Pith. sign in

REVIEW 3 major objections 6 minor 7 cited by

From Experts to a Generalist: Toward General Whole-Body Control for Humanoid Robots

T0 review · 3 major / 6 minor · reviewed 2026-08-07 · deepseek-v4-flash

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

Pith's one-line read One general policy tracks 135 seconds of varied humanoid motion.

desk verdict A solid empirical systems paper with a genuine new integration - clustering, per-cluster delta models, and distillation - where the main issues are an overclaim, an internal count inconsistency, and an unresolved odometry-drift caveat. read the letter →

arxiv 2506.12779 v3 pith:JQKUAWER submitted 2025-06-15 cs.RO cs.LG

Yuxuan Wang , Ming Yang , Ziluo Ding , Yu Zhang , Weishuai Zeng , Xinrun Xu , Haobin Jiang , Zongqing Lu This is my paper · ORCID
classification cs.ROcs.LG
keywords whole-bodycontrolhumanoidrobotsexpert-to-generalistmotionclusteringsim-to-realtransferdeltaactionmodelknowledgedistillationreinforcementlearning
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

The paper claims that a single controller can handle a wide range of humanoid whole-body motions—jumping, walking, standing, and more—if the motion data is first clustered into behaviorally similar groups, experts are trained per cluster, corrected with real-world data, and then distilled into one generalist policy. The proposed pipeline, BumbleBee, reports success rates of 89.58% in IsaacGym and 66.84% in MuJoCo, higher than three prior whole-body controllers, and demonstrates a 135-second real-world tracking sequence. The paper's core claim is that grouping motions before training resolves the conflicting gradients that arise when a policy is trained on a mixed dataset, and that expert-specific sim-to-real correction beats a single correction model. A sympathetic reader would care because this offers a concrete path toward agile, general-purpose humanoid control rather than a collection of single-skill policies.

What carries the argument

The central mechanism is an expert-to-generalist pipeline built on three stages. First, an autoencoder with Transformer encoders maps motion sequences (joint positions, root translation and velocity, foot contact states, and foot velocities) alongside text descriptions into a shared latent space, and K-means groups the motions into six clusters. Second, expert MLP tracking policies are initialized from a general tracking policy and fine-tuned within each cluster, then each expert is refined iteratively with a cluster-specific delta action model $\pi_\Delta(s_t,a_t)$ that corrects the simulated action so $s_{t+1} = f^{\text{sim}}(s_t, a_t + \pi_\Delta(s_t,a_t))$ approximates real-world transitions. Third, the experts are distilled into a single generalist using DAgger with a KL divergence loss, and the final generalist uses a Gated Transformer-XL architecture to absorb the diverse expert behaviors.

What would settle it

Instrument the robot with a motion capture or high-precision localization system during delta-action data collection and compare the learned corrections with the odometry-based ones: if the odometry-based corrections move the simulated root away from the motion-capture reference over time, the central real-world improvement claim fails.

Watch

Extended reading notes

Core claim

BumbleBee's central claim is that decomposing a heterogeneous motion dataset into semantically and kinematically coherent clusters, training specialized expert policies, refining each expert with cluster-specific delta action models, and distilling all experts into a Transformer-based generalist produces a whole-body controller that outperforms both a single policy trained on everything and experts derived from randomly split clusters. The ablation evidence is load-bearing: random clustering gives no benefit over the general init in MuJoCo (35.36% versus 33.01% success), while the full BumbleBee pipeline reaches 66.84%. Cluster-specific delta models also beat a single general delta model, notably on the jump cluster (68.92% versus 50.71%). The authors interpret these results as showing that meaningful motion clustering reduces cross-task interference and that per-cluster delta correction more accurately bridges the simulation-to-real gap.

Load-bearing premise

The delta-action correction assumes that the robot's onboard odometry gives an accurate enough estimate of root translation to serve as real-world ground truth; if that estimate drifts during long reference motions, the learned corrections will be wrong and iterative fine-tuning could amplify the error.

Editorial extensions

If this is right

  • A single generalist can be trained by first specializing, and distillation can preserve or even exceed expert performance: the final generalist sometimes outperforms individual experts on the hardest clusters, such as jump and walk-slow.
  • Iterative delta fine-tuning raises expert success rates monotonically across iterations, from 51.49% at iteration 0 to 60.33% at iteration 1 and 70.37% at iteration 2 in MuJoCo.
  • Cluster-specific delta action models are necessary: a general delta model trained on all real-world data improves two of three tested clusters but degrades the jump cluster (50.71% versus 68.92% for the expert-specific model).
  • Real-world foot stability improves measurably with each delta fine-tuning iteration, supporting the claim that the iterative loop transfers benefits beyond simulation.
  • The pipeline's design implies that motion-type interference, not policy capacity alone, is a key bottleneck for general whole-body control.

Reading between the lines

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

  • If odometry bias is the main limitation the paper names, adding high-precision localization such as visual-inertial odometry could further improve real-world tracking, especially for long sequences where drift accumulates.
  • The cluster-then-distill recipe is a transferable design for other multi-skill robot control problems: partition the behavior space, adapt each partition to real dynamics, then compress into one policy.
  • A testable extension is whether the chosen six clusters generalize to new motion categories; the elbow-based choice of K is tied to this specific AMASS-derived dataset, so new datasets may need a different partition.
  • The monotone improvement across delta iterations suggests a possible stopping rule based on marginal gain, which the authors did not explore because of limited compute.
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

3 major / 6 minor

Summary. The paper proposes BumbleBee (BB), a framework for training a general whole-body controller for humanoid robots. The pipeline first clusters a filtered AMASS motion dataset with an autoencoder that combines kinematic features and text descriptions, then trains per-cluster expert tracking policies, refines them with iterative delta-action models to bridge the sim-to-real gap, and finally distills the experts into a single Transformer-based generalist policy. Experiments are conducted in IsaacGym, MuJoCo, and on a Unitree G1 robot, with comparisons against OmniH2O, Exbody2, and Hover. The central claims are that BB achieves state-of-the-art whole-body control performance and that the expert-to-generalist clustering plus per-cluster delta-action adaptation is the source of this advantage.

Significance. If substantiated, the proposed framework is a useful contribution to humanoid whole-body control, combining motion clustering, sim-to-real delta-action adaptation, and expert distillation in a single pipeline. The paper earns credit for reporting confidence intervals in Table 12 and for including a random-cluster ablation in Table 3, both of which are appropriate controls. However, the real-world evidence is qualitative only (Figures 1 and 5), and the delta-action ground truth relies on onboard odometry rather than motion capture, creating a risk of bias that the authors themselves acknowledge in Section 5. The claim of state-of-the-art real-world performance is therefore not yet fully established by the presented evidence.

major comments (3)
  1. [§4.3, Table 1] The text states that "BB outperforms all baselines across nearly all evaluation metrics," but in MuJoCo BB's MPKPE is 294.27 ± 7.923, which is worse than Exbody2's 272.42 ± 7.029. Since MPKPE is one of the three headline metrics, this claim is contradicted by the paper's own table. The authors should qualify the claim or explicitly justify why success rate is the only metric that matters for the superiority statement.
  2. [§3.2.2, Appendix B.1, Section 5] The delta-action models are trained with a root-position reward (Table 8) whose reference comes from onboard odometry because no motion capture system was used. The paper concedes in Section 5 that this "may introduce biases," but the experimental section does not quantify odometry drift or provide any independent localization check. Since iterative delta-action fine-tuning is the mechanism claimed to be responsible for real-world improvement, this is load-bearing. If odometry drifts during a rollout, the delta model treats that drift as real dynamics and the iterative loop can amplify the bias. The real-world results are qualitative only, so the claim that BB achieves state-of-the-art real-world performance is not yet supported. Please provide either a drift analysis for the collected trajectories or quantitative real-world tracking metrics that do not depend on the same odometry reference.
  3. [§4.4, Tables 3 and 4] The iterative-improvement and clustering-ablation results are reported as point estimates without confidence intervals or seed information. Table 4 reports mean success rates across iterations, and Table 3 compares a single "Random" split against General Init and BB. The random-split ablation, in particular, does not report variance across multiple random splits, so the conclusion that random partitioning offers no advantage rests on a single draw. Please report means and standard deviations over at least three seeds or random splits for these central claims.
minor comments (6)
  1. [Abstract and §1] The contributions describe an "auto-regressive clustering method," but the method section and Figure 2 describe an autoencoder-based clustering approach. The term appears again in the Conclusion; please use consistent terminology.
  2. [Appendix A.3] MPJPE is defined as an average position error over joints, yet the stated unit is radians. This is inconsistent; either the definition is actually an angle error or the unit should be a length unit.
  3. [Table 12] The confidence intervals in Table 12 are computed over samples on a single reference trajectory, so they do not characterize variance across the test motion set. Reporting statistics across motions would better support the generalization claims.
  4. [§4.2, Table 2] The cluster names (Jump, Walk-slow, etc.) are inferred from post-hoc kinematic summaries and keyword lists. It would be clearer to state explicitly that these labels are assigned by the authors after clustering, not produced by the clustering algorithm itself.
  5. [Figure 6] The legend and color scheme are difficult to parse because each subplot repeats the same legend with different orderings; a unified legend or a table would improve readability.
  6. [Appendix A.2] There is a typo: "roportional derivative" should be "proportional derivative."

Circularity Check

1 steps flagged · score 2.0 of 10

Minor self-referential cluster validation; the empirical expert-to-generalist pipeline itself is not circular.

  1. self definitional [Section 3.1.2 (AE Clustering, Lcluster) and Section 4.2 (Cluster Analysis)]
    "Our method not only relies solely on autoencoders to reconstruct full motion sequences, but also incorporates textual annotations to align motions at the semantic level. ... Lcluster = LInfoNCE(zl, zm) +L2(zl, zm) +Lhuber( ˆM l, M) +Lhuber( ˆM m, M). ... From the semantic perspective, we extracted the top keywords from each cluster and find that their meanings closely align with the corresponding kinematic characteristics, demonstrating that the clusters are semantically meaningful."

    The text descriptions are an input to the clustering objective via the InfoNCE alignment term, so the latent space is explicitly trained to place semantically similar motions together. Reporting that cluster keywords align with cluster kinematics is therefore reading back the same annotations used to construct the clusters; the alignment is partly enforced by the loss rather than independently discovered. This validation is supporting evidence for the clustering step, not the central claim of superior whole-body control, so it is a minor self-referential step rather than a fatal circularity.

full rationale

BumbleBee is an empirical pipeline rather than a formal derivation. The main chain -- motion clustering, per-cluster expert RL, real-world delta-action fine-tuning (following ASAP), and DAgger distillation into a Transformer generalist -- is self-contained: the delta models are fit to real rollouts, and the final MuJoCo/IsaacGym numbers are head-to-head comparisons against external baselines (OmniH2O, Exbody2, Hover) trained on the same filtered AMASS data. No load-bearing claim reduces to a fitted parameter renamed as a prediction, and no uniqueness theorem or central premise is imported from the authors' own prior work; the Jaeger citation [25] is incidental related work. The only self-referential aspect is the semantic validation of clusters, where the same text annotations used in the clustering loss are later used as evidence of semantic meaningfulness. The odometry-based root translation in the delta reward (Appendix B.1, Table 8) is a real correctness risk for the sim-to-real claim, but it is a measurement limitation, not a circular derivation.

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

The paper's results rest on standard RL and imitation learning machinery. The hand-set choices (K, termination thresholds, delta action DoF) are the main free parameters. No new physical entities are introduced.

free parameters (4)
  • Number of clusters K = 6
    Selected by elbow method on within-cluster sum of squares; directly determines expert specialization and downstream distillation.
  • Test termination keypoint threshold = 0.8 for walking, 0.4 for other tasks
    Hand-set thresholds define Success Rate; tuning these changes the headline SR numbers.
  • Delta action DoF = 4 ankle joints only
    Inherited from ASAP; restricts the sim-to-real correction to ankles, an assumption about where model error concentrates.
  • Termination curriculum threshold = 0.8 down to 0.3 during training
    Hand-set curriculum schedule for the keypoint termination condition; affects training difficulty and success.
assumptions (4)
  • domain assumption The filtered AMASS dataset with PHC filtering contains 8,179 high-quality trajectories representative of diverse human motions.
    Section 3.1.1; the whole pipeline depends on this dataset being a valid training distribution for a humanoid whole-body controller.
  • domain assumption HumanML3D text annotations align with the AMASS sequences and provide semantic information that improves clustering.
    Section 3.1.2; text alignment is used in the clustering loss, but coverage gaps in HumanML3D are not discussed.
  • domain assumption The learned delta action model, when added to simulator actions, reproduces real-world dynamics closely enough to fine-tune the policy.
    Section 3.2.2; this is the core sim-to-real transfer assumption inherited from ASAP.
  • domain assumption The Gated Transformer-XL policy architecture has sufficient capacity to distill six expert policies without losing performance.
    Section 3.3; the paper states MLP capacity was insufficient, but does not provide an architecture study.

how reviews work

0 comments
Cite this review

Pith. "Pith review of From Experts to a Generalist: Toward General Whole-Body Control for Humanoid Robots." pith.science (2026). https://pith.science/paper/JQKUAWER

@misc{pith2026250612779,
  author       = {Pith},
  title        = {Pith review of: From Experts to a Generalist: Toward General Whole-Body Control for Humanoid Robots},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/JQKUAWER}},
  note         = {Machine review of arXiv:2506.12779}
}
read the original abstract

Achieving general agile whole-body control on humanoid robots remains a major challenge due to diverse motion demands and data conflicts. While existing frameworks excel in training single motion-specific policies, they struggle to generalize across highly varied behaviors due to conflicting control requirements and mismatched data distributions. In this work, we propose BumbleBee (BB), an expert-generalist learning framework that combines motion clustering and sim-to-real adaptation to overcome these challenges. BB first leverages an autoencoder-based clustering method to group behaviorally similar motions using motion features and motion descriptions. Expert policies are then trained within each cluster and refined with real-world data through iterative delta action modeling to bridge the sim-to-real gap. Finally, these experts are distilled into a unified generalist controller that preserves agility and robustness across all motion types. Experiments on two simulations and a real humanoid robot demonstrate that BB achieves state-of-the-art general whole-body control, setting a new benchmark for agile, robust, and generalizable humanoid performance in the real world. The project webpage is available at https://beingbeyond.github.io/BumbleBee/.

Figures

Figures reproduced from arXiv: 2506.12779 by the authors.

Figure 1
Figure 1. A general whole-body control policy in the real world tracks the motion of two consecutive long [PITH_FULL_IMAGE:figures/full_fig_p001_1.png] view at source ↗
Figure 2
Figure 2. Overview of the BumbleBee framework. The left section illustrates the data curation stage, which [PITH_FULL_IMAGE:figures/full_fig_p004_2.png] view at source ↗
Figure 3
Figure 3. Elbow method showing the trade-off between cluster number and within-cluster sum of squares, with K = 6 selected as our selected point. Whether the clustering results are meaningful in terms of both kinematics and semantics? [PITH_FULL_IMAGE:figures/full_fig_p007_3.png] view at source ↗
Figures from the paper (3 more)
Figure 4
Figure 4. Figure 4: Evaluation of expert vs. generalist models in MuJoCo, measured by success rate. [PITH_FULL_IMAGE:figures/full_fig_p008_4.png]
Figure 5
Figure 5. Figure 5: Visualization of expert performance across iterations in the real world. We deploy the [PITH_FULL_IMAGE:figures/full_fig_p009_5.png]
Figure 6
Figure 6. Figure 6: Complete Iterative Comparison [PITH_FULL_IMAGE:figures/full_fig_p017_6.png]

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 7 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. EgoHTR: Egocentric 4D Demonstrations of Human Terrain Traversal

    cs.RO 2026-07 conditional novelty 6.0 of 10

    EgoHTR is a 55-sequence, 150k-frame egocentric 4D human-terrain dataset with a reconstruction pipeline, MoCap-validated benchmark, and perceptive locomotion policies deployed on a Unitree G1.

  2. Behavior Foundations for Quadruped Robots: ABot-C0 Technical Report

    cs.RO 2026-07 conditional novelty 6.0 of 10

    A multi-source 16,074-clip quadruped motion library plus a flow-matching generalist tracker shows empirical data scaling and zero-shot unseen tracking, integrated with all-terrain locomotion and real-robot deployment.

  3. Athena-WBC: Capability-Aligned Policy Experts for Long-Tail Humanoid Whole-Body Control

    cs.RO 2026-07 conditional novelty 5.5 of 10

    Capability-aligned dynamic and balance experts recover residual humanoid whole-body tracking failures better than data reallocation alone, then distill into one stronger deployable controller.

  4. ComplexMimic: Human-Scene Interaction Imitation in Complex 3D Environments

    cs.CV 2026-07 unverdicted novelty 5.5 of 10

    Dual-expert RL plus difficulty-aware multi-teacher distillation improves physics-based human–scene interaction imitation under complex 3D geometry versus prior single-policy baselines.

  5. SkillPlug: Unsupervised Skill Mining for Few-Shot Adaptation in Robotic Manipulation

    cs.RO 2026-07 conditional novelty 5.0 of 10

    Unsupervised skill mining with self-supervised compactness, alignment, and disentanglement losses yields a fixed skill library that improves multi-task and few-shot robotic manipulation when plugged into ACT and OpenVLA-OFT.

  6. RobotDancing: Residual-Action Reinforcement Learning Enables Robust Long-Horizon Humanoid Motion Tracking

    cs.RO 2025-09 conditional novelty 5.0 of 10

    Residual-action reinforcement learning, with selective corrections on hip and knee pitch joints, enables zero-shot long-horizon dance tracking on real humanoid robots.

  7. Integrating Diffusion-based Multi-task Learning with Online Reinforcement Learning for Robust Quadruped Robot Control

    cs.RO 2025-07 conditional novelty 5.0 of 10

    A diffusion policy pretrained on offline gait data and then finetuned with PPO achieves robust language-conditioned quadruped control with 50 Hz onboard inference.

Reference graph

Works this paper leans on

45 extracted references · 25 canonical work pages · cited by 7 Pith papers

  1. [1]

    Deep learning for detecting robotic grasps.The International Journal of Robotics Research, 34(4-5):705–724, 2015

    Ian Lenz, Honglak Lee, and Ashutosh Saxena. Deep learning for detecting robotic grasps.The International Journal of Robotics Research, 34(4-5):705–724, 2015

    work page 2015

  2. [2]

    Robotic grasping using deep reinforcement learning

    Shirin Joshi, Sulabh Kumra, and Ferat Sahin. Robotic grasping using deep reinforcement learning. In2020 IEEE 16th International Conference on Automation Science and Engineering (CASE), pages 1461–1466. IEEE, 2020

    work page 2020

  3. [3]

    Graspit! a versatile simulator for robotic grasping.IEEE Robotics & Automation Magazine, 11(4):110–122, 2004

    Andrew T Miller and Peter K Allen. Graspit! a versatile simulator for robotic grasping.IEEE Robotics & Automation Magazine, 11(4):110–122, 2004

    work page 2004

  4. [4]

    Guoguang Du, Kai Wang, Shiguo Lian, and Kaiyong Zhao. Vision-based robotic grasping from object localization, object pose estimation to grasp estimation for parallel grippers: a review.Artificial IntelligenceReview, 54(3):1677– 1734, 2021

    work page 2021

  5. [5]

    Unigrasp: Learning a unified model to grasp with multifingered robotic hands

    Lin Shao, Fabio Ferreira, Mikael Jorda, Varun Nambiar, Jianlan Luo, Eugen Solowjow, Juan Aparicio Ojea, Oussama Khatib, and Jeannette Bohg. Unigrasp: Learning a unified model to grasp with multifingered robotic hands. IEEE Robotics and Automation Letters, 5(2):2286–2293, 2020

    work page 2020

  6. [6]

    Coordination of multiple mobile robots in an object carrying task using implicit communication

    Guilherme AS Pereira, Bruno S Pimentel, Luiz Chaimowicz, and Mário FM Campos. Coordination of multiple mobile robots in an object carrying task using implicit communication. InProceedings 2002 IEEE International Conference on Robotics and Automation (Cat. No. 02CH37292), volume 1, pages 281–286. IEEE, 2002. 10

    work page 2002

  7. [7]

    Sim-to-real learning for humanoid box loco-manipulation

    Jeremy Dao, Helei Duan, and Alan Fern. Sim-to-real learning for humanoid box loco-manipulation. In2024 IEEE International Conference on Robotics and Automation (ICRA), pages 16930–16936. IEEE, 2024

    work page 2024

  8. [8]

    Human-humanoid collaborative carrying.IEEE Transactions on Robotics, 35(4):833–846, 2019

    Don Joven Agravante, Andrea Cherubini, Alexander Sherikov, Pierre-Brice Wieber, and Abderrahmane Kheddar. Human-humanoid collaborative carrying.IEEE Transactions on Robotics, 35(4):833–846, 2019

    work page 2019

Show all 45 references
  1. [9]

    Asap: Aligning simulation and real-world physics for learning agile humanoid whole-body skills

    Tairan He, Jiawei Gao, Wenli Xiao, Yuanhang Zhang, Zi Wang, Jiashun Wang, Zhengyi Luo, Guanqi He, Nikhil Sobanbabu, Chaoyi Pan, Zeji Yi, Guannan Qu, Kris Kitani, Jessica Hodgins, Linxi "Jim" Fan, Yuke Zhu, Changliu Liu, and Guanya Shi. Asap: Aligning simulation and real-world ...

    2025 arXiv

  2. [10]

    Perceiver-actor: Amulti-tasktransformerforroboticmanipulation

    MohitShridhar, LucasManuelli, andDieterFox. Perceiver-actor: Amulti-tasktransformerforroboticmanipulation. In Conference on Robot Learning, pages 785–799. PMLR, 2023

    2023

  3. [11]

    Transformer-based deep imitation learning for dual-arm robot manipulation

    Heecheol Kim, Yoshiyuki Ohmura, and Yasuo Kuniyoshi. Transformer-based deep imitation learning for dual-arm robot manipulation. In 2021 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 8965–8972. IEEE, 2021

    2021

  4. [12]

    Rt-1: Robotics transformer for real-world control at scale.arXiv preprint arXiv:2212.06817, 2022

    Anthony Brohan, Noah Brown, Justice Carbajal, Yevgen Chebotar, Joseph Dabis, Chelsea Finn, Keerthana Gopalakrishnan, Karol Hausman, Alex Herzog, Jasmine Hsu, et al. Rt-1: Robotics transformer for real-world control at scale.arXiv preprint arXiv:2212.06817, 2022

    2022 arXiv

  5. [13]

    Humanoid locomotion as next token prediction

    Ilija Radosavovic, Jathushan Rajasegaran, Baifeng Shi, Bike Zhang, Sarthak Kamat, Koushil Sreenath, Trevor Darrell, and Jitendra Malik. Humanoid locomotion as next token prediction. InThe Thirty-eighth Annual Conference on Neural Information Processing Systems, 2024

    2024

  6. [14]

    Exbody2: Advanced expressive humanoid whole-body control.arXiv preprint arXiv:2412.13196, 2024

    Mazeyu Ji, Xuanbin Peng, Fangchen Liu, Jialong Li, Ge Yang, Xuxin Cheng, and Xiaolong Wang. Exbody2: Advanced expressive humanoid whole-body control.arXiv preprint arXiv:2412.13196, 2024

    2024 arXiv

  7. [15]

    Outrageously large neural networks: The sparsely-gated mixture-of-experts layer.arXiv preprintarXiv:1701.06538, 2017

    Noam Shazeer, Azalia Mirhoseini, Krzysztof Maziarz, Andy Davis, Quoc Le, Geoffrey Hinton, and Jeff Dean. Outrageously large neural networks: The sparsely-gated mixture-of-experts layer.arXiv preprintarXiv:1701.06538, 2017

    2017 arXiv

  8. [16]

    Gshard: Scaling giant models with conditional computation and automatic sharding, 2020

    Dmitry Lepikhin, HyoukJoong Lee, Yuanzhong Xu, Dehao Chen, Orhan Firat, Yanping Huang, Maxim Krikun, Noam Shazeer, and Zhifeng Chen. Gshard: Scaling giant models with conditional computation and automatic sharding, 2020

    2020

  9. [17]

    Reducing the dimensionality of data with neural networks

    Geoffrey E Hinton and Ruslan R Salakhutdinov. Reducing the dimensionality of data with neural networks. science, 313(5786):504–507, 2006

    2006

  10. [18]

    The mit humanoid robot: Design, motion planning, and control for acrobatic behaviors

    Matthew Chignoli, Donghyun Kim, Elijah Stanger-Jones, and Sangbae Kim. The mit humanoid robot: Design, motion planning, and control for acrobatic behaviors. In2020 IEEE-RAS 20th International Conference on Humanoid Robots (Humanoids), pages 1–8, 2021

    2021

  11. [19]

    Whole-body geometric retargeting for humanoid robots

    Kourosh Darvish, Yeshasvi Tirupachuri, Giulio Romualdi, Lorenzo Rapetti, Diego Ferigo, Francisco Javier Andrade Chavez, and Daniele Pucci. Whole-body geometric retargeting for humanoid robots. In2019 IEEE-RAS 19th International Conference on Humanoid Robots (Humanoids), pages ...

    2019

  12. [20]

    Synchronized human-humanoid motion imitation

    Antonin Dallard, Mehdi Benallegue, Fumio Kanehiro, and Abderrahmane Kheddar. Synchronized human-humanoid motion imitation. IEEE Robotics and Automation Letters, 8(7):4155–4162, 2023

    2023

  13. [21]

    Dario Bellicoso, Vassilios Tsounis, Jemin Hwangbo, Karen Bodie, Peter Fankhauser, Michael Bloesch, Remo Diethelm, Samuel Bachmann, Amir Melzer, and Mark Hoepflinger

    Marco Hutter, Christian Gehring, Dominic Jud, Andreas Lauber, C. Dario Bellicoso, Vassilios Tsounis, Jemin Hwangbo, Karen Bodie, Peter Fankhauser, Michael Bloesch, Remo Diethelm, Samuel Bachmann, Amir Melzer, and Mark Hoepflinger. Anymal - a highly mobile and dynamic quadruped...

    2016

  14. [22]

    The 3d linear inverted pendulum mode: A simple modeling for a biped walking pattern generation

    Shuuji Kajita, Fumio Kanehiro, Kenji Kaneko, Kazuhito Yokoi, and Hirohisa Hirukawa. The 3d linear inverted pendulum mode: A simple modeling for a biped walking pattern generation. InProceedings 2001 IEEE/RSJ International Conference on Intelligent Robots and Systems. Expanding...

    2001

  15. [23]

    Whole-body control of humanoid robots.Humanoid robotics: a reference, pages 1161–1183, 2019

    Federico L Moro and Luis Sentis. Whole-body control of humanoid robots.Humanoid robotics: a reference, pages 1161–1183, 2019

    2019

  16. [24]

    Amo: Adaptive motion optimization for hyper-dexterous humanoid whole-body control.Robotics: Science and Systems 2025, 2025

    Jialong Li, Xuxin Cheng, Tianshu Huang, Shiqi Yang, Rizhao Qiu, and Xiaolong Wang. Amo: Adaptive motion optimization for hyper-dexterous humanoid whole-body control.Robotics: Science and Systems 2025, 2025. 11

    2025

  17. [25]

    Jaeger: Dual-level humanoid whole-body controller, 2025

    Ziluo Ding, Haobin Jiang, Yuxuan Wang, Zhenguo Sun, Yu Zhang, Xiaojie Niu, Ming Yang, Weishuai Zeng, Xinrun Xu, and Zongqing Lu. Jaeger: Dual-level humanoid whole-body controller, 2025

    2025

  18. [26]

    Omnih2o: Universal and dexterous human-to-humanoid whole-body teleoperation and learning

    Tairan He, Zhengyi Luo, Xialin He, Wenli Xiao, Chong Zhang, Weinan Zhang, Kris Kitani, Changliu Liu, and Guanya Shi. Omnih2o: Universal and dexterous human-to-humanoid whole-body teleoperation and learning. 2024

    2024

  19. [27]

    Learning human-to-humanoid real-time whole-body teleoperation.arXiv preprint arXiv:2403.04436, 2024

    Tairan He, Zhengyi Luo, Wenli Xiao, Chong Zhang, Kris Kitani, Changliu Liu, and Guanya Shi. Learning human-to-humanoid real-time whole-body teleoperation.arXiv preprint arXiv:2403.04436, 2024

    2024 arXiv

  20. [28]

    Humanplus: Humanoid shadowing and imitation from humans

    Zipeng Fu, Qingqing Zhao, Qi Wu, Gordon Wetzstein, and Chelsea Finn. Humanplus: Humanoid shadowing and imitation from humans. InConference on Robot Learning (CoRL), 2024

    2024

  21. [29]

    Hover: Versatile neural whole-body controller for humanoid robots.arXiv preprint arXiv:2410.21229, 2024

    Tairan He, Wenli Xiao, Toru Lin, Zhengyi Luo, Zhenjia Xu, Zhenyu Jiang, Changliu Liu, Guanya Shi, Xiaolong Wang, Linxi Fan, and Yuke Zhu. Hover: Versatile neural whole-body controller for humanoid robots.arXiv preprint arXiv:2410.21229, 2024

    arXiv

  22. [30]

    Troje, Gerard Pons-Moll, and Michael J

    Naureen Mahmood, Nima Ghorbani, Nikolaus F. Troje, Gerard Pons-Moll, and Michael J. Black. AMASS: Archive of motion capture as surface shapes. InInternational Conference on Computer Vision, pages 5442–5451, October 2019

    2019

  23. [31]

    System identification—a survey.Automatica, 7(2):123–162, 1971

    Karl Johan Åström and Peter Eykhoff. System identification—a survey.Automatica, 7(2):123–162, 1971

    1971

  24. [32]

    System identification techniques.Structural safety, 3(3-4):269–316, 1986

    F Kozin and HG Natke. System identification techniques.Structural safety, 3(3-4):269–316, 1986

    1986

  25. [33]

    Agile continuous jumping in discontinuous terrains.arXiv preprint arXiv:2409.10923, 2024

    Yuxiang Yang, Guanya Shi, Changyi Lin, Xiangyun Meng, Rosario Scalise, Mateo Guaman Castro, Wenhao Yu, Tingnan Zhang, Ding Zhao, Jie Tan, et al. Agile continuous jumping in discontinuous terrains.arXiv preprint arXiv:2409.10923, 2024

    2024 arXiv

  26. [34]

    Policy transfer with strategy optimization

    Wenhao Yu, C Karen Liu, and Greg Turk. Policy transfer with strategy optimization. arXiv preprint arXiv:1810.05751, 2018

    2018 arXiv

  27. [35]

    Learning fast adaptation with meta strategy optimization

    Wenhao Yu, Jie Tan, Yunfei Bai, Erwin Coumans, and Sehoon Ha. Learning fast adaptation with meta strategy optimization. IEEE Robotics and Automation Letters, 5(2):2950–2957, 2020

    2020

  28. [36]

    Matthew Loper, Naureen Mahmood, Javier Romero, Gerard Pons-Moll, and Michael J. Black. SMPL: A skinned multi-person linear model.ACM Trans. Graphics (Proc. SIGGRAPH Asia), 34(6):248:1–248:16, October 2015

    2015

  29. [37]

    Winkler, Kris Kitani, and Weipeng Xu

    Zhengyi Luo, Jinkun Cao, Alexander W. Winkler, Kris Kitani, and Weipeng Xu. Perpetual humanoid control for real-time simulated avatars. InInternational Conference on Computer Vision (ICCV), 2023

    2023

  30. [38]

    Generating diverse and natural 3d human motions from text

    Chuan Guo, Shihao Zou, Xinxin Zuo, Sen Wang, Wei Ji, Xingyu Li, and Li Cheng. Generating diverse and natural 3d human motions from text. InProceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), pages 5152–5161, June 2022

    2022

  31. [39]

    Black, and Gül Varol

    Mathis Petrovich, Michael J. Black, and Gül Varol. TMR: Text-to-motion retrieval using contrastive 3D human motion synthesis. InInternational Conference on Computer Vision (ICCV), 2023

    2023

  32. [40]

    Black, and Gül Varol

    Mathis Petrovich, Michael J. Black, and Gül Varol. TEMOS: Generating diverse human motions from textual descriptions. In European Conference on Computer Vision (ECCV), 2022

    2022

  33. [41]

    Bert: Pre-training of deep bidirectional transformers for language understanding

    Jacob Devlin, Ming-Wei Chang, Kenton Lee, and Kristina Toutanova. Bert: Pre-training of deep bidirectional transformers for language understanding. InProceedings of the 2019 conference of the North American chapter of the association for computational linguistics: human langua...

    2019

  34. [42]

    A reduction of imitation learning and structured prediction to no-regret online learning

    Stéphane Ross, Geoffrey Gordon, and Drew Bagnell. A reduction of imitation learning and structured prediction to no-regret online learning. InProceedings of the fourteenth international conference on artificial intelligence and statistics, pages 627–635. JMLR Workshop and Conf...

    2011

  35. [43]

    Thorndike

    Robert L. Thorndike. Who belongs in the family?Psychometrika, 18(4):267–276, 1953

    1953

  36. [44]

    Robocentric visual–inertial odometry.The International Journal of Robotics Research, 41(7):667–689, 2022

    Zheng Huai and Guoquan Huang. Robocentric visual–inertial odometry.The International Journal of Robotics Research, 41(7):667–689, 2022

    2022

  37. [45]

    Proximal policy optimization algorithms

    John Schulman, Filip Wolski, Prafulla Dhariwal, Alec Radford, and Oleg Klimov. Proximal policy optimization algorithms. arXiv preprint arXiv:1707.06347, 2017. 12 Appendix A Environment Details A.1 RL environment We provide a detailed training and test environment setting in th...

    2017 arXiv

Pith tools

Reviewed August 7, 2026 · model on record in the stance chip above.