The reviewed record of science sign in
Pith

arxiv: 2607.06052 · v1 · pith:4J3MY6KF · submitted 2026-07-07 · cs.RO

ThorArena: Benchmarking Humanoid Physical Interaction with Human Motion-Force Demonstrations

Chenhao Yu , Hongwu Wang , Weitao Zhang , Youhao Hu , Jiachen Zhang , Gangyang Li , Alois Knoll , Shaqi Luo This is my paper

Reviewed by Pith T0 review T1 audit T2 compute T3 formal T4 kernel 2026-07-08 17:59 UTCglm-5.2pith:4J3MY6KFrecord.jsonopen to challenge →

classification cs.RO
keywords humanoid roboticsforce-aware evaluationwhole-body controlmotion-force datasetbenchmarkcontact-rich interactionsimulation-based evaluation
0
0 comments X

The pith

Force Replay Exposes Hidden Weaknesses in Humanoid Robot Control

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

This paper argues that conventional evaluation of humanoid robot control policies — which only checks how well a robot tracks motion in free space — systematically hides critical weaknesses that emerge when the robot must also withstand physical interaction forces. The authors build ThorArena, a benchmark that pairs synchronized human motion and hand-force recordings across six contact-rich tasks (table wiping, object lifting/lowering, chair pushing/pulling, cooperative carrying), then replays those recorded forces onto a simulated humanoid's hands while a control policy executes the corresponding motion. The central metric is the Force-Aware Tracking Score (FATS), which weights tracking error more heavily in high-force regimes and combines it with episode survival. The paper's key empirical claim: four representative whole-body control policies that look nearly identical under no-force evaluation (survival rates near 1.0, FATS scores within a 6-point band) diverge sharply when forces are replayed — survival on the push-chair task drops to 0.73–0.81 for three of the four policies, and FATS spreads widen by over 30 points on that task. The benchmark thus demonstrates that force-aware evaluation is necessary, not optional, for assessing whether a humanoid policy is ready for contact-rich deployment.

Core claim

The paper's central finding is that interaction forces act as a stress test that separates policies which appear equivalent under kinematic-only evaluation. Under no-force conditions, all four tested policies (Thor2, TWIST2, GMT, SONIC) achieve survival rates above 0.99 and FATS scores between 78.4 and 84.3 — a narrow range suggesting comparable capability. When recorded human hand forces are replayed in simulation, the same policies exhibit qualitatively different failure modes: Thor2 maintains near-perfect survival (1.0) and the lowest tracking error; SONIC preserves balance through compliant behavior but accumulates the largest upper-body tracking errors; TWIST2 and GMT suffer substantial

What carries the argument

Force-Aware Tracking Score (FATS): a composite metric that stratifies timesteps into low-, medium-, and high-force regimes using the 33rd and 66th percentiles of total applied hand force, computes keypoint RMSE in each regime, weights them (0.2, 0.3, 0.5) to emphasize high-force behavior, and combines the weighted error with an episode survival factor via an exponential scoring function. The force-replay protocol transforms recorded sensor-frame force vectors into the simulated hand-local frame and applies them as external forces on the robot's hand bodies during policy rollout, with a scalar coefficient set to 1 for force evaluation and 0 for baseline comparison.

If this is right

  • If the force-replay protocol is valid, then any humanoid control policy evaluated only under no-force conditions may be certified as deployment-ready while hiding force-induced instability that would cause failures in real contact-rich tasks.
  • The FATS framework could be extended to other embodied AI domains where external perturbations matter — for instance, evaluating quadruped locomotion over uneven terrain with replayed ground reaction forces, or assessing bimanual manipulation under varying payload forces.
  • The finding that different policies exhibit distinct trade-offs under force (tracking accuracy vs. survival vs. control effort) suggests that force-aware evaluation could drive a new axis of policy design: not just motion fidelity, but force-robustness profiles tailored to expected interaction types.
  • The paired motion-force dataset (360 sequences across six tasks) could enable training policies that explicitly anticipate interaction forces rather than merely tolerating them, shifting from reactive robustness to predictive force-aware control.

Where Pith is reading between the lines

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

  • If the simulated force-replay protocol does not faithfully reproduce real contact dynamics — because the robot's own motion changes contact location, timing, and force profile — then the performance differences revealed by ThorArena may be artifacts of the replay method rather than genuine policy weaknesses. The paper's own acknowledgment that physical-robot validation is future work means the ben
  • The six-task set, while representative, covers only hand-mediated interactions with relatively static objects. Humanoid deployment scenarios likely involve richer contact patterns (whole-arm contacts, unexpected collisions, multi-contact scenarios) that the current force-replay protocol cannot capture, suggesting the benchmark may under-estimate the range of force-induced failure modes.
  • The FATS weighting scheme (0.2, 0.3, 0.5 for low/mid/high force) implicitly assumes that high-force regimes are the most informative for policy evaluation. This is plausible but untested — it is possible that medium-force regimes, where policies transition from nominal to stressed behavior, are actually more diagnostic of control quality.

Load-bearing premise

The benchmark assumes that replaying recorded human hand forces as fixed external disturbances in simulation is a valid proxy for the contact forces a real humanoid would experience during task execution. In reality, the robot's own motion changes where, when, and how hard it contacts objects, so the replayed forces may not match what the robot would actually encounter.

What would settle it

If a policy that scores poorly on ThorArena's force-replay benchmark performs well on a real humanoid robot executing the same contact-rich tasks, or if a policy that scores well fails in the real world, the benchmark's predictive validity for real-world deployment would be undermined.

Figures

Figures reproduced from arXiv: 2607.06052 by Alois Knoll, Chenhao Yu, Gangyang Li, Hongwu Wang, Jiachen Zhang, Shaqi Luo, Weitao Zhang, Youhao Hu.

Figure 1
Figure 1. Figure 1: Overview of ThorArena. ThorArena integrates real-world motion–force demonstration collection, synchronized interaction-force replay in simulation, and standardized evaluation of humanoid whole-body control policies. The acquisition system records synchronized whole-body human motion and forces exerted through both hands across six representative physical interaction tasks. The captured motions are retarget… view at source ↗
Figure 2
Figure 2. Figure 2: Real-world motion–force acquisition system and task set. (a) The operator uses a VR-based system for whole-body motion capture, while two force sensors coupled with 3D-printed hooks record the forces exerted through both hands and provide stable attachment points for object interaction. (b) Retargeted humanoid motions for the six force-interaction tasks: Clean Table (table wiping), Liftdown Water (lowering… view at source ↗
Figure 3
Figure 3. Figure 3: ThorArena benchmark evaluation pipeline. The benchmark replays retargeted reference motions and recorded two-hand forces in simulation, evaluates different whole-body control policies through a unified policy adapter, and reports FATS and complementary diagnostic metrics under force and no-force settings. episode i, the keypoint RMSE in the three regimes, E low i , E mid i , and E high i , is aggregated in… view at source ↗
Figure 4
Figure 4. Figure 4: Task-level FATS comparison with and without external forces. [PITH_FULL_IMAGE:figures/full_fig_p005_4.png] view at source ↗
Figure 6
Figure 6. Figure 6: Keypoint tracking errors across low-force and high-force segments. [PITH_FULL_IMAGE:figures/full_fig_p006_6.png] view at source ↗
Figure 5
Figure 5. Figure 5: Tracking score and survival distribution under external forces. [PITH_FULL_IMAGE:figures/full_fig_p006_5.png] view at source ↗
read the original abstract

Humanoid robots are increasingly expected to perform contact-rich tasks that require not only accurate whole-body motion but also robust physical interaction with surrounding objects and humans. Although recent advances in humanoid motion imitation and whole-body control have achieved remarkable tracking performance, existing datasets and benchmarks primarily focus on kinematic motion while largely overlooking synchronized interaction forces. As a result, current evaluations fail to capture how external interaction forces affect tracking accuracy, stability, and control robustness. In this paper, we present ThorArena, a benchmark for evaluating force-aware humanoid interaction based on human demonstrations with synchronized motion and force measurements. We collect a real-world interaction dataset that simultaneously captures whole-body human motion and forces exerted by both hands across six representative physical interaction tasks. Based on these demonstrations, we propose force-aware evaluation metrics that jointly assess whole-body tracking accuracy, robustness under different force levels, control effort, and episode survival through the Force-Aware Tracking Score (FATS) and complementary diagnostic metrics. We further establish a unified benchmark protocol that replays recorded interaction forces in simulation and provides a standardized evaluation interface for different humanoid control policies. Experiments on representative whole-body control policies demonstrate that force-aware evaluation reveals substantial performance differences that remain largely hidden under conventional no-force evaluation. ThorArena provides a practical and reproducible framework for studying force-aware humanoid interaction and offers a new benchmark for evaluating contact-rich humanoid behaviors.

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

3 major / 6 minor

Summary. The paper presents ThorArena, a benchmark for evaluating force-aware humanoid physical interaction. It contributes three components: (1) a real-world dataset of synchronized whole-body human motion and two-hand interaction forces across six contact-rich tasks (60 sequences per task, 360 total), (2) the Force-Aware Tracking Score (FATS) and complementary diagnostic metrics (robustness ratio, power overhead, survival rate), and (3) a simulation-based force-replay protocol with a unified policy-adapter interface that applies recorded interaction forces to simulated humanoid hand bodies during policy rollout. Four whole-body control policies (Thor2, TWIST2, GMT, SONIC) are evaluated under matched force and no-force conditions, showing wider performance spreads under external forces.

Significance. The synchronized motion-force dataset fills a genuine gap in the humanoid benchmarking literature, where existing resources provide kinematic trajectories but not paired interaction forces. The force-replay protocol and policy-adapter interface are practical design choices that lower the barrier to evaluating diverse policies under consistent disturbance conditions. The FATS metric provides a reasonable composite score that stratifies performance by force regime. The no-force vs. force comparison (Tables I–III) does demonstrate that survival rates and tracking spreads change under external forces, supporting the paper's motivation. The dataset is stated to be released, which adds reproducibility value.

major comments (3)
  1. §I, Contributions bullet 1; §IV, Tables I–III: The paper states that the paired motion-force data 'have been used to train the force-aware humanoid control policy Thor2,' but never specifies whether the evaluation sequences are disjoint from Thor2's training set. Thor2 ranks first on all six subtasks under external forces (Table I: FATS 81.71 vs. 70.04–73.78). The central claim that force-aware evaluation reveals 'substantial performance differences' is load-bearing on this comparison. If Thor2 trained on the same (or distributionally identical) force profiles that are replayed during evaluation, its advantage may reflect familiarity with the force characteristics rather than genuine force robustness. The paper should either (a) explicitly state the train/evaluation split and confirm no sequence overlap, or (b) include at least one policy not trained on this dataset as a control, or (c)退
  2. §III.B, Eq. (2)–(3): The FATS weights (w_low, w_mid, w_high) = (0.2, 0.3, 0.5) and σ = 0.15 m are hand-chosen without justification or sensitivity analysis. Since FATS is the primary metric and the relative ranking of policies could shift under different weight choices (e.g., if a policy degrades more in the mid-force regime than the high-force regime), a sensitivity analysis over at least 2–3 alternative weight configurations would strengthen the claim that the reported differences are robust to metric design choices rather than artifacts of the chosen parameters.
  3. §III.C, Force-replay protocol: The protocol applies recorded human sensor-frame forces to simulated robot hand bodies, but real contact dynamics during humanoid task execution differ — the robot's own motion changes contact location, timing, and force profile. The paper acknowledges this gap as future work (§V), but the central claim of revealing 'substantial performance differences' depends on whether these simulated force perturbations reflect real-world conditions. At minimum, the paper should discuss the expected direction and magnitude of the sim-to-real discrepancy (e.g., does open-loop force replay over- or under-estimate disturbance severity?) and ideally report results with varied force coefficients (currently only 0 and 1 are used) to characterize how sensitive the policy rankings are to force magnitude.
minor comments (6)
  1. §III.B, Eq. (4): The robustness ratio ρ = E_low/E_high is defined as RMSE_low/RMSE_high, but the text later says 'ρ → 1 indicates slower degradation.' This is correct but could confuse readers since a ratio of low/high error approaching 1 could also mean uniformly poor tracking. The paper partially addresses this in §IV.A ('its high ratio partly reflects a larger low-force baseline error'), but this caveat should appear at the metric definition, not only in the results discussion.
  2. Table II: The 'Low KP' and 'High KP' rows report values of 27.8/37.5/37.8/43.9 and 33.8/40.7/42.4/54.1 respectively. The units (mm) are stated in the text but not in the table header. Adding units to the table would improve readability.
  3. §III.A: The paper states 60 raw sequences per task but does not report the duration of each sequence, the sampling rate of the force sensors, or the synchronization accuracy between motion capture and force measurements. These details are important for reproducibility of the dataset.
  4. Fig. 4: The radar plots are small and the axis labels are difficult to read. Consider enlarging or providing a tabular companion.
  5. §IV.A: The paper mentions 'about 25% lower' for Thor2's keypoint error vs. the second best, but does not specify which policy is the second best in each regime. Specifying this would make the comparison more precise.
  6. §II: The related work section does not discuss any existing force-aware or contact-rich benchmarks in manipulation (e.g., benchmarks that measure interaction forces in grasping or pushing). Even if these are from a different subfield, briefly positioning ThorArena relative to force-aware manipulation benchmarks would strengthen the novelty claim.

Simulated Author's Rebuttal

3 responses · 1 unresolved

We thank the referee for the careful and constructive review. The referee raises three major points: (1) potential train/evaluation overlap for Thor2, (2) lack of sensitivity analysis for FATS hyperparameters, and (3) the sim-to-real gap in open-loop force replay. We agree with all three points and will address each in revision. Specifically, we will add an explicit train/evaluation split statement and include TWIST2, GMT, and SONIC as controls not trained on our dataset; add a sensitivity analysis over alternative FATS weight configurations; and add a discussion of expected sim-to-real discrepancy direction plus results with intermediate force coefficients. We cannot fully resolve the sim-to-real gap within this revision (physical robot validation is future work), which we note as a standing limitation.

read point-by-point responses

  1. Referee: Train/evaluation overlap for Thor2: the paper does not specify whether evaluation sequences are disjoint from Thor2's training set, and Thor2 ranks first on all subtasks. If Thor2 trained on the same force profiles, its advantage may reflect familiarity rather than genuine robustness.

    Authors: The referee is correct that this is a critical clarification. In the revised manuscript, we will explicitly state the train/evaluation split: the 360 collected sequences are partitioned such that the evaluation set used in ThorArena is disjoint from the data used to train Thor2. We will add this statement to both the Contributions section (§I) and the Experimental Setup (§IV). Furthermore, we note that TWIST2, GMT, and SONIC serve as the requested controls: none of these three policies were trained on our dataset, yet they are evaluated under the same force-replay protocol. The performance spread among these three policies (FATS 70.04–73.78 under external forces) already demonstrates that force-aware evaluation reveals differences independent of training-set familiarity. We will make this control structure explicit in the revision. That said, we acknowledge that Thor2's advantage over the other policies cannot be fully disentangled from its force-aware training: Thor2 was designed with force awareness in mind, so its superior performance under forces is expected and not necessarily a confound. The benchmark's purpose is to reveal such differences, not to claim that all policies should be force-naive. We will clarify this framing. revision: yes

  2. Referee: FATS weights (0.2, 0.3, 0.5) and σ = 0.15 m are hand-chosen without justification or sensitivity analysis. A sensitivity analysis over alternative weight configurations would strengthen the claim that reported differences are robust to metric design choices.

    Authors: We agree. The weight choice was motivated by the intuition that the high-force regime is most informative for force-aware evaluation, but this rationale was not stated in the manuscript and the robustness of rankings to this choice was not verified. In the revision, we will: (1) add a brief justification for the weight selection, (2) include a sensitivity analysis with at least three alternative configurations (e.g., uniform weights (1/3, 1/3, 1/3), low-force-emphasizing (0.5, 0.3, 0.2), and mid-force-emphasizing (0.2, 0.5, 0.3)), and (3) report whether policy rankings change under these alternatives. Based on our preliminary assessment, the ranking of Thor2 as top performer is robust because it achieves the lowest keypoint error in all three force regimes individually (Table II: Low KP 27.8, High KP 33.8, both best). However, the relative ordering of TWIST2, GMT, and SONIC may shift, and we will report this transparently. We will also add a brief note on σ sensitivity. revision: yes

  3. Referee: Open-loop force replay does not account for how robot motion changes contact dynamics. The paper should discuss expected direction/magnitude of sim-to-real discrepancy and report results with varied force coefficients (currently only 0 and 1).

    Authors: This is a fair and important point. We will address it in two ways. First, we will add a discussion of the expected sim-to-real discrepancy direction: open-loop force replay likely over-estimates disturbance severity in some respects (because a real robot's compliant motion would reduce peak contact forces through give) and under-estimates in others (because real contact involves friction, deformation, and intermittent contact that a fixed-force application does not capture). We will frame this as a known limitation of open-loop replay and note that closed-loop force adaptation is future work, as already acknowledged in §V. Second, we will add experiments with intermediate force coefficients (e.g., 0.25, 0.5, 0.75) to characterize how policy rankings and FATS scores vary with force magnitude. This will show whether the reported performance differences are specific to full-force replay or hold across a range of disturbance intensities. We cannot, within this revision, provide physical robot validation to directly measure the sim-to-real gap; this remains a standing limitation that we will state explicitly. revision: partial

standing simulated objections not resolved
  • We cannot provide direct sim-to-real validation within this revision. Physical robot experiments to quantify the discrepancy between open-loop force replay and real contact dynamics are planned as future work (§V) but are beyond the scope of the current submission. We will state this limitation explicitly in the revised manuscript.

Circularity Check

0 steps flagged

No significant circularity in the benchmark methodology; one minor self-citation concern (Thor2 trained on same dataset) that is not load-bearing for the central methodological claim.

full rationale

The paper's central claim is that force-aware evaluation (FATS + force replay) reveals performance differences hidden under no-force evaluation. This claim is supported empirically by running the same protocol on four policies and observing wider performance spreads under force (Table I: FATS 70.04–81.71) versus no-force (Table III: 78.43–84.31). The FATS metric (Eqs. 1–3) is defined independently of any policy's training data: it computes keypoint RMSE stratified by force regime, weighted by fixed design constants (w_low, w_mid, w_high) = (0.2, 0.3, 0.5) and σ = 0.15 m, combined with a survival factor. The force-replay protocol (§III.C) applies recorded sensor-frame forces to simulated hand bodies — this is a protocol definition, not a derivation that reduces to its inputs. No parameter is fitted to evaluation data and then presented as a prediction. No uniqueness theorem or self-citation chain is invoked to force the conclusion. The one concern is that Thor2, the authors' own policy, was trained on the same motion-force dataset used for benchmark evaluation (§I: 'The paired motion–force data have been used to train the force-aware humanoid control policy Thor2'), and Thor2 ranks first on all six subtasks. This is a legitimate train-on-test bias concern (correctness risk), but it does not make the benchmark methodology circular: the metric, protocol, and evaluation interface are defined independently of Thor2, and the claim about force-aware evaluation revealing hidden differences holds even among the three non-Thor2 policies (survival drops to 0.73–0.81 under force on push chair for all three, while near 1.0 under no-force). The self-citation is minor and does not undermine the central methodological contribution.

Axiom & Free-Parameter Ledger

3 free parameters · 3 axioms · 0 invented entities

The paper introduces no new physical entities, particles, forces, or dimensions. The free parameters are metric design choices (weights, sensitivity, thresholds) rather than physical constants fitted to data. The axioms are domain assumptions about the validity of simulation-based force replay and metric design, all of which are standard in robotics benchmarking but carry unstated consequences.

free parameters (3)
  • FATS force-regime weights (w_low, w_mid, w_high) = (0.2, 0.3, 0.5)
    Hand-chosen to emphasize high-force regime; no sensitivity analysis or justification provided beyond 'emphasizes the high-force regime' (§III.B).
  • FATS error sensitivity σ = 0.15m
    Controls the exponential decay of score with tracking error; chosen without stated justification or comparison to alternatives (§III.B).
  • Force-regime thresholds (33rd, 66th percentiles) = 33rd and 66th percentiles
    Used to stratify timesteps into low/medium/high force; the choice of tertiles is conventional but unstated in rationale (§III.B).
axioms (3)
  • domain assumption Recorded human hand forces can be validly replayed as external disturbances on simulated humanoid hand bodies
    The benchmark's central evaluation protocol depends on this assumption (§III.C). The paper acknowledges it needs real-robot validation (§V) but proceeds with simulation-only results.
  • domain assumption Root-relative keypoint error is the appropriate primary tracking metric
    Reference keypoints are re-anchored to the robot's root pose at every timestep (§III.B), making the metric invariant to global drift. Global drift is reported separately as a diagnostic. This is a standard choice in humanoid tracking but affects all reported numbers.
  • domain assumption The six selected tasks are representative of contact-rich humanoid interaction
    The task set (table wiping, object lifting/lowering, chair pushing/pulling, cooperative carrying) covers pushing, pulling, lifting, and cooperative patterns, but the paper does not argue for exhaustiveness or representativeness beyond listing the tasks (§III.A).

pith-pipeline@v1.1.0-glm · 12428 in / 2754 out tokens · 469333 ms · 2026-07-08T17:59:32.699131+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

27 extracted references · 27 canonical work pages · 12 internal anchors

  1. [1]

    Learning human-to-humanoid real-time whole-body teleoperation,

    T. He, Z. Luo, W. Xiao, C. Zhang, K. Kitani, C. Liu, and G. Shi, “Learning human-to-humanoid real-time whole-body teleoperation,” in2024 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS). IEEE, 2024, pp. 8944–8951

    work page 2024

  2. [2]

    HumanPlus: Humanoid Shadowing and Imitation from Humans

    Z. Fu, Q. Zhao, Q. Wu, G. Wetzstein, and C. Finn, “Humanplus: Humanoid shadowing and imitation from humans,”arXiv preprint arXiv:2406.10454, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  3. [3]

    OmniH2O: Universal and Dexterous Human-to-Humanoid Whole-Body Teleoperation and Learning

    T. He, Z. Luo, X. He, W. Xiao, C. Zhang, W. Zhang, K. Kitani, C. Liu, and G. Shi, “Omnih2o: Universal and dexterous human- to-humanoid whole-body teleoperation and learning,”arXiv preprint arXiv:2406.08858, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  4. [4]

    HOMIE: Humanoid Loco-Manipulation with Isomorphic Exoskeleton Cockpit

    Q. Ben, F. Jia, J. Zeng, J. Dong, D. Lin, and J. Pang, “Homie: Humanoid loco-manipulation with isomorphic exoskeleton cockpit,” arXiv preprint arXiv:2502.13013, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  5. [5]

    TWIST: Teleoperated Whole-Body Imitation System

    Y . Ze, Z. Chen, J. P. Ara ´ujo, Z. ang Cao, X. B. Peng, J. Wu, and C. K. Liu, “Twist: Teleoperated whole-body imitation system,”arXiv preprint arXiv:2505.02833, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  6. [6]

    Twist2: Scalable, portable, and holistic humanoid data collection system,

    Y . Ze, S. Zhao, W. Wang, A. Kanazawa, R. Duan, P. Abbeel, G. Shi, J. Wu, and C. K. Liu, “Twist2: Scalable, portable, and holistic humanoid data collection system,”arXiv preprint arXiv:2511.02832, 2025

    work page arXiv 2025

  7. [7]

    Clone: Closed-loop whole-body humanoid teleoperation for long- horizon tasks,

    Y . Li, Y . Lin, J. Cui, T. Liu, W. Liang, Y . Zhu, and S. Huang, “Clone: Closed-loop whole-body humanoid teleoperation for long- horizon tasks,” 2025

    work page 2025

  8. [8]

    Clot: Closed-loop global motion tracking for whole-body humanoid teleoperation,

    T. Zhu, G. Cai, Y . Zhaohui, G. Ren, H. Xie, Z. Wang, J. Wu, J. Wang, X. Yang, Y . Muet al., “Clot: Closed-loop global motion tracking for whole-body humanoid teleoperation,”arXiv preprint arXiv:2602.15060, 2026

    work page arXiv 2026

  9. [9]

    Perpetual humanoid control for real-time simulated avatars,

    Z. Luo, J. Cao, A. W. Winkler, K. Kitani, and W. Xu, “Perpetual humanoid control for real-time simulated avatars,” inInternational Conference on Computer Vision (ICCV), 2023

    work page 2023

  10. [10]

    Retargeting matters: General motion retargeting for humanoid motion tracking,

    J. P. Araujo, Y . Ze, P. Xu, J. Wu, and C. K. Liu, “Retargeting matters: General motion retargeting for humanoid motion tracking,” arXiv preprint arXiv:2510.02252, 2025

    work page arXiv 2025

  11. [11]

    OmniRetarget: Interaction-Preserving Data Generation for Humanoid Whole-Body Loco-Manipulation and Scene Interaction

    L. Yang, X. Huang, Z. Wu, A. Kanazawa, P. Abbeel, C. Sferrazza, C. K. Liu, R. Duan, and G. Shi, “Omniretarget: Interaction-preserving data generation for humanoid whole-body loco-manipulation and scene interaction,”arXiv preprint arXiv:2509.26633, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  12. [12]

    Hover: Versatile neural whole-body controller for humanoid robots,

    T. He, W. Xiao, T. Lin, Z. Luo, Z. Xu, Z. Jiang, J. Kautz, C. Liu, G. Shi, X. Wanget al., “Hover: Versatile neural whole-body controller for humanoid robots,” in2025 IEEE International Conference on Robotics and Automation (ICRA). IEEE, 2025, pp. 9989–9996

    work page 2025

  13. [13]

    Amo: Adaptive motion optimization for hyper-dexterous humanoid whole- body control,

    J. Li, X. Cheng, T. Huang, S. Yang, R. Qiu, and X. Wang, “Amo: Adaptive motion optimization for hyper-dexterous humanoid whole- body control,”Robotics: Science and Systems 2025, 2025

    work page 2025

  14. [14]

    GMT: General Motion Tracking for Humanoid Whole-Body Control

    Z. Chen, M. Ji, X. Cheng, X. Peng, X. B. Peng, and X. Wang, “Gmt: General motion tracking for humanoid whole-body control,”arXiv preprint arXiv:2506.14770, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  15. [15]

    SONIC: Supersizing Motion Tracking for Natural Humanoid Whole-Body Control

    Z. Luo, Y . Yuan, T. Wang, C. Li, S. Chen, F. Casta ˜neda, Z.-A. Cao, J. Li, D. Minor, Q. Ben, X. Da, R. Ding, C. Hogg, L. Song, E. Lim, E. Jeong, T. He, H. Xue, W. Xiao, Z. Wang, S. Yuen, J. Kautz, Y . Chang, U. Iqbal, L. Fan, and Y . Zhu, “Sonic: Supersizing motion tracking for natural humanoid whole-body control,”arXiv preprint arXiv:2511.07820, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  16. [16]

    Smpl: A skinned multi-person linear model,

    M. Loper, N. Mahmood, J. Romero, G. Pons-Moll, and M. J. Black, “Smpl: A skinned multi-person linear model,” inSeminal Graphics Papers: Pushing the Boundaries, V olume 2, 2023, pp. 851–866

    work page 2023

  17. [17]

    Amass: Archive of motion capture as surface shapes,

    N. Mahmood, N. Ghorbani, N. F. Troje, G. Pons-Moll, and M. J. Black, “Amass: Archive of motion capture as surface shapes,” in The IEEE International Conference on Computer Vision (ICCV), Oct

    work page

  18. [18]

    Available: https://amass.is.tue.mpg.de

    [Online]. Available: https://amass.is.tue.mpg.de

    work page

  19. [19]

    Robust mo- tion in-betweening,

    F. G. Harvey, M. Yurick, D. Nowrouzezahrai, and C. Pal, “Robust mo- tion in-betweening,”ACM Transactions on Graphics (TOG), vol. 39, no. 4, pp. 60–1, 2020

    work page 2020

  20. [20]

    Learning from Massive Human Videos for Universal Humanoid Pose Control

    J. Mao, S. Zhao, S. Song, T. Shi, J. Ye, M. Zhang, H. Geng, J. Malik, V . Guizilini, and Y . Wang, “Learning from massive human videos for universal humanoid pose control,”arXiv preprint arXiv:2412.14172, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  21. [21]

    PHUMA: Physically Reliable Humanoid Locomotion Dataset

    K. Lee, S. Kim, Y . Lee, M. Park, H. Kim, D. Hwang, D. Kim, H. Lee, and J. Choo, “Phuma: Physically reliable humanoid locomo- tion dataset,”arXiv preprint arXiv:2510.26236, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  22. [22]

    Humoto: A 4d dataset of mocap human object interactions,

    J. Lu, C.-H. P. Huang, U. Bhattacharya, Q. Huang, and Y . Zhou, “Humoto: A 4d dataset of mocap human object interactions,” in Proceedings of the IEEE/CVF International Conference on Computer Vision (ICCV), October 2025, pp. 10 886–10 897

    work page 2025

  23. [23]

    Humanoid Everyday: A Comprehensive Robotic Dataset for Open-World Humanoid Manipulation

    Z. Zhao, H. Jing, X. Liu, J. Mao, A. Jha, H. Yang, R. Xue, S. Zakharor, V . Guizilini, and Y . Wang, “Humanoid everyday: A comprehensive robotic dataset for open-world humanoid manipulation,”arXiv preprint arXiv:2510.08807, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  24. [24]

    Mimicking-Bench: A Benchmark for Generalizable Humanoid-Scene Interaction Learning via Human Mimicking

    Y . Liu, B. Yang, L. Zhong, H. Wang, and L. Yi, “Mimicking-bench: A benchmark for generalizable humanoid-scene interaction learning via human mimicking,”arXiv preprint arXiv:2412.17730, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  25. [25]

    Distinguishing Imitation Error from Intrinsic Motion Learning Difficulty

    Z. Meng, L. Yin, X. Chen, A. Chen, S. Guo, and Y . Qin, “Bench- marking humanoid imitation learning with motion difficulty,”arXiv preprint arXiv:2512.07248, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  26. [26]

    Omniclone: Engineering a robust, all- rounder whole-body humanoid teleoperation system,

    Y . Li, L. Ma, Y . Lin, Y . Du, M. Liu, K. Hu, J. Cui, Y . Zhu, W. Liang, B. Jia, and S. Huang, “Omniclone: Engineering a robust, all- rounder whole-body humanoid teleoperation system,”arXiv preprint arXiv:2603.14327, 2026

    work page arXiv 2026

  27. [27]

    Towards motion turing test: Evaluating human-likeness in humanoid robots,

    M. Li, M. Liu, Z. Wu, X. Lin, J. Zhang, M. Yan, Z. Xie, C. Zhang, C. Wen, L. Xuet al., “Towards motion turing test: Evaluating human-likeness in humanoid robots,” inProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, 2026, pp. 16 486–16 498

    work page 2026