. Vanderborght, T. Verstraten, and D. Lefeber, “Scaling laws for parallel motor-gearbox arrangements,” inIEEE/RSJ International Conference on Intelligent Robots and Systems, 2020, pp. 6339–6346

    work page 2020

  34. [34]

    Design Principles for a Family of Direct-Drive Legged Robots,

    G. Kenneally, A. De, and D. E. Koditschek, “Design Principles for a Family of Direct-Drive Legged Robots,”IEEE Robotics and Automation Letters, vol. 1, no. 2, pp. 900–907, July 2016

    work page 2016

  35. [35]

    Scaling laws for robotic transmissions,

    E. Saerens, S. Crispel, P. L. Garcia, T. Verstraten, V . Ducastel, B. Van- derborght, and D. Lefeber, “Scaling laws for robotic transmissions,” Mechanism and Machine Theory, vol. 140, pp. 601–621, 2019

    work page 2019

  36. [36]

    Drake: Model-based design and verification for robotics,

    R. Tedrake and Drake Development Team, “Drake: Model-based design and verification for robotics,” https://drake.mit.edu/, 2019

    work page 2019