Pith. sign in

REVIEW 3 major objections 5 minor 40 cited by

ExBody2: Advanced Expressive Humanoid Whole-Body Control

T0 review · 3 major / 5 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 ExBody2 claims that a humanoid robot can track diverse human whole-body motions with a single control policy when the controller decouples local keypoint tracking from root-velocity tracking and the training data is automatically filtered…

desk verdict Solid whole-body control engineering with real-robot results; the curation threshold is selected and evaluated on the same split, so that specific gain is not yet established, but the rest of the pipeline holds up. read the letter →

arxiv 2412.13196 v2 pith:WDCQVJMP submitted 2024-12-17 cs.RO cs.AIcs.LG

Mazeyu Ji , Xuanbin Peng , Fangchen Liu , Jialong Li , Ge Yang , Xuxin Cheng , Xiaolong Wang This is my paper · ORCID
classification cs.ROcs.AIcs.LG
keywords humanoidwhole-bodycontrolmotionimitationteacher-studentpolicydistillationreinforcementlearningsim-to-realtransferdatasetfilteringexpressivetrackingmotion-velocitydecoupled
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

ExBody2 argues that a humanoid robot can imitate expressive, dynamic human whole-body motions with a single control policy if the training data is automatically pruned to keep only motions the robot can physically track, and if the controller separates two goals: matching body landmarks in the robot's own coordinate frame and following the commanded root velocity. The paper builds a teacher-student pipeline in which a privileged reinforcement-learned teacher produces the data used to filter the motion corpus, and a student policy is distilled for deployment on a real humanoid. On the evaluation corpus, the authors report that this one generalist policy tracks keypoints and joint angles more accurately than the compared whole-body controllers, and that brief fine-tuning on a motion group yields specialist policies with still higher precision. The reason to care is that whole-body expressiveness, not just stable walking, is the missing piece between humanoid hardware and natural behavior in human spaces.

What carries the argument

The load-bearing mechanism is the motion-velocity decoupled control strategy combined with automated feasibility filtering. Instead of chasing global keypoint coordinates, the controller maps keypoints into the robot's current body frame and tracks them locally, while a separate channel tracks the commanded root linear velocity, direction, and orientation; this prevents small positional errors from accumulating and allows a long dancing sequence to stay stable. The filtering step uses the initial policy $\pi_0$ as an oracle: for each motion sequence $s$ it computes $e(s) = \alpha E_{\text{key}}(s) + \beta E_{\text{dof}}(s)$ with $\alpha = 0.1$ and $\beta = 0.9$, and the paper selects $\tau^*$ so that $D_{\tau^*} = \{s : e(s) \le \tau^*\}$ is the training set for the final generalist, embodying a Feasibility-Diversity Principle: keep upper-body diversity while removing lower-body motions the robot cannot physically track. A teacher-student distillation step then transfers the privileged teacher's behavior to a student that observes only proprioception and reference targets, with history length and online imitation-style rollouts shown to be necessary for the final accuracy.

What would settle it

Hold out a random half of the motion corpus before training. Select the error cutoff using only the retained half, train the generalist on the filtered retained half, and measure its error on the held-out half against an unfiltered policy. If the filtered policy is not better on the held-out half, the claimed curation gain is overfitting to the threshold-selection set. A second check: rerun the filtering with a much stronger initial policy and see whether the same motion clips are discarded; if the rankings flip substantially, the filter is measuring the weakness of the first policy, not the feasibility of the motions.

Watch

Extended reading notes

Core claim

The paper's central claim is that the right formulation of whole-body tracking is two decoupled objectives - local body-landmark tracking plus global motion-velocity tracking - and that the main obstacle to expressive control is not policy capacity but the feasibility of the training data. Its automated curation scores every motion clip by the initial teacher's lower-body tracking error $e(s) = \alpha E_{\text{key}}(s) + \beta E_{\text{dof}}(s)$ with $\alpha = 0.1$, $\beta = 0.9$, keeps the clips below a threshold $\tau^*$ chosen to maximize performance on the full evaluation corpus, and trains the generalist policy on this subset. The reported result is that this generalist, trained and deployed as a single policy, achieves lower keypoint and joint tracking errors than the compared baselines in both simulation and real-world experiments, and that resuming from it to fine-tune on a specialized motion group gives specialists that outperform both the generalist and models trained from scratch on matched compute. The authors frame the trade-off explicitly: filtering slightly increases velocity-tracking error while substantially improving stability and precision, and fine-tuning gains accuracy on target motions at the cost of broad coverage.

Load-bearing premise

The whole gain of the generalist policy hinges on the assumption that the first, unfiltered policy's lower-body tracking errors are the right measure of which motions the robot can never physically perform, and that the cutoff chosen on the evaluation corpus works on other motion data without retuning.

Editorial extensions

If this is right

  • A single ExBody2 generalist policy can replace separate controllers for walking, crouching, dancing, and upper-body gestures on the same humanoid platform.
  • New motion capture data can be pruned automatically by the same error-threshold procedure, removing the need for manual or language-based filtering.
  • Fine-tuning the generalist on a small, motion-specific dataset yields higher tracking accuracy than training from scratch under matched compute, with the advantage growing as motions become more difficult.
  • The accepted cost of filtering is a slightly larger root-velocity error, traded for substantially lower keypoint and joint errors and better real-world stability.

Reading between the lines

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

  • A testable consequence the paper leaves open: apply the threshold-selection procedure to a held-out motion library and verify that the $\tau^*$ tuned on the main corpus is still near-optimal; the paper asserts this transfer but does not evaluate it.
  • Because the filter is defined by the initial policy's errors, better initial policies would likely change which clips are labeled infeasible, so the curation and the policy are coupled rather than the filter being a property of the data alone.
  • The acknowledged limitation that specialist policies cannot be recombined in one session points to a natural next step: a router or gating network that blends specialists in real time based on the incoming motion class.
  • If decoupled velocity control rather than filtering is the main source of gain, an ablation that keeps the filter and swaps in global-keypoint tracking should degrade performance more than removing any single reward term; the paper does not run this isolation.
Share X Bluesky LinkedIn Reddit HN

Signed reviews

No signed human review yet.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

3 major / 5 minor

Summary. The paper presents ExBody2, a teacher-student reinforcement-learning framework for whole-body tracking on a Unitree G1 humanoid. A teacher policy trained with privileged information is distilled into a student policy via DAgger, and the method adds three claimed innovations: automated dataset curation that filters infeasible motion clips using the tracking error of an initial policy, a generalist-to-specialist fine-tuning paradigm, and a decoupled motion-velocity control strategy that tracks local keypoints while using root velocity for global guidance. Experiments on the CMU dataset and on the real robot report lower keypoint and joint errors than ExBody, ExBody*, and OmniH2O* baselines, and show that specialist fine-tuning improves accuracy over the generalist and from-scratch training. The paper also reports ablations of history length and DAgger distillation, and an appendix explores dataset size and feasibility-diversity trade-offs.

Significance. If the results hold, ExBody2 would be a strong step for expressive whole-body control: a single generalist policy outperforming prior whole-body trackers in simulation and on real hardware, with a principled, automated answer to the question of which mocap clips are worth training on. The paper's strengths are its real-robot deployment, the consistent use of multiple error metrics, and the informative ablations of history length and DAgger distillation, as well as the direct comparison of generalist, specialist, and from-scratch training in Table IV. The central unresolved issue is that the curation threshold is chosen and evaluated on the same benchmark, the decoupled-control innovation is never isolated, and the main tables lack variance reporting; these gaps weaken the attribution of the reported gains to the paper's specific contributions. If the missing holdout or cross-dataset validation and proper statistical reporting are added, the paper would be a solid contribution to the field.

major comments (3)
  1. [§III.A.1 and §IV.C] The automated-curation threshold is selected and evaluated on the same benchmark. The objective 'τ* = arg max_τ E_{s∈D}[Performance(π_τ,s)]' in §III.A.1 is instantiated with D = D_CMU: Fig. 4 chooses τ*=0.15 from policies evaluated on D_CMU, and Table II then reports ExBody2 versus ExBody2-w/o-Filter on that same dataset. The curation benefit (e.g., Empjpe 0.1355→0.1079, Elower_mpjpe 0.1579→0.1253) can therefore include selection overfit, and the statement in §IV.C that the threshold 'generalizes to other datasets' is not backed by any experiment that fixes τ*=0.15 and evaluates on a held-out split or an independent dataset. In addition, the filter score e(s)=αE_key(s)+βE_dof(s) is produced by the initial policy π0; because π0 is suboptimal, motions that a later, better policy could track are removed if π0 finds them hard, conflating 'difficult for π0' with 'infeasible.' Please add a holdout/cross-dataset validation of the threshold and an explicit test of the infeasibility proxy before claiming the curation gain.
  2. [§III.B.3] The decoupled motion-velocity control strategy is not ablated in isolation. The paper's comparison of ExBody2 with OmniH2O* in Tables II and III changes observation space, privileged information, reward design, and teacher-student training simultaneously, so the better tracking cannot be attributed specifically to the decoupled local-keypoint/velocity architecture described in §III.B.3. Please add a variant that keeps all other components fixed and replaces the local keypoint tracking plus velocity term with global keypoint tracking (or removes the decoupled velocity term), and report its errors on D_CMU and on the real-world subset.
  3. [Tables II–IV] Headline results are reported without variance or significance. Tables II, III, IV and Fig. 4 give single numbers, whereas Appendix Table X says 'Statistically significant results are highlighted in bold across 5 random seeds'; no seed-level standard deviations or tests are reported for the main claims. Without error bars, small differences—for example ExBody2-w/o-Filter versus ExBody2 on upper-body metrics in Table II—cannot be distinguished from run-to-run noise. Please report mean±std over at least five seeds (or trials, for the real-robot results) for all headline comparisons, and state the number of seeds used.
minor comments (5)
  1. [§IV.D.2] The text states that the specialist 'significantly outperforms' both the generalist and scratch policies on DACCAD, but in Table IV(d) the Generalist policy has a lower velocity error (Evel=0.3361) than the Specialist (0.4021); please qualify this claim and discuss the velocity-versus-pose trade-off.
  2. [§III.B.1] In the reward-design paragraph, 'would discussed' should read 'will be discussed.'
  3. [References] References [3] and [4] are the same arXiv paper (Expressive Whole-Body Control for Humanoid Robots) and should be merged.
  4. [Title/Abstract] The name of the method is spelled inconsistently as 'Exbody2' in the abstract and text and 'ExBody2' in the title; please standardize.
  5. [§III.A.1] The notation 'Performance(πτ, s)' in the threshold objective is never formally defined; please specify which metric over a sequence is used (e.g., mean MPJPE) so that the optimization is reproducible.

Circularity Check

1 steps flagged · score 6.0 of 10

The filtering threshold is selected and evaluated on the same D_CMU corpus, so the reported curation gain and the "generalizable" tau* are fit results rather than independent predictions.

  1. fitted input called prediction [Section III.A.1 (Eq. tau* objective); Section IV.C (Evaluation step, Fig. 4)]
    "τ ∗ = arg max τ Es∈D[Performance(πτ , s)], where πτ is trained on Dτ . ... All the resulting policies were evaluated on the full DCMU dataset, measuring tracking performance across multiple metrics. The results are visualized in Figure 4. ... The optimal threshold τ ∗, identified through this process, exhibits generalizability and can be effectively applied to other motion datasets, ensuring robust training and improved performance."

    tau* is defined by maximizing Performance on D, and the paper's evaluation of filtering (Fig. 4 and Table II) measures the same Performance on the same D_CMU. Thus the statement that tau=0.15 is optimal and that filtering improves tracking is a restatement of the argmax used to select the threshold, not an independent empirical finding. The additional sentence claiming that tau* 'exhibits generalizability' to other datasets promotes this fitted value to a universal claim without any holdout or cross-dataset experiment, so the curation benefit is presented as a validated prediction when it is actually a selection result.

full rationale

ExBody2's automated-data-curation contribution contains one selection-on-the-evaluation-set loop. In Section III.A.1, tau* is chosen as the threshold maximizing Es in D[Performance(pi_tau, s)]; in Section IV.C, the same D_CMU is used both to pick among {0.075, 0.1, 0.125, 0.15, 0.175} and to report the resulting errors (Fig. 4, Table II). Consequently, the 'discovery' that tau=0.15 is best and that filtering helps is not an independently tested prediction: it is the definition of the argmax. The claim that tau* generalizes to other datasets is asserted without any holdout split or cross-dataset evaluation, so that transfer claim is unsupported. This circularity is partial, not total: the unfiltered Exbody2-w/o-Filter still beats ExBody, ExBody†, and OmniH2O* on D_CMU, and the decoupled velocity-control and specialist-finetuning comparisons (Deasy, Dmoderate, Dhard, DACCAD) are evaluated on separate task groups rather than on the threshold-selection objective. No load-bearing self-citation chain or imported uniqueness theorem was found. The score of 6 reflects one fitted prediction embedded in a central claimed innovation, while the broader outperformance and specialist claims retain independent content.

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

The central claim rests on a small number of hand-picked weights and thresholds (α, β, τ*) and on standard domain assumptions about simulation-to-real transfer and reward design. No new physical entities are introduced.

free parameters (4)
  • α, β in e(s) = α=0.1, β=0.9
    Hand-chosen weights in the feasibility score; changing them changes which motions are filtered (Fig. 4 caption).
  • Filtering threshold τ* = 0.15
    Selected by greedy search to maximize performance on D_CMU, then evaluated on the same D_CMU; a fitted value, not an independent prediction.
  • Reward weights = Various (see Tables I and IX)
    Hand-tuned coefficients for tracking and regularization rewards; central to policy behavior but common in RL.
  • History length = 10
    Chosen by ablation (Table XI), not derived.
assumptions (4)
  • domain assumption The CMU MoCap dataset, after retargeting to Unitree G1, is a sufficient source of diverse human motions for training expressive whole-body control.
    Used throughout; no analysis of dataset coverage vs. target tasks.
  • ad hoc to paper Tracking error under the initial policy is a valid proxy for physical infeasibility of a motion.
    Defines the filtering rule in Section III.A.1; conflates hard for current policy with infeasible.
  • domain assumption PPO and DAgger distillation in IsaacGym transfer to the real robot without additional real-world data.
    Standard sim-to-real premise in this literature, invoked for all deployment claims.
  • domain assumption The hand-designed reward function correctly encodes the desired trade-off between expressiveness and stability.
    Reward in Table I and IX is the objective being optimized; no comparison to alternative reward designs.

how reviews work

0 comments
Cite this review

Pith. "Pith review of ExBody2: Advanced Expressive Humanoid Whole-Body Control." pith.science (2026). https://pith.science/paper/WDCQVJMP

@misc{pith2026241213196,
  author       = {Pith},
  title        = {Pith review of: ExBody2: Advanced Expressive Humanoid Whole-Body Control},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/WDCQVJMP}},
  note         = {Machine review of arXiv:2412.13196}
}
read the original abstract

This paper tackles the challenge of enabling real-world humanoid robots to perform expressive and dynamic whole-body motions while maintaining overall stability and robustness. We propose Advanced Expressive Whole-Body Control (Exbody2), a method for producing whole-body tracking controllers that are trained on both human motion capture and simulated data and then transferred to the real world. We introduce a technique for decoupling the velocity tracking of the entire body from tracking body landmarks. We use a teacher policy to produce intermediate data that better conforms to the robot's kinematics and to automatically filter away infeasible whole-body motions. This two-step approach enabled us to produce a student policy that can be deployed on the robot that can walk, crouch, and dance. We also provide insight into the trade-off between versatility and the tracking performance on specific motions. We observed significant improvement of tracking performance after fine-tuning on a small amount of data, at the expense of the others.

Figures

Figures reproduced from arXiv: 2412.13196 by the authors.

Figure 1
Figure 1. Humanoid robot executing various expressive whole-body motions in the real world. The robot can (a) walk with a large stride from static standing, (b) dance along a long horizon (43 seconds) choreography, (c) dynamic sidestep with fluid weight shifts, (d) punch with different height configurations, (e) express various upper-body movements while maintaining balance, (f) powerful rightward body hook with dynamic shift… view at source ↗
Figure 2
Figure 2. Exbody2’s framework. (a) Motion retargeting adapts raw human motion datasets to fit the humanoid robot’s morphology, generating a diverse set of training samples. (b) Automated dataset filtering ranks motions based on tracking errors and selects an optimal subset to train a generalist policy, balancing feasibility and diversity. (c) Specialist policy finetuning refines the generalist model for specific motion catego… view at source ↗
Figure 3
Figure 3. Teacher-student framework for humanoid motion learning, where the teacher uses privileged information, and the student learns from past observations to generate control actions. shelf reinforcement learning (RL) algorithm, PPO [51], with privileged information that can be obtained only in simulators. For the second stage, we replace the privileged information with observations which are aligned with the real world, … view at source ↗
Figures from the paper (5 more)
Figure 4
Figure 4. Figure 4: Impact of dataset filtering thresholds on policy tracking errors. The figure shows the tracking error trends across different dataset fil￾tering thresholds. Policies trained on datasets with filtering thresholds that balance diversity and stability (e.g., πτ=0.150) ach…
Figure 5
Figure 5. Figure 5: A sequence of a robot performing the Cha-Cha dance. From top to bottom: the reference motion represented by an avatar, our algorithm’s performance in the simulation, and its performance on a real robot. The bottom three rows show the per-frame errors: whole￾body joint …
Figure 6
Figure 6. Figure 6: Illustration of ExBody2’s multi-source application, demonstrating how VR, RGB, motion capture, and generative models can be combined to produce diverse humanoid behaviors. (a) Motion Datasets: specialized policies (e.g., kung fu, dancing) finetuned on specialist motion…
Figure 7
Figure 7. Figure 7: Sim-to-real experiment results showcasing diverse motions across SMPL, simulation, and real-world environments. Examples include: (a) Clasping Fists, (b) Clapping Twist, (c) Greeting Gesture, (d) Punching, (e) Crouching, and (f) Defensive Pose. large dataset (DCMU )—wh…
Figure 8
Figure 8. Figure 8: Empirical CDF of the base policy’s error metric e(s) on the entire DCMU dataset. The horizontal axis indicates the percentile of motion sequences from 0% (lowest error) to 100% (highest error), while the vertical axis shows e(s). We overlay dashed horizontal lines at k…

Discussion (0). Continue with ORCID to comment.

Forward citations

Cited by 40 Pith papers

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

  1. Transformer Transformer: A Unified Model for Motion-Conditioned Robot Co-design

    cs.RO 2026-07 conditional novelty 7.0 of 10

    A single diffusion transformer trains on tokenized robot bodies and motions to generate and optimize robot designs for unseen rewards and trajectories, outpacing evolutionary search in speed and often in reward.

  2. Deep Sensorimotor Control by Imitating Predictive Models of Human Motion

    cs.RO 2025-08 conditional novelty 7.0 of 10

    A predictive model of human hand motion, trained on human interaction data, can reward a robot policy for tracking predicted future keypoints and enable learning of dexterous manipulation from sparse rewards.

  3. Visual Imitation Enables Contextual Humanoid Control

    cs.RO 2025-05 conditional novelty 7.0 of 10

    A single policy trained from 123 monocular videos, fine-tuned in simulation, and distilled to heightmap plus root-direction inputs lets a Unitree G1 climb stairs and sit and stand on real furniture.

  4. LangWBC: Language-directed Humanoid Whole-Body Control via End-to-end Learning

    cs.RO 2025-04 conditional novelty 7.0 of 10

    LangWBC maps text commands and proprioceptive history directly to humanoid actions via a conditional variational autoencoder distilled from an RL teacher, enabling diverse, smoothly transitioning whole-body motions on...

  5. $\omega$-0: A Latent Predictive World Action Model for Concurrent Humanoid Loco-Manipulation

    cs.RO 2026-08 conditional novelty 6.0 of 10

    A single whole-body model with latent future prediction outperforms prior robot policies on 11 real-world humanoid household loco-manipulation tasks.

  6. Teleopit: A Full-Embodiment Humanoid Teleoperation System

    cs.RO 2026-08 conditional novelty 6.0 of 10

    Teleopit combines VR body, hand, and head tracking with a learned whole-body tracker and a cross-hand retargeter, and teleop-collected demos train ACT and GR00T policies to around 90 to 95 percent success on a humanoi...

  7. LooperMuscle: Fast and Stable Learning of Humanoid Whole-Body Tracking via Structured Mixture-of-Experts

    cs.RO 2026-08 conditional novelty 6.0 of 10

    LooperMuscle trains a humanoid whole-body tracking policy in about 45 minutes, cutting body error by 34% versus FastSAC and reaching 72% of PPO's reward, versus PPO's 6 hours.

  8. What Matters in Humanoid General Motion Tracking? An Empirical Study

    cs.RO 2026-07 conditional novelty 6.0 of 10

    A controlled ablation of humanoid motion-tracking pipelines shows that explicit reference joint velocities and a short observation history improve tracking, while residual actions and teacher-student training yield on...

  9. ContactMimic: Humanoid Object Interaction via Contact Control

    cs.RO 2026-07 conditional novelty 6.0 of 10

    A humanoid tracking policy is trained with contact-following rewards and trajectory augmentation to decouple physical contact from keypoint geometry, enabling runtime contact control.

  10. 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.

  11. VOFA: Visual Object Goal Pushing with Force-Adaptive Control for Humanoids

    cs.RO 2026-05 unverdicted novelty 6.0 of 10

    VOFA combines a high-level visuomotor policy with a low-level force-adaptive controller to let humanoids push objects up to 17 kg to arbitrary goals using only noisy onboard vision, achieving over 80% real-world success.

  12. Thor: Towards Human-Level Whole-Body Reactions for Intense Contact-Rich Environments

    cs.RO 2025-10 conditional novelty 6.0 of 10

    A decoupled whole-body RL policy with a force-based lean reward enables a Unitree G1 humanoid to pull with up to 167.7 N, beating prior controllers by 69–75%.

  13. PHUMA: Physically Reliable Humanoid Locomotion Dataset

    cs.RO 2025-10 conditional novelty 6.0 of 10

    PHUMA is a curated 73-hour humanoid locomotion corpus whose physical-reliability metrics are partly defined by the same losses used to optimize it, and whose imitation success claims are confounded by in-distribution ...

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

    cs.RO 2025-10 conditional novelty 6.0 of 10

    A 10,300-demonstration, 260-task multimodal humanoid manipulation dataset with baseline policy evaluations and a cloud evaluation platform.

  15. TOP: Time Optimization Policy for Stable and Accurate Standing Manipulation with Humanoid Robots

    cs.RO 2025-08 conditional novelty 6.0 of 10

    A reinforcement-learned time optimization policy that adaptively slows upper-body motion clips improves stability and precision of humanoid standing manipulation at a modest time cost.

  16. In-between Motion Generation Based Multi-Style Quadruped Robot Locomotion

    cs.RO 2025-07 conditional novelty 6.0 of 10

    A CVAE-based in-between motion generator creates multi-style quadruped gaits from sparse motion data, and the trained controller runs gallop, tripod, trotting, and pacing on a real robot.

  17. Being-H0: Vision-Language-Action Pretraining from Large-Scale Human Videos

    cs.CV 2025-07 conditional novelty 6.0 of 10

    A dexterous VLA pretrained on a 2.5M-instance human hand motion dataset transfers skills to a real robot hand, outperforming baselines in manipulation tasks.

  18. Learning Motion Skills with Adaptive Assistive Curriculum Force in Humanoid Robots

    cs.RO 2025-06 conditional novelty 6.0 of 10

    A2CF uses an adaptive assistive-force agent to guide humanoid robots through training, yielding faster convergence and robust policies that work without the external force.

  19. GMT: General Motion Tracking for Humanoid Whole-Body Control

    cs.RO 2025-06 conditional novelty 6.0 of 10

    GMT trains a single unified humanoid policy using adaptive sampling and mixture-of-experts, achieving lower tracking errors than a re-implemented ExBody2 across diverse whole-body motions.

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

    cs.RO 2025-06 conditional novelty 6.0 of 10

    BumbleBee, an expert-to-generalist pipeline using autoencoder-based motion clustering and per-cluster delta action models, reports state-of-the-art whole-body control on a Unitree G1 humanoid, with success rates of 89...

  21. SkillBlender: Towards Versatile Humanoid Whole-Body Loco-Manipulation via Skill Blending

    cs.RO 2025-06 conditional novelty 6.0 of 10

    SkillBlender pretrains reusable goal-conditioned skills and blends them with softmax per-joint weights to solve simulated humanoid loco-manipulation tasks with one or two reward terms.

  22. MoRE: Mixture of Residual Experts for Humanoid Lifelike Gaits Learning on Complex Terrains

    cs.RO 2025-06 conditional novelty 6.0 of 10

    A two-stage reinforcement learning pipeline with a mixture of latent residual experts gives a Unitree G1 humanoid multiple commanded human-like gaits over complex terrains.

  23. HuB: Learning Extreme Humanoid Balance

    cs.RO 2025-05 conditional novelty 6.0 of 10

    HuB combines reference motion refinement, balance shaping rewards, and robustness training to enable a G1 humanoid to hold extreme single-leg poses that prior tracking methods fail to maintain.

  24. JAEGER: Dual-Level Humanoid Whole-Body Controller

    cs.RO 2025-05 conditional novelty 6.0 of 10

    JAEGER decouples upper- and lower-body control of a humanoid robot into two jointly trained policies and reports lower pose- and velocity-tracking errors than prior whole-body controllers in simulation.

  25. Towards Embodiment Scaling Laws in Robot Locomotion

    cs.RO 2025-05 conditional novelty 6.0 of 10

    A policy trained on about one thousand simulated robot bodies generalizes progressively better to unseen bodies as the number of training bodies grows, and it transfers zero-shot to two real robots.

  26. TWIST: Teleoperated Whole-Body Imitation System

    cs.RO 2025-05 conditional novelty 6.0 of 10

    Human MoCap drives a Unitree G1 humanoid in real time through a single teacher-student RL+BC controller that transfers zero-shot from simulation.

  27. BeamDojo: Learning Agile Humanoid Locomotion on Sparse Footholds

    cs.RO 2025-02 conditional novelty 6.0 of 10

    A two-stage RL framework with a polygonal-foot foothold reward and double critic enables a Unitree G1 humanoid to traverse sparse footholds in simulation and the real world.

  28. A Unified and General Humanoid Whole-Body Controller for Versatile Locomotion

    cs.RO 2025-02 conditional novelty 6.0 of 10

    A single RL policy controls walking, jumping, and standing gaits of a humanoid with tunable foot and posture parameters, plus a separate policy for hopping, and supports real-time upper-body intervention for loco-mani...

  29. Perception-and-action system for humanoid robot task execution in construction

    cs.RO 2026-08 conditional novelty 5.0 of 10

    A humanoid robot learned eight construction motions from human video demonstrations via pose retargeting and reinforcement learning, with an average 82.45 mm joint tracking error.

  30. ZeroWBC: Learning Natural Whole-Body Humanoid Interaction from Human Egocentric Data

    cs.RO 2026-03 conditional novelty 5.0 of 10

    An open-loop generation-then-tracking system maps one egocentric image plus language into Unitree G1 whole-body interactions using only human egocentric motion data.

  31. 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.

  32. Towards Immersive Human-X Interaction: A Real-Time Framework for Physically Plausible Motion Synthesis

    cs.CV 2025-08 unverdicted novelty 5.0 of 10

    Human-X jointly predicts actions and reactions in real time to produce physically plausible human-machine interaction motion.

  33. Hierarchical Vision-Language Planning for Multi-Step Humanoid Manipulation

    cs.RO 2025-06 conditional novelty 5.0 of 10

    A three-layer hierarchical system using a VLM planner and VLM skill monitor with imitation-learned skills and an RL tracking policy achieved 73% success on a real humanoid pick-and-place task.

  34. Multi-Embodiment Robotic Retargeting via Guided Diffusion Model

    cs.RO 2025-05 reject novelty 5.0 of 10

    A graph-conditioned diffusion model retargets motions across heterogeneous robot embodiments without needing target-robot motion data, yet lacks baseline comparisons and error bars in its validation.

  35. SMAP: Self-supervised Motion Adaptation for Physically Plausible Humanoid Whole-body Control

    cs.RO 2025-05 conditional novelty 5.0 of 10

    SMAP uses a vector-quantized periodic autoencoder to adapt human motion into physically plausible humanoid motion, then distills an RL teacher policy into a student policy for whole-body control.

  36. AMO: Adaptive Motion Optimization for Hyper-Dexterous Humanoid Whole-Body Control

    cs.RO 2025-05 conditional novelty 5.0 of 10

    A hybrid trajectory-optimization and RL framework lets a humanoid robot flex its torso and legs to reach and manipulate objects beyond the range of prior controllers.

  37. ASAP: Aligning Simulation and Real-World Physics for Learning Agile Humanoid Whole-Body Skills

    cs.RO 2025-02 conditional novelty 5.0 of 10

    ASAP trains a residual action model on real-world rollouts and fine-tunes simulation policies through it, reducing humanoid whole-body motion tracking error in sim-to-real transfer.

  38. Robust RL Control for Bipedal Locomotion with Closed Kinematic Chains

    cs.RO 2025-07 conditional novelty 4.0 of 10

    A reinforcement-learning gait controller that explicitly models closed kinematic chains outperforms one trained on a simplified serial model, both in simulation and on the physical TopA robot.

  39. A Survey: Learning Embodied Intelligence from Physical Simulators and World Models

    cs.RO 2025-07 conditional novelty 4.0 of 10

    Embodied intelligence learning is reviewed through the complementary lenses of physical simulators and world models, with a proposed IR-L0 to IR-L4 robot capability taxonomy.

  40. Feature-Based vs. GAN-Based Learning from Demonstrations: When and Why

    cs.LG 2025-07 conditional novelty 3.0 of 10

    Feature-based and GAN-based imitation learning should be selected by task priorities (fidelity, diversity, interpretability, adaptability), not by paradigm loyalty.

Reference graph

Works this paper leans on

70 extracted references · 32 canonical work pages · cited by 40 Pith papers

  1. [1]

    Carnegie-Mellon mocap database

    Carnegie Mellon University. Carnegie-Mellon mocap database. http://mocap.cs.cmu.edu/, Mar 2007. [Online]

    work page 2007

  2. [2]

    Arcap: Collecting high-quality human demonstrations for robot learning with augmented reality feedback

    Sirui Chen, Chen Wang, Kaden Nguyen, Li Fei-Fei, and C Karen Liu. Arcap: Collecting high-quality human demonstrations for robot learning with augmented reality feedback. arXiv preprint arXiv:2410.08464 , 2024

    arXiv 2024

  3. [4]

    Expressive whole- body control for humanoid robots

    Xuxin Cheng, Yandong Ji, Junming Chen, Ruihan Yang, Ge Yang, and Xiaolong Wang. Expressive whole- body control for humanoid robots. arXiv preprint arXiv:2402.16796, 2024

    arXiv 2024

  4. [5]

    Universal manipulation interface: In- the-wild robot teaching without in-the-wild robots

    Cheng Chi, Zhenjia Xu, Chuer Pan, Eric Cousineau, Benjamin Burchfiel, Siyuan Feng, Russ Tedrake, and Shuran Song. Universal manipulation interface: In- the-wild robot teaching without in-the-wild robots. In Proceedings of Robotics: Science and Systems (RSS) , 2024

    2024

  5. [6]

    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. In 2020 IEEE-RAS 20th International Conference on Humanoid Robots (Humanoids), pages 1–8. IEEE, 2021

    work page 2020

  6. [7]

    Synchronized human- humanoid motion imitation

    Antonin Dallard, Mehdi Benallegue, Fumio Kanehiro, and Abderrahmane Kheddar. Synchronized human- humanoid motion imitation. IEEE Robotics and Automa- tion Letters , 8(7):4155–4162, 2023. doi: 10.1109/LRA. 2023.3280807

    arXiv 2023

  7. [8]

    Whole body humanoid control from human motion descriptors

    Behzad Dariush, Michael Gienger, Bing Jian, Christian Goerick, and Kikuo Fujimura. Whole body humanoid control from human motion descriptors. In 2008 IEEE International Conference on Robotics and Automation , pages 2677–2684. IEEE, 2008

    work page 2008

  8. [9]

    Whole- body geometric retargeting for humanoid robots

    Kourosh Darvish, Yeshasvi Tirupachuri, Giulio Ro- mualdi, Lorenzo Rapetti, Diego Ferigo, Francisco Javier Andrade Chavez, and Daniele Pucci. Whole- body geometric retargeting for humanoid robots. In 2019 IEEE-RAS 19th International Conference on Hu- manoid Robots (Humanoids), pages 679–686, 2019. doi: 10.1109/Humanoids43949.2019.9035059

    arXiv 2019

Show all 70 references
  1. [10]

    Learning vision-based bipedal locomotion for challenging terrain

    Helei Duan, Bikram Pandit, Mohitvishnu S Gadde, Bart Jaap van Marum, Jeremy Dao, Chanho Kim, and Alan Fern. Learning vision-based bipedal locomotion for challenging terrain. arXiv preprint arXiv:2309.14594, 2023

    2023 arXiv

  2. [11]

    Learning multi-modal whole- body control for real-world humanoid robots, 2024

    Pranay Dugar, Aayam Shrestha, Fangzhou Yu, Bart van Marum, and Alan Fern. Learning multi-modal whole- body control for real-world humanoid robots, 2024. URL https://arxiv.org/abs/2408.07295

    2024 arXiv

  3. [12]

    Adversarial motion priors make good substitutes for complex reward functions

    Alejandro Escontrela, Xue Bin Peng, Wenhao Yu, Tingnan Zhang, Atil Iscen, Ken Goldberg, and Pieter Abbeel. Adversarial motion priors make good substitutes for complex reward functions. 2022 ieee. In International Conference on Intelligent Robots and Systems (IROS) , volume 2, 2022

    2022

  4. [13]

    Humanplus: Humanoid shadowing and imitation from humans, 2024

    Zipeng Fu, Qingqing Zhao, Qi Wu, Gordon Wetzstein, and Chelsea Finn. Humanplus: Humanoid shadowing and imitation from humans, 2024. URL https://arxiv.org/abs/ 2406.10454

    2024 arXiv

  5. [14]

    Mo- bile aloha: Learning bimanual mobile manipulation with low-cost whole-body teleoperation

    Zipeng Fu, Tony Z Zhao, and Chelsea Finn. Mo- bile aloha: Learning bimanual mobile manipulation with low-cost whole-body teleoperation. arXiv preprint arXiv:2401.02117, 2024

    2024 arXiv

  6. [15]

    Umi on legs: Making manipulation policies mobile with manipulation-centric whole-body controllers

    Huy Ha, Yihuai Gao, Zipeng Fu, Jie Tan, and Shuran Song. Umi on legs: Making manipulation policies mobile with manipulation-centric whole-body controllers. arXiv preprint arXiv:2407.10353, 2024

    2024 arXiv

  7. [17]

    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,

  8. [18]

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

    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

  9. [19]

    Hover: Versatile neural whole-body controller for humanoid robots

    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

  10. [20]

    The development of honda humanoid robot

    Kazuo Hirai, Masato Hirose, Yuji Haikawa, and Toru Takenaka. The development of honda humanoid robot. In Proceedings. 1998 IEEE international conference on robotics and automation (Cat. No. 98CH36146) , vol- ume 2, pages 1321–1326. IEEE, 1998

    1998

  11. [21]

    Huang, Edwin Olson, and David C

    Albert S. Huang, Edwin Olson, and David C. Moore. Lcm: Lightweight communications and marshalling. 2010 IEEE/RSJ International Conference on Intelligent Robots and Systems , pages 4057–4062, 2010. URL https://api.semanticscholar.org/CorpusID:10900899

    2010

  12. [22]

    Anymal-a highly mobile and dynamic quadrupedal robot

    Marco Hutter, Christian Gehring, Dominic Jud, Andreas Lauber, C Dario Bellicoso, Vassilios Tsounis, Jemin Hwangbo, Karen Bodie, Peter Fankhauser, Michael Bloesch, et al. Anymal-a highly mobile and dynamic quadrupedal robot. In IROS, 2016

    2016

  13. [23]

    Efficient multitask learning with an embodied predictive model for door opening and entry with whole- body control

    Hiroshi Ito, Kenjiro Yamamoto, Hiroki Mori, and Tetsuya Ogata. Efficient multitask learning with an embodied predictive model for door opening and entry with whole- body control. Science Robotics, 7(65):eaax8177, 2022

    2022

  14. [24]

    Learning whole- body manipulation for quadrupedal robot

    Seunghun Jeon, Moonkyu Jung, Suyoung Choi, Beomjoon Kim, and Jemin Hwangbo. Learning whole- body manipulation for quadrupedal robot. arXiv preprint arXiv:2308.16820, 2023

    2023 arXiv

  15. [25]

    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. In Proceedings 2001 IEEE/RSJ Inter- national Conference on Intelligent Robots and Systems. Expand...

    2001

  16. [26]

    Egomimic: Scaling imitation learning via egocentric video, 2024

    Simar Kareer, Dhruv Patel, Ryan Punamiya, Pranay Mathur, Shuo Cheng, Chen Wang, Judy Hoffman, and Danfei Xu. Egomimic: Scaling imitation learning via egocentric video, 2024. URL https://arxiv.org/abs/2410. 24221

    2024

  17. [27]

    Development of wabot 1

    Ichiro Kato. Development of wabot 1. Biomechanism, 2:173–214, 1973

    1973

  18. [28]

    Rma: Rapid motor adaptation for legged robots

    Ashish Kumar, Zipeng Fu, Deepak Pathak, and Jitendra Malik. Rma: Rapid motor adaptation for legged robots. arXiv preprint arXiv:2107.04034 , 2021

    2021 arXiv

  19. [29]

    Learning quadrupedal locomotion over challenging terrain

    Joonho Lee, Jemin Hwangbo, Lorenz Wellhausen, Vladlen Koltun, and Marco Hutter. Learning quadrupedal locomotion over challenging terrain. Science robotics, 5 (47):eabc5986, 2020

    2020

  20. [30]

    Hybrik: A hybrid analytical- neural inverse kinematics solution for 3d human pose and shape estimation

    Jiefeng Li, Chao Xu, Zhicun Chen, Siyuan Bian, Lixin Yang, and Cewu Lu. Hybrik: A hybrid analytical- neural inverse kinematics solution for 3d human pose and shape estimation. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition, pages 3383–3393, 2021

    2021

  21. [31]

    Okami: Teaching humanoid robots manipulation skills through single video imitation

    Jinhan Li, Yifeng Zhu, Yuqi Xie, Zhenyu Jiang, Mingyo Seo, Georgios Pavlakos, and Yuke Zhu. Okami: Teaching humanoid robots manipulation skills through single video imitation. In 8th Annual Conference on Robot Learning , 2024

    2024

  22. [32]

    Reinforcement learning for robust parameterized loco- motion control of bipedal robots

    Zhongyu Li, Xuxin Cheng, Xue Bin Peng, Pieter Abbeel, Sergey Levine, Glen Berseth, and Koushil Sreenath. Reinforcement learning for robust parameterized loco- motion control of bipedal robots. In 2021 IEEE Interna- tional Conference on Robotics and Automation (ICRA) , pages 28...

    2021

  23. [33]

    Robust and versatile bipedal jumping control through multi-task re- inforcement learning

    Zhongyu Li, Xue Bin Peng, Pieter Abbeel, Sergey Levine, Glen Berseth, and Koushil Sreenath. Robust and versatile bipedal jumping control through multi-task re- inforcement learning. arXiv preprint arXiv:2302.09450 , 2023

    2023 arXiv

  24. [34]

    Reinforce- ment learning for versatile, dynamic, and robust bipedal locomotion control

    Zhongyu Li, Xue Bin Peng, Pieter Abbeel, Sergey Levine, Glen Berseth, and Koushil Sreenath. Reinforce- ment learning for versatile, dynamic, and robust bipedal locomotion control. arXiv preprint arXiv:2401.16889 , 2024

    2024 arXiv

  25. [35]

    Berkeley hu- manoid: A research platform for learning-based control

    Qiayuan Liao, Bike Zhang, Xuanyu Huang, Xiaoyu Huang, Zhongyu Li, and Koushil Sreenath. Berkeley hu- manoid: A research platform for learning-based control. arXiv preprint arXiv:2407.21781 , 2024

    2024 arXiv

  26. [36]

    Character controllers using motion vaes

    Hung Yu Ling, Fabio Zinno, George Cheng, and Michiel Van De Panne. Character controllers using motion vaes. ACM Transactions on Graphics (TOG), 39(4):40–1, 2020

    2020

  27. [37]

    Kitani, and Weipeng Xu

    Zhengyi Luo, Jinkun Cao, Josh Merel, Alexander Win- kler, Jing Huang, Kris M. Kitani, and Weipeng Xu. Universal humanoid motion representations for physics- based control. In The Twelfth International Confer- ence on Learning Representations , 2024. URL https: //openreview.net...

    2024

  28. [38]

    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. In The IEEE International Conference on Computer Vision (ICCV) , Oct 2019. URL https://amass.is.tue.mpg.de

    2019

  29. [39]

    Isaac gym: High performance gpu-based physics simulation for robot learning

    Viktor Makoviychuk, Lukasz Wawrzyniak, Yunrong Guo, Michelle Lu, Kier Storey, Miles Macklin, David Hoeller, Nikita Rudin, Arthur Allshire, Ankur Handa, et al. Isaac gym: High performance gpu-based physics simulation for robot learning. arXiv preprint arXiv:2108.10470, 2021

    2021 arXiv

  30. [40]

    Dynamic walk of a biped

    Hirofumi Miura and Isao Shimoyama. Dynamic walk of a biped. IJRR, 1984

    1984

  31. [41]

    Whole-body control of humanoid robots

    Federico L Moro and Luis Sentis. Whole-body control of humanoid robots. Humanoid Robotics: A reference, Springer, Dordrecht, 2019

    2019

  32. [42]

    A multimode teleoperation framework for humanoid loco-manipulation: An application for the icub robot

    Luigi Penco, Nicola Scianca, Valerio Modugno, Leonardo Lanari, Giuseppe Oriolo, and Serena Ivaldi. A multimode teleoperation framework for humanoid loco-manipulation: An application for the icub robot. IEEE Robotics and Automation Magazine , 26(4):73–82,

  33. [43]

    Learning agile robotic locomotion skills by imitating animals

    Xue Bin Peng, Erwin Coumans, Tingnan Zhang, Tsang- Wei Edward Lee, Jie Tan, and Sergey Levine. Learning agile robotic locomotion skills by imitating animals. In Robotics: Science and Systems , 07 2020. doi: 10.15607/ RSS.2020.XVI.064

    2020

  34. [44]

    Amp: Adversarial motion priors for stylized physics-based character control

    Xue Bin Peng, Ze Ma, Pieter Abbeel, Sergey Levine, and Angjoo Kanazawa. Amp: Adversarial motion priors for stylized physics-based character control. ACM Transac- tions on Graphics (ToG) , 40(4):1–20, 2021

    2021

  35. [45]

    Ase: Large-scale reusable adversarial skill embeddings for physically simulated characters

    Xue Bin Peng, Yunrong Guo, Lina Halper, Sergey Levine, and Sanja Fidler. Ase: Large-scale reusable adversarial skill embeddings for physically simulated characters. ACM Trans. Graph., 41(4), July 2022

    2022

  36. [46]

    From one hand to multiple hands: Imitation learning for dexterous manipulation from single-camera teleoperation

    Yuzhe Qin, Hao Su, and Xiaolong Wang. From one hand to multiple hands: Imitation learning for dexterous manipulation from single-camera teleoperation. IEEE Robotics and Automation Letters , 7(4):10873–10881, 2022

    2022

  37. [47]

    Real- world humanoid locomotion with reinforcement learning

    Ilija Radosavovic, Tete Xiao, Bike Zhang, Trevor Dar- rell, Jitendra Malik, and Koushil Sreenath. Real- world humanoid locomotion with reinforcement learning. arXiv:2303.03381, 2023

    2023 arXiv

  38. [48]

    Humanoid locomotion as next token prediction

    Ilija Radosavovic, Bike Zhang, Baifeng Shi, Jathushan Rajasegaran, Sarthak Kamat, Trevor Darrell, Koushil Sreenath, and Jitendra Malik. Humanoid locomotion as next token prediction. arXiv:2402.19469, 2024

    2024 arXiv

  39. [49]

    Dynamic loco- motion synchronization of bipedal robot and human operator via bilateral feedback teleoperation

    Joao Ramos and Sangbae Kim. Dynamic loco- motion synchronization of bipedal robot and human operator via bilateral feedback teleoperation. Sci- ence Robotics , 4(35):eaav4282, 2019. doi: 10.1126/ scirobotics.aav4282. URL https://www.science.org/doi/ abs/10.1126/scirobotics.aav4282

    2019 doi

  40. [50]

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

    St ´ephane Ross, Geoffrey Gordon, and Drew Bagnell. A reduction of imitation learning and structured prediction to no-regret online learning. In Proceedings of the four- teenth international conference on artificial intelligence and statistics, 2011

    2011

  41. [51]

    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

    2017 arXiv

  42. [52]

    Deep imitation learning for humanoid loco-manipulation through human teleoperation

    Mingyo Seo, Steve Han, Kyutae Sim, Seung Hyeon Bang, Carlos Gonzalez, Luis Sentis, and Yuke Zhu. Deep imitation learning for humanoid loco-manipulation through human teleoperation. In 2023 IEEE-RAS 22nd International Conference on Humanoid Robots (Hu- manoids), pages 1–8. IEEE, 2023

    2023

  43. [53]

    Blind bipedal stair traversal via sim-to-real reinforcement learning

    Jonah Siekmann, Kevin Green, John Warila, Alan Fern, and Jonathan Hurst. Blind bipedal stair traversal via sim-to-real reinforcement learning. arXiv preprint arXiv:2105.08328, 2021

    2021 arXiv

  44. [54]

    Hrp: Human affordances for robotic pre-training

    Mohan Kumar Srirama, Sudeep Dasari, Shikhar Bahl, and Abhinav Gupta. Hrp: Human affordances for robotic pre-training. In Proceedings of Robotics: Science and Systems, Delft, Netherlands, 2024

    2024

  45. [55]

    Humanmimic: Learning nat- ural locomotion and transitions for humanoid robot via wasserstein adversarial imitation

    Annan Tang, Takuma Hiraoka, Naoki Hiraoka, Fan Shi, Kento Kawaharazuka, Kunio Kojima, Kei Okada, and Masayuki Inaba. Humanmimic: Learning nat- ural locomotion and transitions for humanoid robot via wasserstein adversarial imitation. arXiv preprint arXiv:2309.14225, 2023

    2023 arXiv

  46. [56]

    Calm: Conditional adversarial latent models for directable virtual characters

    Chen Tessler, Yoni Kasten, Yunrong Guo, Shie Mannor, Gal Chechik, and Xue Bin Peng. Calm: Conditional adversarial latent models for directable virtual characters. In ACM SIGGRAPH 2023 Conference Proceedings, SIG- GRAPH ’23, New York, NY , USA, 2023. Association for Computing M...

    2023

  47. [57]

    Maskedmimic: Unified physics-based character control through masked motion inpainting

    Chen Tessler, Yunrong Guo, Ofir Nabati, Gal Chechik, and Xue Bin Peng. Maskedmimic: Unified physics-based character control through masked motion inpainting. ACM Transactions on Graphics (TOG) , 2024

    2024

  48. [58]

    Mimicplay: Long-horizon imitation learning by watching human play

    Chen Wang, Linxi Fan, Jiankai Sun, Ruohan Zhang, Li Fei-Fei, Danfei Xu, Yuke Zhu, and Anima Anand- kumar. Mimicplay: Long-horizon imitation learning by watching human play. arXiv preprint arXiv:2302.12422, 2023

    2023 arXiv

  49. [59]

    Karen Liu

    Chen Wang, Haochen Shi, Weizhuo Wang, Ruohan Zhang, Li Fei-Fei, and C. Karen Liu. Dexcap: Scalable and portable mocap data collection system for dexterous manipulation. arXiv preprint arXiv:2403.07788 , 2024

    2024 arXiv

  50. [60]

    Strategy and skill learning for physics-based table tennis animation

    Jiashun Wang, Jessica Hodgins, and Jungdam Won. Strategy and skill learning for physics-based table tennis animation. In ACM SIGGRAPH 2024 Conference Papers, pages 1–11, 2024

    2024

  51. [61]

    Hybrid zero dynamics of planar biped walkers

    Eric R Westervelt, Jessy W Grizzle, and Daniel E Koditschek. Hybrid zero dynamics of planar biped walkers. IEEE transactions on automatic control , 48(1): 42–56, 2003

    2003

  52. [62]

    Simbicon: Simple biped locomotion control

    KangKang Yin, Kevin Loken, and Michiel Van de Panne. Simbicon: Simple biped locomotion control. ACM Trans- actions on Graphics , 2007

    2007

  53. [63]

    Learning physically simulated tennis skills from broadcast videos

    Haotian Zhang, Ye Yuan, Viktor Makoviychuk, Yun- rong Guo, Sanja Fidler, Xue Bin Peng, and Kayvon Fatahalian. Learning physically simulated tennis skills from broadcast videos. ACM Trans. Graph. , 42(4), jul

  54. [64]

    Learning fine-grained bimanual manipulation with low-cost hardware

    Tony Z Zhao, Vikash Kumar, Sergey Levine, and Chelsea Finn. Learning fine-grained bimanual manipulation with low-cost hardware. arXiv preprint arXiv:2304.13705 , 2023. APPENDIX A. Real-world Deployment Our real robot employs a Unitree G1 platform, with an onboard Jetson Orin N...

    2023 arXiv

  55. [68]

    Evel ↓ Empkpe ↓ Eupper mpkpe ↓ Elower mpkpe ↓ Empjpe ↓ Eupper mpjpe ↓ Elower mpjpe ↓ (a) Eval

    Diverse enough (especially in upper-body movements) Metrics Training Dataset In dist. Evel ↓ Empkpe ↓ Eupper mpkpe ↓ Elower mpkpe ↓ Empjpe ↓ Eupper mpjpe ↓ Elower mpjpe ↓ (a) Eval. on D50 D50 ✓ 0.1375 0.0627 0.0571 0.0682 0.0753 0.0626 0.0928 D250 ✓ 0.1454 0.0669 0.0600 0.0738...

    1979

  56. [69]

    To illustrate how we arrived at this principle, We manually design three datasets of varying sizes, where the largest being the complete CMU dataset

    Feasible enough that lower-body motions do not exceed the robot’s mechanical limits, avoiding extreme samples (e.g., tumbling, handstands) that hamper training. To illustrate how we arrived at this principle, We manually design three datasets of varying sizes, where the larges...

  57. [70]

    D50 (in-distribution for the simplest data)

  58. [71]

    DCM U(the full, more complex dataset)

  59. [72]

    DACCAD , an out-of-distribution set containing actions not found in any of the training subsets. Table X summarizes our findings: • Evaluation on D50: Policies trained on D50 unsur- prisingly achieve the highest tracking accuracy for in- distribution actions, as reflected in m...

  60. [2019]

    doi: 10.1109/MRA.2019.2941245

    2019

  61. [2023]

    doi: 10.1145/3592408

    ISSN 0730-0301. doi: 10.1145/3592408. URL https://doi.org/10.1145/3592408

    doi

  62. [2024]

    URL https://arxiv.org/abs/2406.08858

    arXiv

Pith tools

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