Pith. sign in

REVIEW 4 major objections 6 minor 57 references

Quadruped motion tracking scales with data: a generalist flow policy improves as the library grows to thousands of clips.

Reviewed by Pith at T0; open to challenge. T0 means a machine referee read the full paper against a public rubric. the ladder, T0–T4 →

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

T0 review · grok-4.5

2026-07-10 19:09 UTC pith:ATAJP5N5

load-bearing objection Solid systems report: first large-scale quadruped generalist tracker with a real data-scaling curve, but “scaling law” and product claims outrun the shared-pool held-out design and mostly qualitative hardware. the 4 major comments →

arxiv 2607.07370 v2 pith:ATAJP5N5 submitted 2026-07-08 cs.RO cs.AIcs.HCcs.LG

Behavior Foundations for Quadruped Robots: ABot-C0 Technical Report

Xufeng Zhao , Fuzhi Yang , Jianhui Chen , Li Gao , Zhang Meng , Jie Gao , Yao Zheng , Congyang Zhao
show 12 more authors
Tianxiong Lv Menglin Yang Minqi Gu Yaru Zhao Wenyu Liu Honglin Han Shihui Su Zixiao Tang Liu Liu Mu Xu Yang Cai Wenbin Tang
This is my paper
classification cs.RO cs.AIcs.HCcs.LG
keywords quadruped robotsmotion trackingflow matchingbehavior foundation modelsall-terrain locomotionvideo-to-motion generationsim-to-realhuman-robot interaction
verification ladder T0 review T1 audit T2 compute T3 formal T4 reserved

The pith

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

Humanoid whole-body control has ridden large motion-capture libraries and motion tracking; quadrupeds have not, because animal motion is scarce and hard to retarget. This technical report argues that a multi-source data pyramid—video generation with identity-consistent conditioning, motion capture, teleoperation, and artist design—can produce tens of thousands of physically filtered clips that make a generalist quadruped controller possible. On that library, a specialist-to-generalist flow-matching tracker is the first to show a clear data scaling law: as training motions grow from tens to thousands, held-out tracking error falls and success rises, with zero-shot tracking of unseen clips. The same stack then layers biomimetic omnidirectional locomotion, a three-stage privileged-to-LiDAR all-terrain controller, a hand-shaking interaction case study, and a unified multi-policy deployment layer. The authors use urban navigation and companion-style multimodal interaction to argue that quadrupeds can move from isolated demos toward product-level behavioral intelligence.

Core claim

ABot-C0 claims that a scalable multi-source motion library of 16,074 physically feasible clips enables the first generalist quadruped motion-tracking controller, and that this controller exhibits a data scaling law: with a specialist-to-generalist flow-matching policy, increasing training motions from 30 to 7,076 systematically improves unseen tracking (MPJPE 24.61→14.79 mm, success 84.30%→88.54%) and narrows the seen–unseen gap, with further gains from manifold-calibrated reference conditioning.

What carries the argument

Specialist-to-generalist Flow-Matching distillation with Manifold-Calibrated Reference Conditioning (MCRC): per-clip PPO specialists are distilled via DAgger into one flow policy, then conditioned on a VAE latent of the local reference window so the student tracks a learned motion manifold rather than raw frame commands alone.

Load-bearing premise

That video-generated motions filtered by re-render similarity, reprojection thresholds, and per-clip simulation rollouts form a distribution whose closed-loop success will transfer to real robots without a large residual domain gap.

What would settle it

Train the same flow-matching generalist on increasing motion budgets (30 → full set) and measure held-out MPJPE and success on a fixed unseen set of 1,000 clips; if unseen error does not fall and the seen–unseen gap does not shrink as reported, the claimed scaling law fails.

Watch this falsifier — get emailed when new claim-graph text bears on it.

Editorial analysis

A structured set of objections, weighed in public.

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

Referee Report

4 major / 6 minor

Summary. ABot-C0 is a systems technical report for generalist quadruped motion control. It contributes (i) a multi-source data engine (teleoperation, artist design, MoCap, and identity-consistent video-to-motion generation with multi-stage filtering) yielding 16,074 physically validated clips; (ii) a specialist-to-generalist Flow-Matching motion tracker with Dynamic-Aware (PRF) curation and Manifold-Calibrated Reference Conditioning (MCRC), plus residual RL; (iii) a progressive locomotion stack (robust baseline with Barlow/NP3O, Diff-CAST biomimetic omnidirectional gait, and three-stage privileged-to-perceptive LiDAR all-terrain control); (iv) a compositional scene-interaction pipeline (hand-shaking via perception, goal locomotion, IK references, and tracking); and (v) a unified multi-policy deployment stack on the Tutu platform, demonstrated in companion HRI and all-terrain navigation. The headline empirical claim is a data scaling law for quadruped motion tracking (Table 4 / Fig. 6): as training motions grow from 30 to 7,076, unseen MPJPE falls and success rises, with further gains from MCRC (Table 6).

Significance. If the scaling and deployment claims hold under stronger distributional tests, this is a substantial systems contribution: quadruped motion data has lagged humanoid MoCap/video pipelines, and a reproducible specialist-to-generalist Flow-Matching tracker with ablations (Tables 3–6), locomotion safety/terrain ablations (Tables 7–12), and a working multi-policy real-robot stack would be valuable to the field. Strengths include clear specialist-vs-multi-motion-RL-vs-flow comparisons, fixed-budget PRF curation ablations, MCRC observation ablations, NP3O hardware-safety stress tests, and explicit limitations on multi-policy vs unified BFM design. The work is primarily empirical systems engineering rather than a closed-form derivation; its significance rests on whether “unseen” scaling and product-level demos generalize beyond the filtered multi-source pool and qualitative hardware showcases.

major comments (4)
  1. [§3.1 / Table 4 / Fig. 6] Abstract, §3.1, Table 4, Fig. 6: The central “scaling law / zero-shot unseen tracking” claim is only partially supported. Held-out motions (1,000) are drawn from the same multi-source library after the same CLIP, reprojection, and specialist physical-feasibility gates (§2.1.3), with video generation alone contributing 7,488/16,074 clips. Monotonic unseen improvement can therefore reflect denser coverage of an already-filtered manifold rather than distributional OOD generalization. Please either (a) report a truly external OOD split (e.g., held-out source type, animal MoCap not used in generation/retargeting, or real teleop-only holdout), or (b) reframe the claim as in-distribution scaling within a filtered multi-source pool and qualify “zero-shot” accordingly.
  2. [§2.1.3] §2.1.3 physical feasibility gate and §3.1 specialist pipeline: Defining “physically feasible” as “a per-motion specialist can complete a full-length sim rollout without termination” couples dataset construction to the same tracking family later distilled into the generalist. This is a reasonable engineering filter, but it biases the foundation set toward motions already solvable by the specialist recipe and weakens the claim that the data pyramid independently enables generalist scaling. Report rejection rates by failure mode, sensitivity to specialist hyperparameters, and at least one alternative feasibility criterion (e.g., trajectory optimization / dynamics residual thresholds without RL success).
  3. [§5.1 / §6] Sections 4–6 vs §5.1: Simulation tracking evidence is relatively strong (Tables 3–6), but real-robot support for the generalist tracker is largely qualitative (deployment architecture, companion demos, navigation). The weakest load-bearing assumption for product-level claims is sim-to-real transfer of filtered video-generated motions. Please add quantitative hardware tracking metrics on a fixed motion suite (MPJPE or joint/root errors, success/fall rates, energy) for seen vs held-out clips, and state how many video-generated vs MoCap/teleop references were executed on hardware.
  4. [Abstract / §1 / §7] §1 and abstract position ABot-C0 as establishing “behavior foundations” / a BFM-like stack, while §7 correctly notes it remains a coordinated multi-policy system. The title and abstract over-claim relative to the architecture (separate tracking, locomotion, interaction policies with arbitration). Tighten the framing to “systems foundations toward a quadruped BFM” unless a single conditioned policy is demonstrated, and clarify what is novel versus concurrent self-group pipelines (video generation, Diff-CAST, QuadFM) cited as data/method sources.
minor comments (6)
  1. [Table 1] Table 1: DogML comparison notes “redundant retargeted sequences”; make the unique-event definition and retargeting protocol fully explicit so diversity claims are auditable.
  2. [§3.1.1] Eqs. (3)–(5): Specify the Beta(1.5,1.0) schedule rationale and whether D=5 ODE steps was ablated for tracking quality vs latency on hardware.
  3. [Table 4] Table 4: Seen success declines slightly at full scale (92.74%) while unseen improves; discuss capacity/interference or curation effects rather than only the gap reduction.
  4. [§5.3] §5.3 Table 13: Final hand-target execution error (~13.7 cm mean) is large relative to IK planning (~1.2 cm). Clarify whether this is acceptable for contact HRI and how compliance/gain reduction contributes.
  5. [Fig. 1] Figure 1 / system diagrams are useful but dense; ensure all acronyms (MCRC, PRF, NP3O, SACC, Diff-CAST) are defined at first use in the main text consistently.
  6. [§1 / References] Several concurrent arXiv citations from the same group supply core data/methods; a short related-work paragraph disentangling prior vs new contributions would help readers and reviewers.

Circularity Check

1 steps flagged

Empirical systems report: scaling and MCRC are measured, not forced by definition; only mild selection via specialist feasibility gating.

specific steps
  1. other [§2.1.3 Physical feasibility gate; linked to §3.1 specialist-to-generalist tracking]
    "Physical feasibility gate. Even geometrically faithful trajectories may be physically infeasible. For each surviving trajectory, we train a per-motion specialist tracking policy using the controller described below and run a full-length rollout in simulation. Trajectories triggering any termination condition (fall, root divergence, velocity explosion) are discarded (97.6% pass rate)."

    “Physically feasible” is defined by successful closed-loop tracking with the same specialist controllers later distilled into the generalist. Motions that fail specialist rollouts are removed before generalist training/eval, so high success on the retained library is partly ensured by that selection. Effect is mild (only ~2.4% discarded) and does not by itself produce the reported data-scaling curve on held-out clips.

full rationale

ABot-C0 is a multi-component systems/technical report. Its central “scaling law” (Table 4 / Fig. 6) is an empirical train–eval curve: more training clips from a fixed multi-source library improve held-out MPJPE/success under the same specialist→DAgger→flow pipeline. That is standard ML measurement, not a first-principles derivation that reduces to its inputs. Held-out clips sharing the same generators and filters is a generalization-scope issue, not circularity by construction. Self-group citations (Diff-CAST [6], QuadFM/video pipelines [14,24]) supply concurrent components and data; they do not import a uniqueness theorem or smuggle an ansatz that forces the scaling numbers. The only mild circularity-adjacent step is operationalizing “physically feasible” via per-clip specialist rollouts that discard failures before generalist training—so retained-library trackability is partly selected for—but the pass rate is high (97.6%) and does not force the monotonic unseen scaling trend. Score 2 reflects that minor selection bias without elevating it to a load-bearing circular derivation.

Axiom & Free-Parameter Ledger

6 free parameters · 5 axioms · 4 invented entities

The central scaling and generalist-tracking claims rest on engineering choices and domain assumptions rather than a small set of mathematical axioms: video-model identity losses, filter thresholds, specialist-then-DAgger distillation, flow-matching action generation, and sim-to-real domain randomization. Free parameters are numerous (standard for deep RL systems). Invented named modules (MCRC, PRF, Diff-CAST, NP3O usage) are methods, not new physical entities; their independent evidence is the paper’s own ablations.

free parameters (6)
  • Identity consistency hinge margin m_id and weight λ on L_IC
    Chosen training knobs for Wan2.2 I2V fine-tuning that determine which generated videos keep rigid robot identity.
  • CLIP / reprojection filter thresholds (mean <20px, max <100px; CLIP pass rates reported)
    Hard cutoffs that define which recovered trajectories enter the training library; directly shape the scaling dataset.
  • PRF curation weights λ_p=0.45, λ_r=0.35, λ_f=0.20 and complexity binning
    Hand-set mixture that selects the fixed-budget motion subset used in curation experiments and manifold training.
  • Residual RL scale s=0.2 and clip c=0.5; action scale α=0.25; D=5 ODE steps
    Control and inference hyperparameters that affect reported tracking fidelity and deployable actions.
  • VAE latent dim 32, window H=20, β KL weight; history length 10
    Architecture choices for MCRC and the transformer-history student that drive the best reported unseen success.
  • NP3O constraint margins/penalty schedules and Diff-CAST diffusion/SACC hyperparameters
    Safety and style-prior knobs that determine locomotion safety and omnidirectional tracking tables.
axioms (5)
  • domain assumption Physics simulators (Isaac/MuJoCo) plus domain randomization sufficiently approximate real Tutu dynamics for transferred policies.
    Underpins all sim training and the claim of reliable real-world operation (Sections 3–6).
  • ad hoc to paper A trajectory that a specialist tracker can complete without termination is “physically feasible” enough to keep in the foundation dataset.
    Section 2.1.3 physical feasibility gate operationalizes feasibility via controller success, not independent biomechanical validation.
  • domain assumption Specialist ensemble + DAgger on student-induced states yields a valid multi-modal expert action distribution for flow matching.
    Standard imitation assumption adapted from humanoid pipelines; load-bearing for the generalist tracker (Section 3.1.1).
  • domain assumption Privileged terrain/dynamics labels in sim are adequate teachers for LiDAR memory students under sensor noise curricula.
    Three-stage all-terrain pipeline (Section 3.2.3) depends on this privileged-to-perceptive transfer.
  • standard math Standard RL/optimization math (PPO, flow matching, VAE ELBO, Barlow Twins) behaves as in prior literature.
    Background methods used throughout Sections 3.1–3.2 without re-proof.
invented entities (4)
  • ABot-C0 behavior foundation stack no independent evidence
    purpose: Name the integrated data+tracking+locomotion+deployment system claimed as quadruped BFM foundations.
    System brand rather than a physical entity; evidence is the paper’s own demos and tables.
  • Manifold-Calibrated Reference Conditioning (MCRC) no independent evidence
    purpose: Add VAE latent (and optional recon error) of reference windows as student conditions to improve generalist tracking.
    Method introduced in Section 3.1.1; support is internal ablation Table 6, not external replication.
  • PRF-score dynamic-aware motion curation no independent evidence
    purpose: Rank motions by physical feasibility, rollout executability, and flow confidence within complexity bins.
    Selection rule defined in Section 3.1.1; gains shown only in-paper under fixed budget.
  • Diff-CAST (diffusion prior + SACC) biomimetic locomotion module no independent evidence
    purpose: Produce naturalistic omnidirectional gaits without AMP mode collapse/heading drift.
    Presented as part of ABot-C0 and cross-cited to the authors’ related arXiv; independent external validation not shown here.

pith-pipeline@v1.1.0-grok45 · 29819 in / 4088 out tokens · 49939 ms · 2026-07-10T19:09:48.884628+00:00 · methodology

0 comments
read the original abstract

The motion controller is one of the most fundamental modules in embodied intelligence systems. Driven by large-scale human motion-capture data and the motion-tracking paradigm, humanoid control has achieved remarkable progress in recent years. However, migrating this recipe to the quadrupedal setting is far less straightforward: animal motion data is scarcer and harder to capture at scale than human data, and cross-embodiment retargeting remains fragile. We present ABot-C0, a generalist motion-control system for quadruped robots that establishes three complementary behavior foundations: a scalable multi-source motion-data pipeline, robust policy learning across motion tracking, locomotion, and scene interaction, and a unified deployment stack for reliable real-world operation. Fundamentally, we construct a data pyramid through conditional video-generation synthesis, annotated motion capture, teleoperation, and human design, producing 16,074 physically feasible motion clips as the data foundation for diverse motion-learning demands. With large-scale motion data, a Flow-Matching generalist policy demonstrates, for the first time, a scaling law for quadruped motion tracking: performance improves consistently as training scales up, with zero-shot capability to track unseen motions. We then go a step further toward robust all-terrain locomotion by adopting a three-stage privileged-to-perceptive framework with temporal LiDAR memory and terrain-predictive supervision. Collectively, these components form a motion generalist that coordinates multi-policy execution, smooth behavior transitions, energy-efficient control, and safety mechanisms for real-world deployment. Extensive experiments on urban-terrain autonomous navigation and companion-style multimodal interaction demonstrate that quadruped robots can move beyond functional demos toward product-level behavioral intelligence.

Review history (2 revisions) →

discussion (0)

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

Reference graph

Works this paper leans on

57 extracted references · 57 canonical work pages · 17 internal anchors

  1. [1]

    Michael Ahn, Anthony Brohan, Noah Brown, Yevgen Chebotar, Omar Cortes, Byron David, Chelsea Finn, Chuyuan Fu, Keerthana Gopalakrishnan, Karol Hausman, Alex Herzog, Daniel Ho, Jasmine Hsu, Julian Ibarz, Brian Ichter, Alex Irpan, Eric Jang, Rosario Jauregui Ruano, Kyle Jeffrey, Sally Jesmonth, Nikhil J. Joshi, Ryan Julian, Dmitry Kalashnikov, Yuheng Kuang, ...

    work page 2022

  2. [2]

    Visual imitation enables contextual humanoid control,

    Arthur Allshire, Hongsuk Choi, Junyi Zhang, David McAllister, Anthony Zhang, Chung Min Kim, Trevor Darrell, Pieter Abbeel, Jitendra Malik, and Angjoo Kanazawa. Visual imitation enables contextual humanoid control,

    work page

  3. [3]

    URLhttps://arxiv.org/abs/2505.03729

    work page internal anchor Pith review Pith/arXiv arXiv

  4. [4]

    Genesis: A generative and universal physics engine for robotics and beyond, 2024

    Genesis Authors. Genesis: A generative and universal physics engine for robotics and beyond, 2024. URL https://github.com/Genesis-Embodied-AI/Genesis

    work page 2024

  5. [5]

    Kevin Black, Noah Brown, James Darpinian, Karan Dhabalia, Danny Driess, Adnan Esmail, Michael Equi, Chelsea Finn, Niccolo Fusai, et al.π0.5: a vision-language-action model with open-world generalization.arXiv preprint arXiv:2504.16054, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  6. [6]

    Bones-seed: Skeletal everyday embodiment dataset, 2025

    Bones Studio. Bones-seed: Skeletal everyday embodiment dataset, 2025. URL https://huggingface.co/ datasets/bones-studio/seed. 142,220 annotated motion sequences from 522 actors

    work page 2025

  7. [7]

    Constraint-Aware Diffusion Priors for High-Fidelity and Versatile Quadruped Locomotion

    Jianhui Chen, Ruixin Zhan, Liu Liu, Yang Cai, and Ziqiao Li. Constraint-aware diffusion priors for high-fidelity and versatile quadruped locomotion, 2026. URLhttps://arxiv.org/abs/2605.08804

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  8. [8]

    HoloMotion-1 Technical Report

    Maiyue Chen, Kaihui Wang, Bo Zhang, Xihan Ma, Zhiyuan Yang, Yi Ren, Qijun Huang, Zihao Zhu, Yucheng Wang, and Zhizhong Su. Holomotion-1 technical report, 2026. URLhttps://arxiv.org/abs/2605.15336

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  9. [9]

    GMT: General Motion Tracking for Humanoid Whole-Body Control

    Zixuan Chen, Mazeyu Ji, Xuxin Cheng, Xuanbin Peng, Xue Bin Peng, and Xiaolong Wang. GMT: General motion tracking for humanoid whole-body control.arXiv preprint arXiv:2506.14770, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  10. [10]

    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

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  11. [11]

    Extreme parkour with legged robots

    Xuxin Cheng, Kexin Shi, Ananye Agarwal, and Deepak Pathak. Extreme parkour with legged robots. InIEEE International Conference on Robotics and Automation (ICRA), 2024

    work page 2024

  12. [12]

    Danny Driess, Fei Xia, Mehdi S. M. Sajjadi, Corey Lynch, Aakanksha Chowdhery, Brian Ichter, Ayzaan Wahid, Jonathan Tompson, Quan Vuong, Tianhe Yu, Wenlong Huang, Yevgen Chebotar, Pierre Sermanet, Daniel Duckworth, Sergey Levine, Vincent Vanhoucke, Karol Hausman, Marc Toussaint, Klaus Greff, Andy Zeng, Igor Mordatch, and Pete Florence. PaLM-E: An embodied ...

    work page 2023

  13. [13]

    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, 2024

    work page 2024

  14. [14]

    Doubly-fused vit: Fuse information from vision transformer doubly with local representation

    Li Gao, Dong Nie, Bo Li, and Xiaofeng Ren. Doubly-fused vit: Fuse information from vision transformer doubly with local representation. InEuropean Conference on Computer Vision, pages 744–761. Springer, 2022

    work page 2022

  15. [15]

    Quadfm: Foundational text-driven quadruped motion dataset for generation and control, 2026

    Li Gao, Fuzhi Yang, Jianhui Chen, Liu Liu, Yao Zheng, Yang Cai, and Ziqiao Li. Quadfm: Foundational text-driven quadruped motion dataset for generation and control, 2026. URLhttps://arxiv.org/abs/2603.24021

    work page arXiv 2026

  16. [16]

    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. arXiv preprint arXiv:2406.08858, 2024

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  17. [17]

    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, et al. ASAP: Aligning simulation and real-world physics for learning agile humanoid whole-body skills.arXiv preprint arXiv:2502.01143, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  18. [18]

    ExBody2: Advanced Expressive Humanoid Whole-Body Control

    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. 26

    work page internal anchor Pith review Pith/arXiv arXiv 2024

  19. [19]

    OpenVLA: An Open-Source Vision-Language-Action Model, September 2024

    Moo Jin Kim, Karl Pertsch, Siddharth Karamcheti, Ted Xiao, Ashwin Balakrishna, Suraj Nair, Rafael Rafailov, Ethan Foster, Grace Lam, Pannag Sanketi, Quan Vuong, Thomas Kollar, Benjamin Burchfiel, Russ Tedrake, Dorsa Sadigh, Sergey Levine, Percy Liang, and Chelsea Finn. OpenVLA: An Open-Source Vision-Language-Action Model, September 2024

    work page 2024

  20. [20]

    Phuma: Physically-grounded humanoid locomotion dataset

    Kyungmin Lee, Sibeen Kim, Minho Park, Hyunseung Kim, Dongyoon Hwang, Hojoon Lee, and Jaegul Choo. Phuma: Physically-grounded humanoid locomotion dataset. 2025

    work page 2025

  21. [21]

    Bfm-zero: A promptable behavioral foundation model for humanoid control using unsupervised reinforcement learning, 2025

    Yitang Li, Zhengyi Luo, Tonghe Zhang, Cunxi Dai, Anssi Kanervisto, Andrea Tirinzoni, Haoyang Weng, Kris Kitani, Mateusz Guzek, Ahmed Touati, Alessandro Lazaric, Matteo Pirotta, and Guanya Shi. Bfm-zero: A promptable behavioral foundation model for humanoid control using unsupervised reinforcement learning, 2025. URLhttps://arxiv.org/abs/2511.04131

    work page arXiv 2025

  22. [22]

    Truong, Xiaoyu Huang, Yuman Gao, Guy Tevet, Koushil Sreenath, and C

    Qiayuan Liao, Takara E. Truong, Xiaoyu Huang, Yuman Gao, Guy Tevet, Koushil Sreenath, and C. Karen Liu. BeyondMimic: From Motion Tracking to Versatile Humanoid Control via Guided Diffusion, November 2025

    work page 2025

  23. [23]

    Visual whole-body control for legged loco-manipulation.The 8th Conference on Robot Learning, 2024

    Minghuan Liu, Zixuan Chen, Xuxin Cheng, Yandong Ji, Rizhao Qiu, Ruihan Yang, and Xiaolong Wang. Visual whole-body control for legged loco-manipulation.The 8th Conference on Robot Learning, 2024

    work page 2024

  24. [24]

    Comatrack: Competitive multi-agent game-theoretic tracking with vision-language-action models.arXiv preprint arXiv:2603.22846, 2026

    Youzhi Liu, Li Gao, Liu Liu, Mingyang Lv, and Yang Cai. Comatrack: Competitive multi-agent game-theoretic tracking with vision-language-action models.arXiv preprint arXiv:2603.22846, 2026

    work page arXiv 2026

  25. [25]

    Unleashing Infinite Motion: Scaling Expressive Quadrupedal Motion via Generative Video Priors

    Youzhi Liu, Li Gao, Yifei Qian, Liu Liu, Yang Cai, and Ziqiao Li. Unleashing infinite motion: Scaling expressive quadrupedal motion via generative video priors.arXiv preprint arXiv:2606.28237, 2026. URLhttps://arxiv. org/abs/2606.28237

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  26. [26]

    Omnixtreme: Breaking the generality barrier in high-dynamic humanoid control.arXiv preprint arXiv:2602.23843, 2026

    Zhengyi Luo, Wenhan Guo, Dhruv Tirumala, and Xue Bin Peng. Omnixtreme: Breaking the generality barrier in high-dynamic humanoid control.arXiv preprint arXiv:2602.23843, 2026

    work page arXiv 2026

  27. [27]

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

    Zhengyi Luo, Ye Yuan, Tingwu Wang, Chenran Li, Fernando Castañeda, Sirui Chen, Zi-Ang Cao, Jiefeng Li, David Minor, Qingwei Ben, Jinhyung Park, David Sami, Zi Wang, Xingye Da, Runyu Ding, Cyrus Hogg, Lina Song, Edy Lim, Eugene Jeong, Tairan He, Haoru Xue, Wenli Xiao, Simon Yuen, Jan Kautz, Yan Chang, Umar Iqbal, Linxi "Jim" Fan, and Yuke Zhu. Sonic: Super...

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  28. [28]

    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. InThe IEEE International Conference on Computer Vision (ICCV), Oct

    work page

  29. [29]

    URLhttps://amass.is.tue.mpg.de

    work page

  30. [30]

    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, and Gavriel State. Isaac gym: High performance gpu-based physics simulation for robot learning. InThirty-fifth Conference on Neural Information Processing Systems Datasets and Benchmarks Track, 2021

    work page 2021

  31. [31]

    Walk these ways: Tuning robot control for generalization with multiplicity of behavior

    Gabriel B Margolis and Pulkit Agrawal. Walk these ways: Tuning robot control for generalization with multiplicity of behavior. InConference on Robot Learning (CoRL), 2023

    work page 2023

  32. [32]

    Learning robust perceptive locomotion for quadrupedal robots in the wild.Science Robotics, 7(62):eabk2822, 2022

    Takahiro Miki, Joonho Lee, Jemin Hwangbo, Lorenz Wellhauer, Vladlen Koltun, and Marco Hutter. Learning robust perceptive locomotion for quadrupedal robots in the wild.Science Robotics, 7(62):eabk2822, 2022

    work page 2022

  33. [33]

    Orbit: A Unified Simulation Framework for Interactive Robot Learning Environments

    Mayank Mittal, Calvin Yu, Qinxi Yu, Jingzhou Liu, Nikita Rudin, David Hoeller, Jia Lin Yuan, Ritvik Singh, Yunrong Guo, Hammad Mazhar, et al. Isaac lab: A unified and modular framework for robot learning.arXiv preprint arXiv:2301.04195, 2023

    work page internal anchor Pith review Pith/arXiv arXiv 2023

  34. [34]

    Dreamwaq++: Obstacle-aware quadrupedal locomotion with resilient multimodal reinforcement learning.IEEE Transactions on Robotics, 42:819–836, 2026

    I Made Aswin Nahrendra, Byeongho Yu, Minho Oh, Dongkyu Lee, Seunghyun Lee, Hyeonwoo Lee, Hyungtae Lim, and Hyun Myung. Dreamwaq++: Obstacle-aware quadrupedal locomotion with resilient multimodal reinforcement learning.IEEE Transactions on Robotics, 42:819–836, 2026. ISSN 1941-0468. doi: 10.1109/tro.2026.3653774. URLhttp://dx.doi.org/10.1109/TRO.2026.3653774

    work page doi:10.1109/tro.2026.3653774 2026

  35. [35]

    Newton: Nvidia’s next-generation physics engine for robot learning, 2025

    NVIDIA. Newton: Nvidia’s next-generation physics engine for robot learning, 2025. URLhttps://developer. nvidia.com/newton. Announced at GTC 2025

    work page 2025

  36. [36]

    Maxime Oquab, Timothée Darcet, Theo Moutakanni, Huy V. Vo, Marc Szafraniec, Vasil Khalidov, Pierre Fernandez, Daniel Haziza, Francisco Massa, Alaaeldin El-Nouby, Russell Howes, Po-Yao Huang, Hu Xu, Vasu Sharma, Shang-Wen Li, Wojciech Galuba, Mike Rabbat, Mido Assran, Nicolas Ballas, Gabriel Synnaeve, Ishan 27 Misra, Herve Jegou, Julien Mairal, Patrick Lab...

    work page 2023

  37. [37]

    ASE: Large-Scale Reusable Adversarial Skill Embeddings for Physically Simulated Characters.ACM Transactions on Graphics, 41(4):1–17, July 2022

    Xue Bin Peng, Yunrong Guo, Lina Halper, Sergey Levine, and Sanja Fidler. ASE: Large-Scale Reusable Adversarial Skill Embeddings for Physically Simulated Characters.ACM Transactions on Graphics, 41(4):1–17, July 2022. ISSN 0730-0301, 1557-7368. doi: 10.1145/3528223.3530110

    work page doi:10.1145/3528223.3530110 2022

  38. [38]

    Learning Human-like Hand Reaching for Human-Robot Handshaking, March 2021

    Vignesh Prasad, Ruth Stock-Homburg, and Jan Peters. Learning Human-like Hand Reaching for Human-Robot Handshaking, March 2021

    work page 2021

  39. [39]

    Learning Transferable Visual Models From Natural Language Supervision

    Alec Radford, Jong Wook Kim, Chris Hallacy, Aditya Ramesh, Gabriel Goh, Sandhini Agarwal, Girish Sastry, Amanda Askell, Pamela Mishkin, Jack Clark, Gretchen Krueger, and Ilya Sutskever. Learning transferable visual models from natural language supervision, 2021. URLhttps://arxiv.org/abs/2103.00020

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  40. [40]

    Gordon, and Drew Bagnell

    Stéphane Ross, Geoffrey J. Gordon, and Drew Bagnell. A reduction of imitation learning and structured prediction to no-regret online learning. InInternational Conference on Artificial Intelligence and Statistics (AISTATS), 2011

    work page 2011

  41. [41]

    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

    work page internal anchor Pith review Pith/arXiv arXiv 2017

  42. [42]

    Non-verbal Interaction and Interface with a Quadruped Robot using Body and Hand Gestures: Design and User Experience Evaluation, August 2024

    Soohyun Shin, Trevor Evetts, Hunter Saylor, Hyunji Kim, Soojin Woo, Wonhwha Rhee, and Seong-Woo Kim. Non-verbal Interaction and Interface with a Quadruped Robot using Body and Hand Gestures: Design and User Experience Evaluation, August 2024

    work page 2024

  43. [43]

    Hierarchical adaptive loco-manipulation control for quadruped robots

    Mohsen Sombolestan and Quan Nguyen. Hierarchical adaptive loco-manipulation control for quadruped robots. In2023 IEEE International Conference on Robotics and Automation (ICRA), pages 12156–12162, 2023. doi: 10.1109/ICRA48891.2023.10160523

    work page doi:10.1109/icra48891.2023.10160523 2023

  44. [44]

    Mujoco: A physics engine for model-based control

    Emanuel Todorov, Tom Erez, and Yuval Tassa. Mujoco: A physics engine for model-based control. In2012 IEEE/RSJ International Conference on Intelligent Robots and Systems, pages 5026–5033. IEEE, 2012

    work page 2012

  45. [45]

    Advanced skills through multiple adversarial motion priors in reinforcement learning, 2022

    Eric Vollenweider, Marko Bjelonic, Victor Klemm, Nikita Rudin, Joonho Lee, and Marco Hutter. Advanced skills through multiple adversarial motion priors in reinforcement learning, 2022. URLhttps://arxiv.org/abs/2203. 14912

    work page 2022

  46. [46]

    Wan: Open and Advanced Large-Scale Video Generative Models

    Team Wan, Ang Wang, Baole Ai, Bin Wen, Chaojie Mao, Chen-Wei Xie, Di Chen, Feiwu Yu, Haiming Zhao, Jianxiao Yang, Jianyuan Zeng, Jiayu Wang, Jingfeng Zhang, Jingren Zhou, Jinkai Wang, Jixuan Chen, Kai Zhu, Kang Zhao, Keyu Yan, Lianghua Huang, Mengyang Feng, Ningyi Zhang, Pandeng Li, Pingyu Wu, Ruihang Chu, Ruili Feng, Shiwei Zhang, Siyang Sun, Tao Fang, T...

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  47. [47]

    T2qrm: Text-driven quadruped robot motion generation

    Minghui Wang, Zixu Wang, Hongbin Xu, Kun Hu, Zhiyong Wang, and Wenxiong Kang. T2qrm: Text-driven quadruped robot motion generation. InProceedings of the 6th ACM International Conference on Multimedia in Asia, MMAsia ’24, New York, NY, USA, 2024. Association for Computing Machinery. ISBN 9798400712739. doi: 10.1145/3696409.3700230. URLhttps://doi.org/10.11...

    work page doi:10.1145/3696409.3700230 2024

  48. [48]

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

    Yuxuan Wang, Ming Yang, Ziluo Ding, Yu Zhang, Weishuai Zeng, Xinrun Xu, Haobin Jiang, and Zongqing Lu. From experts to a generalist: Toward general whole-body control for humanoid robots.arXiv preprint arXiv:2506.12779, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  49. [49]

    ENPIRE: Agentic Robot Policy Self-Improvement in the Real World

    Wenli Xiao, Jia Xie, Tonghe Zhang, Haotian Lin, Letian "Max" Fu, Haoru Xue, Jalen Lu, Yi Yang, Cunxi Dai, Zi Wang, Jimmy Wu, Guanzhi Wang, S. Shankar Sastry, Ken Goldberg, Linxi "Jim" Fan, Yuke Zhu, and Guanya Shi. ENPIRE: Agentic Robot Policy Self-Improvement in the Real World. https://arxiv.org/abs/2606.19980v1, June 2026

    work page internal anchor Pith review Pith/arXiv arXiv 2026

  50. [50]

    ViTPose: Simple vision transformer baselines for human pose estimation

    Yufei Xu, Jing Zhang, Qiming Zhang, and Dacheng Tao. ViTPose: Simple vision transformer baselines for human pose estimation. InAdvances in Neural Information Processing Systems, 2022

    work page 2022

  51. [51]

    Barlow twins: Self-supervised learning via redundancy reduction

    Jure Zbontar, Li Jing, Ishan Misra, Yann LeCun, and Stéphane Deny. Barlow twins: Self-supervised learning via redundancy reduction. InInternational conference on machine learning, pages 12310–12320. PMLR, 2021. 28

    work page 2021

  52. [52]

    Twist2: Scalable, portable, and holistic humanoid data collection system.arXiv preprint arXiv:2511.02832, 2025

    Yanjie Ze, Siheng Zhao, Weizhuo Wang, Angjoo Kanazawa, Rocky Duan, Pieter Abbeel, Guanya Shi, Jiajun Wu, and C Karen Liu. Twist2: Scalable, portable, and holistic humanoid data collection system.arXiv preprint arXiv:2511.02832, 2025

    work page arXiv 2025

  53. [53]

    Track any motions under any disturbances.arXiv preprint arXiv:2509.13833, 2025

    Zhikai Zhang, Jun Guo, Chao Chen, Jilong Wang, Chenghuai Lin, Yunrui Lian, Han Xue, Zhenrong Wang, Maoqi Liu, Jiangran Lyu, et al. Track any motions under any disturbances.arXiv preprint arXiv:2509.13833, 2025

    work page arXiv 2025

  54. [54]

    Chat with the environment: Interactive multimodal perception using large language models

    Xufeng Zhao, Mengdi Li, Cornelius Weber, Muhammad Burhan Hafez, and Stefan Wermter. Chat with the environment: Interactive multimodal perception using large language models. In2023 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 3590–3596, 2023. doi: 10.1109/IROS55552.2023. 10342363

    work page doi:10.1109/iros55552.2023 2023

  55. [55]

    Agentic skill discovery.Robotics and Autonomous Systems, 196(C), June 2026

    Xufeng Zhao, Cornelius Weber, and Stefan Wermter. Agentic skill discovery.Robotics and Autonomous Systems, 196(C), June 2026. ISSN 0921-8890. doi: 10.1016/j.robot.2025.105248

    work page doi:10.1016/j.robot.2025.105248 2026

  56. [56]

    Robot parkour learning

    Ziwen Zhuang, Zipeng Fu, Jianren Wang, Christopher Atkeson, Sören Schwertfeger, Chelsea Finn, and Hang Zhao. Robot parkour learning. InConference on Robot Learning (CoRL), 2023

    work page 2023

  57. [57]

    Sanketi, Grecia Salazar, Michael S

    Brianna Zitkovich, Tianhe Yu, Sichun Xu, Peng Xu, Ted Xiao, Fei Xia, Jialin Wu, Paul Wohlhart, Stefan Welker, Ayzaan Wahid, Quan Vuong, Vincent Vanhoucke, Huong Tran, Radu Soricut, Anikait Singh, Jaspiar Singh, Pierre Sermanet, Pannag R. Sanketi, Grecia Salazar, Michael S. Ryoo, Krista Reymann, Kanishka Rao, Karl Pertsch, Igor Mordatch, Henryk Michalewski...

    work page 2023