Pith. sign in

REVIEW 4 major objections 4 minor 35 references

Learning Speed-Adaptive Walking Agent Using Imitation Learning with Physics-Informed Simulation

T0 review · 4 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

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

Pith's one-line read This paper reports a single control policy for a skeletal humanoid that produces stable, human-like walking across speeds from 0.65 to 1.85 m/s, trained by adversarial imitation learning on synthetic expert data derived from an…

desk verdict Useful speed-adaptive imitation pipeline, but the headline RMSE validates against the synthetic training data, not human gait. read the letter →

arxiv 2412.03949 v1 pith:QVKAVUR3 submitted 2024-12-05 cs.RO cs.LG

Yi-Hung Chiu , Ung Hee Lee , Changseob Song , Manaen Hu , Inseung Kang This is my paper · ORCID
classification cs.ROcs.LG
keywords imitationlearningspeed-adaptivewalkingsyntheticgaitdatadigitaltwinlocomotioncurriculumbiomechanicsphysics-basedsimulation
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 claims that a single control policy for a skeletal humanoid can produce stable, natural walking across a continuous range of speeds by learning from synthetic motion data rather than new motion capture. The authors build a synthetic data generator that extrapolates an open-source biomechanics dataset across speeds, then train an adversarial imitation learning agent in a physics-based simulator. The trained agent tracks target speeds to within 0.06 m/s and matches synthetic joint-angle profiles to within about 5.24 degrees, significantly beating a baseline that lacks multi-speed demonstrations. This matters because a reusable, speed-adaptive walking agent could serve as a digital twin for studying gait, designing exoskeletons, and planning rehabilitation without expensive data collection.

What carries the argument

The load-bearing machinery is the synthetic motion data generator combined with an adversarial imitation learning loop. The generator fits a linear model to joint angle as a function of walking speed at 21 points in the gait cycle, using averaged data from 22 subjects at speeds from 0.5 to 1.85 m/s; it then applies this model to a single-speed simulation dataset to produce expert kinematics at arbitrary speeds. The learning loop uses a variational discriminator bottleneck (VAIL) to supply imitation reward, an exponential speed-tracking reward, and a progressive curriculum that sweeps target speeds gradually during training; Trust Region Policy Optimization updates the policy.

What would settle it

Measure actual joint-angle trajectories of healthy subjects walking at 0.5, 0.6, and 0.7 m/s and compare them to the linear extrapolation from the 0.5 to 1.85 m/s fit; a systematic, speed-dependent curvature in the residuals would show that the expert demonstrations are biased at slow speeds.

Watch

Extended reading notes

Core claim

The central discovery is that speed adaptability in a physics-based walking agent can be obtained from an entirely synthetic expert: joint-angle profiles linearly interpolated across walking speed from averaged human gait data. Using VAIL, a variational adversarial imitation learning method, with a discriminator that compares expert and agent state-action pairs, a speed reward for center-of-mass velocity, a progressive curriculum over target speeds, and TRPO for policy optimization, the agent achieves an RMSE of 5.24±0.09 degrees for lower-limb joint angles and 0.06±0.01 m/s for COM speed relative to the synthetic ground truth, with R² of 0.83±0.01 and 0.98±0.01. The optimal configuration significantly outperforms a baseline agent that receives only the speed reward and target-speed state but not the multi-speed expert demonstrations (p<0.05). The paper further reports biomechanically plausible joint torques and powers, though with some discrepancies from biological data, attributed to kinematics-only demonstrations and a simplified foot model.

Load-bearing premise

The whole pipeline assumes that human gait kinematics scale linearly with walking speed, so the expert demonstrations are straight-line extrapolations of averaged data; if real gait changes nonlinearly at slow speeds, the agent is trained toward a distorted target.

Editorial extensions

If this is right

  • A single learned policy, rather than a bank of speed-specific controllers, can serve a walking agent across 0.65 to 1.85 m/s, simplifying deployment in assistive devices.
  • Progressive speed curricula are markedly better than random speed presentation, improving joint-angle R² by about 48% and reducing RMSE by about 41%.
  • A balanced combination of imitation and speed rewards (1:1) produces the best kinematic fidelity; heavier speed weighting yields unnatural gait, indicating that pure speed tracking is insufficient for human-like motion.
  • The same synthetic-data-plus-imitation pipeline is directly reusable for other gait conditions, such as slopes or stairs, if a suitable biomechanics dataset is available.
  • The trained agent's speed tracking extends to dynamically varying targets (chirp signals), suggesting it can respond to real-time speed commands.

Reading between the lines

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

  • Because both the expert demonstrations and the evaluation reference are derived from the same linear model, the reported RMSE measures imitation fidelity to synthetic data, not to real human gait; real-world error at extreme speeds is likely larger.
  • The paper's observed kinetic discrepancies, such as an extra hip extension torque and deviant knee power, imply that kinematics-only imitation does not enforce joint moments; adding ground-reaction or torque-profile rewards could produce a more biomechanically faithful twin.
  • Averaging across 22 subjects erases individual gait signatures, so the resulting agent represents an average walker; personalizing the digital twin would require a conditional generator that preserves inter-subject variability.
  • A natural next step is sim-to-real transfer: porting this policy to an exoskeleton controller would test whether the synthetic-data-trained policy retains natural gait under physical contact dynamics.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 4 minor

Summary. The paper presents a pipeline for training a speed-adaptive skeletal humanoid walking agent in LocoMuJoCo using variational adversarial imitation learning (VAIL). Expert demonstrations are generated by a linear model fitted to the Camargo open-source lower-limb biomechanics dataset across 0.5–1.85 m/s, and the agent's observation includes the target speed. Training uses a progressive speed curriculum and a reward that combines a discriminator term with an explicit speed-tracking term. The main reported results are a joint-angle RMSE of 5.24±0.09° and R2=0.83 against the synthetic reference, a target-speed tracking RMSE of 0.06±0.01 m/s with R2=0.98, and a statistically significant improvement over a baseline agent (p<0.05). The paper also includes a discussion of limitations, including the linearity assumption and kinetic discrepancies.

Significance. If the validation were against real human gait, the work would be a significant step toward digital twins of locomotion. The authors make several good methodological choices: they build on an open-source biomechanics dataset and LocoMuJoCo, release their implementation, and run structured ablations of curriculum design and reward weighting. The speed-tracking result is internally valid, and the progressive-curriculum benefit (41% RMSE reduction over random) is a concrete, reproducible contribution. However, the central kinematic fidelity claim is measured against the synthetic data that serve as the training target, not against held-out human motion capture. Given that the synthetic data themselves match the Camargo dataset only moderately (R2=0.59), the stated 'biomechanically realistic motions' claim is not established by the reported metrics. This limits the paper's contribution to a well-engineered speed-adaptive imitation agent with respect to a synthetic reference.

major comments (4)
  1. [III-D, Fig. 5, Abstract] The reported joint-angle RMSE of 5.24±0.09° and R2 of 0.83 are computed against the synthetic linear-model gait profiles of Section II-B, the same generator that produced the expert demonstrations for VAIL training in Section II-C. This makes the metric a measure of imitation fidelity to the training target, not a measure of accuracy to human gait. The Abstract's phrase 'ground-truth kinematics data' and Table I's column title 'Ground Truth vs. Measured Joint Angle' are therefore misleading, especially since the synthetic data themselves deviate from the Camargo dataset by RMSE 8.10° and R2 0.59 (Section III-A). The claim of 'biomechanically realistic motions' is not supported by the current validation. Please reframe the evaluation as a comparison against the synthetic reference and add at least a quantitative comparison of the agent's joint angles to the original Camargo data (or another held-out human gait dataset) for a subset of speeds.
  2. [III-C, Table I] The selection of the optimal reward ratio is not transparent. The text says ratios '0.4 and above 0.5' were excluded, which is ambiguous, and Table I reports no joint-angle entries for the 0.4 speed-reward ratio. The optimal agent is declared to use a speed-reward ratio of 0.5, but the joint-angle R2 at a ratio of 0.1 (0.90±0.01) is higher than at 0.5 (0.83±0.01). The 'aggregated R2' criterion used to choose the optimal setting is never defined (e.g., how joint-angle R2 and speed-tracking R2 are combined across joints and speeds). Please specify the selection rule and justify the choice of 0.5.
  3. [II-C.3] The baseline agent's access to data is underspecified. The text states that the baseline had 'limited access to multi-speed data from our synthetic generator' but then says it received different speed information 'only through the speed-based reward and the target speed state.' It is unclear what demonstrations the baseline was trained on (e.g., a single speed from the synthetic set, the original LocoMuJoCo single-speed dataset, or something else). Because the comparison to the baseline underlies the claim that the synthetic multi-speed generator is beneficial, please describe the baseline's training demonstrations exactly.
  4. [II-D, III-D] The statistical claim that the optimal agent 'significantly outperformed' the baseline (p<0.05) is not fully supported by the described analysis. A paired sample t-test is mentioned, but the number of independent evaluation trials, the pairing structure, and whether multiple comparisons across 13 speeds and three metrics were accounted for are not reported. Please provide the test specification and report effect sizes or confidence intervals for the differences.
minor comments (4)
  1. [II-C, Eq. (2)] The formula contains unmatched parentheses in the discriminator term, which makes the reward expression hard to parse.
  2. [III-C] The sentence 'Speed reward ratios of 0.4 and above 0.5 were excluded...' should be rephrased; the intended meaning appears to be that speed-reward ratios above 0.5 were excluded due to unviable gait cycles, while the 0.4 row lacks joint-angle data.
  3. [II-B] Please clarify how the 21 points are distributed over the gait cycle (e.g., normalized percent) and how synthetic velocities were obtained from the fitted position data.
  4. [Abstract, Fig. 5] The term 'ground-truth data' should be replaced with 'synthetic reference data' to avoid overstating the validation, consistent with Major Comment 1.

Circularity Check

1 steps flagged · score 6.0 of 10

Central kinematic validation is circular: the 5.24° RMSE is measured against the same synthetic linear-model dataset that generated the expert demonstrations the agent was trained to imitate.

  1. fitted input called prediction [Section II-B (Synthetic Motion Data Generator), Section II-C (Training), Section II-D and III-D (Model Evaluation and Optimal Agent Settings), Abstract]
    "By applying the linear model to the LocoMuJoCo’s single-speed walking dataset, we generated artificial data that vary across speeds. ... We evaluated the agent’s performance by calculating the RMSE of the averaged differences in the hip, knee, and ankle joint angles between the synthetic dataset and the generated gait profiles of the trained agent across all speed conditions."

    The VAIL objective trains the policy to imitate expert demonstrations produced by this same synthetic generator; the discriminator reward is maximal when the agent’s state-action pairs match those demonstrations. The evaluation then compares the trained agent’s joint angles to “the synthetic dataset” at each test speed, so the headline RMSE of 5.24° and R²=0.83 measure how well the policy reproduces the fitted linear target, not agreement with real human gait or even with the original open-source dataset. The original biomechanics data differ from the synthetic data by RMSE=8.10° and R²=0.59 (Section III-A), and the only comparison to real biomechanics is qualitative (Fig. 6), where the authors concede notable kinetic discrepancies.

full rationale

The paper contains one concrete circular reduction in its central quantitative claim. The synthetic motion data generator is a linear model fitted to averaged open-source gait data; that same generator supplies the expert demonstrations for variational adversarial imitation learning. The evaluation metric in Sections II-D and III-D computes RMSE/R² between the trained agent’s joint angles and this same synthetic dataset. Thus the headline 5.24±0.09° RMSE and R²=0.83 are self-consistency scores with respect to the training reference, not predictions validated against independent human gait data. The paper’s own Discussion acknowledges the linearity assumption may bias the synthetic data, but that caveat is secondary: even a perfect nonlinear generator would leave the evaluation circular unless the agent were compared against held-out real motion capture data. The speed-tracking result, curriculum comparison, and reward-ratio ablation retain independent content, so the circularity is substantial but not total. No load-bearing self-citation, uniqueness import, or ansatz-smuggling pattern was found. Score 6 reflects that the strongest quantitative claim reduces by construction while other parts of the paper stand on their own.

Assumptions & free parameters 4 free parameters · 5 assumptions · 0 invented entities

The central claim rests on the fitted linear model that generates expert data, the reward weighting lambda, and the evaluation filtering threshold. No new physical entities are introduced; the speed-adaptive behavior is achieved through a conditioned policy rather than a novel mechanism.

free parameters (4)
  • Linear model coefficients (slope and intercept per gait-cycle point per joint) = 21 points per joint for hip, knee, ankle; slopes and intercepts fitted to average profiles from Camargo dataset
    These fitted coefficients define the synthetic expert demonstrations used to train the policy; the agent's target kinematics are derived from them.
  • Speed reward weighting ratio lambda = 0.5 (discriminator weight 0.5)
    Chosen after a sweep over ratios 0 to 1 in 0.1 increments, excluding 0.4 and above 0.5 due to unnatural gait; selection is based on aggregated R2.
  • Gait cycle filtering threshold = 60 time steps (0.6 s minimum stride duration)
    Gait cycles shorter than 0.6 s are discarded in evaluation, which may remove falls or abnormal steps before computing RMSE and R2.
  • Training and inference speed range = 0.65 to 1.85 m/s
    The target speed range is capped at speeds available in the dataset; evaluation uses 0.1 m/s increments within this range.
assumptions (5)
  • ad hoc to paper Gait kinematics vary linearly with walking speed over 0.5 to 1.85 m/s.
    Section II-B fits a linear model to joint angles as a function of speed; Discussion admits nonlinear behavior at slow speeds, so this assumption may bias the expert data.
  • domain assumption The averaged profiles across 22 subjects represent valid expert behavior for a single adult skeletal model.
    Section II-B averages across subjects, removing individual variation; the policy then imitates the average rather than any real person.
  • domain assumption The LocoMuJoCo adult model with fixed upper body and rigid box feet captures the lower-limb dynamics needed for walking.
    Section II-A; Discussion notes the box foot model contributes to the sim-to-real gap and kinetic deviations.
  • domain assumption The linear model fitted on Camargo data can be applied to LocoMuJoCo's 1.25 m/s dataset to produce morphology-consistent synthetic data.
    Section II-B uses this transfer without quantitative validation that the linear model preserves the simulation model's physical consistency.
  • standard math VAIL and TRPO are appropriate and stable algorithms for this training task.
    The paper relies on established algorithms for policy optimization and imitation; these are standard in the literature.

how reviews work

0 comments
Cite this review

Pith. "Pith review of Learning Speed-Adaptive Walking Agent Using Imitation Learning with Physics-Informed Simulation." pith.science (2026). https://pith.science/paper/QVKAVUR3

@misc{pith2026241203949,
  author       = {Pith},
  title        = {Pith review of: Learning Speed-Adaptive Walking Agent Using Imitation Learning with Physics-Informed Simulation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/QVKAVUR3}},
  note         = {Machine review of arXiv:2412.03949}
}
read the original abstract

Virtual models of human gait, or digital twins, offer a promising solution for studying mobility without the need for labor-intensive data collection. However, challenges such as the sim-to-real gap and limited adaptability to diverse walking conditions persist. To address these, we developed and validated a framework to create a skeletal humanoid agent capable of adapting to varying walking speeds while maintaining biomechanically realistic motions. The framework combines a synthetic data generator, which produces biomechanically plausible gait kinematics from open-source biomechanics data, and a training system that uses adversarial imitation learning to train the agent's walking policy. We conducted comprehensive analyses comparing the agent's kinematics, synthetic data, and the original biomechanics dataset. The agent achieved a root mean square error of 5.24 +- 0.09 degrees at varying speeds compared to ground-truth kinematics data, demonstrating its adaptability. This work represents a significant step toward developing a digital twin of human locomotion, with potential applications in biomechanics research, exoskeleton design, and rehabilitation.

Figures

Figures reproduced from arXiv: 2412.03949 by the authors.

Figure 1
Figure 1. Overview of our framework for training a speed-adaptive walking agent. The framework includes two main components: generating expert [PITH_FULL_IMAGE:figures/full_fig_p002_1.png] view at source ↗
Figure 2
Figure 2. Comparison of joint kinematics between open-source biomechanics [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Effect of curriculum learning on the walking agent’s performance [PITH_FULL_IMAGE:figures/full_fig_p004_3.png] view at source ↗
Figures from the paper (3 more)
Figure 5
Figure 5. Figure 5: Walking agent performance with optimal and baseline settings. [PITH_FULL_IMAGE:figures/full_fig_p005_5.png]
Figure 4
Figure 4. Figure 4: Representative walking agent with optimal (top) and suboptimal [PITH_FULL_IMAGE:figures/full_fig_p005_4.png]
Figure 6
Figure 6. Figure 6: Gait biomechanics comparison between the open-source dataset (left) and our optimal walking agent (right) across varying speeds. Each curve [PITH_FULL_IMAGE:figures/full_fig_p006_6.png]

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

35 extracted references · 25 canonical work pages

  1. [1]

    Mobility and aging: new directions for public health action,

    W. A. Satariano, J. M. Guralnik, R. J. Jackson, R. A. Marottoli, E. A. Phelan, and T. R. Prohaska, “Mobility and aging: new directions for public health action,” American journal of public health , vol. 102, no. 8, pp. 1508–1515, 2012

    work page 2012

  2. [2]

    Mobility in older adults: a comprehensive framework,

    S. C. Webber, M. M. Porter, and V . H. Menec, “Mobility in older adults: a comprehensive framework,” The Gerontologist, vol. 50, no. 4, pp. 443–450, 2010

    work page 2010

  3. [3]

    Disability and health data system (dhds),

    Centers for Disease Control and Prevention, “Disability and health data system (dhds),” 2024

    work page 2024

  4. [4]

    Mobility difficulties are not only a problem of old age,

    L. I. Iezzoni, E. P. McCarthy, R. B. Davis, and H. Siebens, “Mobility difficulties are not only a problem of old age,” Journal of General Internal Medicine, vol. 16, no. 4, pp. 235–243, 2001

    work page 2001

  5. [5]

    Gait analysis methods in rehabilitation,

    R. Baker, “Gait analysis methods in rehabilitation,” Journal of Neu- roEngineering and Rehabilitation , vol. 3, no. 1, p. 4, 2006

    work page 2006

  6. [6]

    Opensim: open-source soft- ware to create and analyze dynamic simulations of movement,

    S. L. Delp, F. C. Anderson, A. S. Arnold, P. Loan, A. Habib, C. T. John, E. Guendelman, and D. G. Thelen, “Opensim: open-source soft- ware to create and analyze dynamic simulations of movement,” IEEE transactions on biomedical engineering , vol. 54, no. 11, pp. 1940– 1950, 2007

    work page 1940

  7. [7]

    A survey on digital twin: Definitions, characteristics, applications, and design implications,

    B. R. Barricelli, E. Casiraghi, and D. Fogli, “A survey on digital twin: Definitions, characteristics, applications, and design implications,” IEEE access, vol. 7, pp. 167653–167671, 2019

    work page 2019

  8. [8]

    An interactive graphics-based model of the lower extremity to study orthopaedic surgical procedures,

    S. L. Delp, J. P. Loan, M. G. Hoy, F. E. Zajac, E. L. Topp, and J. M. Rosen, “An interactive graphics-based model of the lower extremity to study orthopaedic surgical procedures,” IEEE Transactions on Biomedical engineering, vol. 37, no. 8, pp. 757–767, 1990

    work page 1990

Show all 35 references
  1. [9]

    Predicting gait adaptations due to ankle plantarflexor muscle weakness and contracture using physics-based musculoskeletal simulations,

    C. F. Ong, T. Geijtenbeek, J. L. Hicks, and S. L. Delp, “Predicting gait adaptations due to ankle plantarflexor muscle weakness and contracture using physics-based musculoskeletal simulations,” PLoS computational biology, vol. 15, no. 10, p. e1006993, 2019

    2019

  2. [10]

    Optimal control based stiffness identification of an ankle-foot orthosis using a predictive walking model,

    M. Sreenivasa, M. Millard, M. L. Felis, K. Mombaur, and S. I. Wolf, “Optimal control based stiffness identification of an ankle-foot orthosis using a predictive walking model,” Frontiers in computational neuroscience, vol. 11, p. 23, 2017

    2017

  3. [11]

    Learning a control policy for fall prevention on an assistive walking device,

    V . C. Kumar, S. Ha, G. Sawicki, and C. K. Liu, “Learning a control policy for fall prevention on an assistive walking device,” in 2020 IEEE International Conference on Robotics and Automation (ICRA) , pp. 4833–4840, IEEE, 2020

    2020

  4. [12]

    A lightweight robotic leg prosthesis replicating the biomechanics of the knee, ankle, and toe joint,

    M. Tran, L. Gabert, S. Hood, and T. Lenzi, “A lightweight robotic leg prosthesis replicating the biomechanics of the knee, ankle, and toe joint,” Science robotics, vol. 7, no. 72, p. eabo3996, 2022

    2022

  5. [13]

    The effect of hip assistance levels on human energetic cost using robotic hip exoskeletons,

    I. Kang, H. Hsu, and A. Young, “The effect of hip assistance levels on human energetic cost using robotic hip exoskeletons,” IEEE Robotics and Automation Letters , vol. 4, no. 2, pp. 430–437, 2019

    2019

  6. [14]

    Crossing the reality gap: A survey on sim-to-real transferability of robot controllers in reinforcement learning,

    E. Salvato, G. Fenu, E. Medvet, and F. A. Pellegrino, “Crossing the reality gap: A survey on sim-to-real transferability of robot controllers in reinforcement learning,” IEEE Access, vol. 9, pp. 153171–153187, 2021

    2021

  7. [15]

    Exploring the sim2real gap using digital twins,

    S. Sudhakar, J. Hanzelka, J. Bobillot, T. Randhavane, N. Joshi, and V . Vineet, “Exploring the sim2real gap using digital twins,” in Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 20418–20427, 2023

    2023

  8. [16]

    Analysis of musculoskeletal systems in the anybody modeling system,

    M. Damsgaard, J. Rasmussen, S. T. Christensen, E. Surma, and M. De Zee, “Analysis of musculoskeletal systems in the anybody modeling system,” Simulation Modelling Practice and Theory, vol. 14, no. 8, pp. 1100–1111, 2006

    2006

  9. [17]

    Myosuite–a contact-rich simulation suite for musculoskeletal motor control,

    V . Caggiano, H. Wang, G. Durandau, M. Sartori, and V . Kumar, “Myosuite–a contact-rich simulation suite for musculoskeletal motor control,” arXiv preprint arXiv:2205.13600 , 2022

    2022 arXiv

  10. [18]

    Mujoco: A physics engine for model-based control,

    E. Todorov, T. Erez, and Y . Tassa, “Mujoco: A physics engine for model-based control,” in 2012 IEEE/RSJ international conference on intelligent robots and systems , pp. 5026–5033, IEEE, 2012

    2012

  11. [19]

    Markerless motion capture through visual hull, articu- lated icp and subject specific model generation,

    S. Corazza, L. M ¨undermann, E. Gambaretto, G. Ferrigno, and T. P. Andriacchi, “Markerless motion capture through visual hull, articu- lated icp and subject specific model generation,” International journal of computer vision , vol. 87, pp. 156–169, 2010

    2010

  12. [20]

    Soft actor-critic: Off- policy maximum entropy deep reinforcement learning with a stochastic actor,

    T. Haarnoja, A. Zhou, P. Abbeel, and S. Levine, “Soft actor-critic: Off- policy maximum entropy deep reinforcement learning with a stochastic actor,” in International conference on machine learning , pp. 1861– 1870, PMLR, 2018

    2018

  13. [21]

    Deepmimic: Example-guided deep reinforcement learning of physics-based char- acter skills,

    X. B. Peng, P. Abbeel, S. Levine, and M. Van de Panne, “Deepmimic: Example-guided deep reinforcement learning of physics-based char- acter skills,” ACM Transactions On Graphics (TOG) , vol. 37, no. 4, pp. 1–14, 2018

    2018

  14. [22]

    Opensim moco: Musculoskeletal optimal control,

    C. L. Dembia, N. A. Bianco, A. Falisse, J. L. Hicks, and S. L. Delp, “Opensim moco: Musculoskeletal optimal control,” PLOS Computa- tional Biology, vol. 16, no. 12, p. e1008493, 2020

    2020

  15. [23]

    Reinforcement learning for control of human locomotion in simulation,

    A. Dashkovets and B. Laschowski, “Reinforcement learning for control of human locomotion in simulation,” in 2024 10th IEEE RAS/EMBS International Conference for Biomedical Robotics and Biomechatronics (BioRob), pp. 43–48, IEEE, 2024

    2024

  16. [24]

    Variational discriminator bottleneck: Improving imitation learning, inverse rl, and gans by constraining information flow,

    X. B. Peng, A. Kanazawa, S. Toyer, P. Abbeel, and S. Levine, “Variational discriminator bottleneck: Improving imitation learning, inverse rl, and gans by constraining information flow,” arXiv preprint arXiv:1810.00821, 2018

    2018 arXiv

  17. [25]

    A compre- hensive, open-source dataset of lower limb biomechanics in multiple conditions of stairs, ramps, and level-ground ambulation and transi- tions,

    J. Camargo, A. Ramanathan, W. Flanagan, and A. Young, “A compre- hensive, open-source dataset of lower limb biomechanics in multiple conditions of stairs, ramps, and level-ground ambulation and transi- tions,” Journal of Biomechanics , vol. 119, p. 110320, 2021

    2021

  18. [26]

    Locomujoco: A comprehensive imitation learning benchmark for locomotion,

    F. Al-Hafez, G. Zhao, J. Peters, and D. Tateo, “Locomujoco: A comprehensive imitation learning benchmark for locomotion,” arXiv preprint arXiv:2311.02496, 2023

    2023 arXiv

  19. [27]

    Muscle contributions to propulsion and support during running,

    S. R. Hamner, A. Seth, and S. L. Delp, “Muscle contributions to propulsion and support during running,” Journal of biomechanics , vol. 43, no. 14, pp. 2709–2716, 2010

    2010

  20. [28]

    Benchmark datasets for bi- lateral lower-limb neuromechanical signals from wearable sensors during unassisted locomotion in able-bodied individuals,

    B. Hu, E. Rouse, and L. Hargrove, “Benchmark datasets for bi- lateral lower-limb neuromechanical signals from wearable sensors during unassisted locomotion in able-bodied individuals,” Frontiers in Robotics and AI , vol. 5, p. 14, 2018

    2018

  21. [29]

    Human kinematic, kinetic and emg data during level walking, toe/heel-walking, stairs ascending/descending,

    T. Lencioni, I. Carpinella, M. Rabuffetti, A. Marzegan, and M. Fer- rarin, “Human kinematic, kinetic and emg data during level walking, toe/heel-walking, stairs ascending/descending,” Figshare https://doi. org/10.6084/m9. figshare. c, vol. 4494755, p. v1, 2019

    doi

  22. [30]

    Lower limb kinematic, kinetic, and emg data from young healthy humans during walking at controlled speeds,

    L. Moreira, J. Figueiredo, P. Fonseca, J. P. Vilas-Boas, and C. P. Santos, “Lower limb kinematic, kinetic, and emg data from young healthy humans during walking at controlled speeds,” Scientific data, vol. 8, no. 1, p. 103, 2021

    2021

  23. [31]

    Generative adversarial net- works,

    I. Goodfellow, J. Pouget-Abadie, M. Mirza, B. Xu, D. Warde-Farley, S. Ozair, A. Courville, and Y . Bengio, “Generative adversarial net- works,” Communications of the ACM , vol. 63, no. 11, pp. 139–144, 2020

    2020

  24. [32]

    Trust region policy optimization,

    J. Schulman, “Trust region policy optimization,” arXiv preprint arXiv:1502.05477, 2015

    2015 arXiv

  25. [33]

    Natural and robust walking using reinforcement learning without demonstrations in high-dimensional musculoskeletal models,

    P. Schumacher, T. Geijtenbeek, V . Caggiano, V . Kumar, S. Schmitt, G. Martius, and D. F. Haeufle, “Natural and robust walking using reinforcement learning without demonstrations in high-dimensional musculoskeletal models,” arXiv preprint arXiv:2309.02976 , 2023

    2023 arXiv

  26. [34]

    Creation and evaluation of human models with varied walking ability from motion capture for assistive device development,

    S. S. Chan, M. Lei, H. Johan, and W. T. Ang, “Creation and evaluation of human models with varied walking ability from motion capture for assistive device development,” in 2023 International Conference on Rehabilitation Robotics (ICORR) , pp. 1–6, IEEE, 2023

    2023

  27. [35]

    Deep reinforcement learning for modeling human locomotion control in neuromechanical simulation,

    S. Song, Ł. Kidzi ´nski, X. B. Peng, C. Ong, J. Hicks, S. Levine, C. G. Atkeson, and S. L. Delp, “Deep reinforcement learning for modeling human locomotion control in neuromechanical simulation,” Journal of neuroengineering and rehabilitation, vol. 18, pp. 1–17, 2021. This wor...

    2021

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.