pith. sign in

arxiv: 2602.03205 · v2 · pith:PT3OQJAMnew · submitted 2026-02-03 · 💻 cs.RO

HUSKY: Humanoid Skateboarding System via Physics-Aware Whole-Body Control

Jinrui Han , Dewei Wang , Chenyun Zhang , Xinzhe Liu , Ping Luo , Chenjia Bai , Xuelong Li This is my paper

Pith reviewed 2026-05-22 11:16 UTC · model grok-4.3

classification 💻 cs.RO
keywords humanoid robotskateboardingwhole-body controlphysics-aware controldynamic balancemotion priorsnon-holonomic constraintsunderactuated platform
0
0 comments X

The pith

A humanoid robot achieves stable real-world skateboarding by modeling the coupling of board tilt to truck steering and combining it with learned pushing motions under physics-guided whole-body control.

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

The paper sets out to show that humanoid robots can manage highly dynamic tasks like skateboarding on underactuated wheeled platforms, where balance, non-holonomic motion, and object interaction all occur at once. Most existing whole-body controllers assume fixed surroundings and therefore cannot handle the rapid changes and coupled forces that arise when a robot stands on a moving skateboard. By first deriving the relationship between how the board tilts and how its trucks steer, the authors create a model that lets them analyze the overall dynamics; they then train human-like pushing actions with adversarial motion priors and steer by leaning according to simple physics rules, using a trajectory planner to switch smoothly between phases. If the approach holds, robots gain a concrete route to performing agile maneuvers on objects that move and tip under their own weight.

Core claim

The authors establish that modeling the coupling relationship between board tilt and truck steering angles supplies the dynamic analysis needed to combine Adversarial Motion Priors for learning pushing motions with a physics-guided, heading-oriented lean-to-steer policy and a trajectory-guided transition mechanism, enabling the Unitree G1 humanoid to execute stable and agile skateboarding maneuvers on real hardware.

What carries the argument

The modeled coupling between board tilt and truck steering angles, which supports analysis of system dynamics and underpins the physics-guided lean-to-steer strategy together with the trajectory-guided phase transitions.

If this is right

  • The robot can generate pushing motions that resemble human technique while remaining balanced on the moving board.
  • Lean-to-steer commands derived from the tilt-steering model produce consistent heading changes without separate steering actuators.
  • Trajectory guidance maintains continuity when the robot switches from pushing to steering phases.
  • Real-world tests on the physical Unitree G1 confirm that the integrated system produces agile maneuvering without falling.

Where Pith is reading between the lines

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

  • The same tilt-coupling modeling technique could be applied to other underactuated balance tasks such as riding a bicycle or balancing on a moving platform.
  • Accurate offline physics models of this kind may reduce the need for dense onboard sensing in future humanoid locomotion systems.
  • Testing the framework on skateboards of varying wheelbase or wheel hardness would reveal how sensitive the closed-loop performance is to changes in the assumed coupling parameters.

Load-bearing premise

The relationship between board tilt and truck steering angles can be captured accurately enough by the model to support closed-loop balance without further online adaptation or extra direct sensing of the board state.

What would settle it

If the robot repeatedly loses balance or falls during sustained turns or when the skateboard encounters small surface irregularities, despite using the modeled coupling, the claim that the physics-guided strategy suffices for stable control would be refuted.

Figures

Figures reproduced from arXiv: 2602.03205 by Chenjia Bai, Chenyun Zhang, Dewei Wang, Jinrui Han, Ping Luo, Xinzhe Liu, Xuelong Li.

Figure 1
Figure 1. Figure 1: Overview. (a) Our proposed framework HUSKY enables the humanoid robot to perform complete real-world skateboarding, including pushing, steering, and phase transitions. (b) Generalization to diverse outdoor scenarios and skateboards with consistent stability and control. (c) Reliable indoor skateboarding performance. (d) Lean-to-steer behaviors achieved by exploiting robot body tilt. (e) Robustness against … view at source ↗
Figure 2
Figure 2. Figure 2: Skateboard Model. We analyze the skateboard kinematic structure and derive the coupling relationships among the board tilt, truck steering, and rake angles, which form the basis of the lean-to-steer behavior. that the truck steering angle is determined by the board tilt angle, with larger tilts producing greater steering deflections. Detailed derivation is provided in Appendix A. This abstraction directly … view at source ↗
Figure 3
Figure 3. Figure 3: Framework of HUSKY. (a) We first analyze and model the humanoid–skateboard system, deriving a physics-inspired lean-to-steer coupling mechanism. Due to the distinct contact dynamics and control objectives across skateboarding phases, we adopt a phase-wise learning strategy. (b) The learning-based whole-body control framework integrates an AMP-based pushing style for active forward propulsion, a steering st… view at source ↗
Figure 4
Figure 4. Figure 4: Steering Trajectories Visualizations. (a) Omitting lean-to-steer coupling prevents effective steering. (b) Incorporating physics-guided tilt guidance substantially increases the reachable heading range and precision [PITH_FULL_IMAGE:figures/full_fig_p007_4.png] view at source ↗
Figure 7
Figure 7. Figure 7: Details of feet motions during transitions. [PITH_FULL_IMAGE:figures/full_fig_p008_7.png] view at source ↗
Figure 8
Figure 8. Figure 8: Effect of Skateboard Physical Identification. [PITH_FULL_IMAGE:figures/full_fig_p008_8.png] view at source ↗
Figure 9
Figure 9. Figure 9: Kinematic geometry of a skateboard truck. [PITH_FULL_IMAGE:figures/full_fig_p011_9.png] view at source ↗
Figure 10
Figure 10. Figure 10: Skateboard Model in Training. Red and blue markers indicate stabel foot placement points, while thelight green areas denote collision zones used for foot-board collision detection. B. Skateboard Model The key components of our simplified skateboard model in MuJoCo are summarized in Table III and Table IV. In simula￾tion, the wheel–ground contact is modeled with six dimensions (condim = 6), including two t… view at source ↗
Figure 11
Figure 11. Figure 11: (a), while the canonical reference poses for the pushing and steering phases are shown in [PITH_FULL_IMAGE:figures/full_fig_p013_11.png] view at source ↗
Figure 12
Figure 12. Figure 12: Humanoid Skateboarding on Diverse Boards. [PITH_FULL_IMAGE:figures/full_fig_p013_12.png] view at source ↗
read the original abstract

While current humanoid whole-body control frameworks predominantly rely on the static environment assumptions, addressing tasks characterized by high dynamism and complex interactions presents a formidable challenge. In this paper, we address humanoid skateboarding, a highly challenging task requiring stable dynamic maneuvering on an underactuated wheeled platform. This integrated system is governed by non-holonomic constraints and tightly coupled human-object interactions. Successfully executing this task requires simultaneous mastery of hybrid contact dynamics and robust balance control on a mechanically coupled, dynamically unstable skateboard. To overcome the aforementioned challenges, we propose HUSKY, a learning-based framework that integrates humanoid-skateboard system modeling and physics-aware whole-body control. We first model the coupling relationship between board tilt and truck steering angles, enabling a principled analysis of system dynamics. Building upon this, HUSKY leverages Adversarial Motion Priors (AMP) to learn human-like pushing motions and employs a physics-guided, heading-oriented strategy for lean-to-steer behaviors. Moreover, a trajectory-guided mechanism ensures smooth and stable transitions between pushing and steering. Experimental results on the Unitree G1 humanoid platform demonstrate that our framework enables stable and agile maneuvering on skateboards in real-world scenarios. The project page is available on https://husky-humanoid.github.io/.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

2 major / 2 minor

Summary. The paper presents HUSKY, a learning-based framework for humanoid skateboarding on an underactuated wheeled platform. It models the coupling between board tilt and truck steering angles, uses Adversarial Motion Priors (AMP) to learn human-like pushing motions, applies a physics-guided heading-oriented lean-to-steer strategy, and incorporates a trajectory-guided mechanism for smooth transitions between pushing and steering. Real-world experiments on the Unitree G1 humanoid are reported to demonstrate stable and agile maneuvering under non-holonomic constraints and hybrid contact dynamics.

Significance. If the central claims hold under detailed scrutiny, this represents a meaningful step forward in humanoid whole-body control for highly dynamic and interactive tasks. The integration of explicit physics modeling of the skateboard coupling with learning-based motion priors addresses a challenging combination of underactuation, non-holonomic constraints, and balance, which could inform future work on agile locomotion and object interaction. The real-robot validation on a standard platform is a positive indicator of practical relevance.

major comments (2)
  1. [§3 (System Modeling and Coupling)] §3 (System Modeling and Coupling): The framework depends on a pre-modeled static coupling relationship between board tilt and truck steering angles to enable the physics-guided lean-to-steer strategy and closed-loop stability. The manuscript should provide analysis or sensitivity results showing that this model remains sufficiently accurate under real-world deviations (e.g., friction changes, wear, or tolerances) without online adaptation or extra board-state sensing, as deviations could undermine trajectory-guided transitions and balance during agile maneuvers.
  2. [§5 (Experimental Validation)] §5 (Experimental Validation): The reported real-robot experiments on the Unitree G1 claim stable and agile maneuvering, but the absence of quantitative metrics (success rates, stability margins, trajectory tracking errors), failure cases, or ablation studies (e.g., with vs. without the physics-guided component) makes it difficult to verify robustness. This weakens the load-bearing claim that the integrated framework reliably handles the dynamic interactions.
minor comments (2)
  1. [Abstract] The abstract could be tightened by explicitly naming the main technical contributions (modeling, AMP, physics-guided strategy, trajectory guidance) rather than listing them narratively.
  2. [Figures] Figure captions and legends should more clearly indicate experimental conditions, such as surface type, speed ranges, or number of trials, to improve reproducibility.

Simulated Author's Rebuttal

2 responses · 0 unresolved

We thank the referee for the constructive feedback on our manuscript. We address each major comment in detail below and describe the revisions we will make to improve clarity and rigor.

read point-by-point responses

  1. Referee: [§3 (System Modeling and Coupling)] §3 (System Modeling and Coupling): The framework depends on a pre-modeled static coupling relationship between board tilt and truck steering angles to enable the physics-guided lean-to-steer strategy and closed-loop stability. The manuscript should provide analysis or sensitivity results showing that this model remains sufficiently accurate under real-world deviations (e.g., friction changes, wear, or tolerances) without online adaptation or extra board-state sensing, as deviations could undermine trajectory-guided transitions and balance during agile maneuvers.

    Authors: We appreciate the referee's emphasis on validating the robustness of the static coupling model. This relationship is derived directly from the fixed geometric parameters of the skateboard trucks and board, which exhibit limited variation in standard hardware. Real-world experiments on the Unitree G1 already demonstrate that the closed-loop system maintains balance and executes agile maneuvers without online adaptation or additional board-state sensors, indicating that the model is sufficiently accurate for the tested conditions. To address the concern more explicitly, we will add a sensitivity analysis to the revised Section 3. This will quantify the impact of parameter deviations (e.g., friction coefficient changes and mechanical tolerances) on steering response and stability margins, using both simulation sweeps and supplementary real-robot trials. revision: yes

  2. Referee: [§5 (Experimental Validation)] §5 (Experimental Validation): The reported real-robot experiments on the Unitree G1 claim stable and agile maneuvering, but the absence of quantitative metrics (success rates, stability margins, trajectory tracking errors), failure cases, or ablation studies (e.g., with vs. without the physics-guided component) makes it difficult to verify robustness. This weakens the load-bearing claim that the integrated framework reliably handles the dynamic interactions.

    Authors: We agree that the current experimental section would benefit from more quantitative support. The manuscript demonstrates feasibility through successful real-world skateboarding trials under non-holonomic and hybrid-contact conditions. In the revision, we will expand Section 5 to report quantitative metrics including trial success rates, root-mean-square trajectory tracking errors, and approximate stability margins derived from center-of-mass and tilt observations. We will also add ablation comparisons (full framework versus variants without the physics-guided lean-to-steer term) and a concise discussion of observed failure modes with recovery behaviors. These additions will strengthen the evidence for robustness. revision: yes

Circularity Check

0 steps flagged

No significant circularity; modeling and control components grounded in external priors and experiments

full rationale

The paper models the board-truck coupling relationship as an input to dynamics analysis and then applies AMP-based learning plus a physics-guided heading strategy for lean-to-steer. These steps rely on stated external priors (non-holonomic constraints, AMP) and real-world Unitree G1 testing rather than reducing any reported performance metric to a quantity defined or fitted inside the same paper. No load-bearing prediction is shown to equal its own input by construction, and self-citations (if present) are not required to justify the central claims. This yields a low circularity score consistent with a self-contained engineering framework.

Axiom & Free-Parameter Ledger

1 free parameters · 1 axioms · 0 invented entities

The central claim rests on the assumption that the modeled coupling between tilt and steering is accurate enough for control, plus standard assumptions from reinforcement learning and physics simulation that are not derived in the paper.

free parameters (1)
  • weights in physics-guided lean-to-steer strategy
    Control gains or scaling factors for converting lean angles to steering commands are likely tuned for the specific robot and board.
axioms (1)
  • domain assumption Non-holonomic constraints govern the skateboard dynamics
    Invoked when describing the integrated system as requiring mastery of hybrid contact dynamics.

pith-pipeline@v0.9.0 · 5771 in / 1305 out tokens · 24244 ms · 2026-05-22T11:16:00.507998+00:00 · methodology

discussion (0)

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

Reference graph

Works this paper leans on

49 extracted references · 49 canonical work pages · 4 internal anchors

  1. [1]

    Visual imitation enables contextual humanoid control.arXiv preprint arXiv:2505.03729, 2025

    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.arXiv preprint arXiv:2505.03729, 2025

    work page arXiv 2025

  2. [2]

    Ai flow: Perspectives, scenarios, and approaches.Vicinagearth, 3(1):1, 2026

    Hongjun An, Wenhan Hu, Sida Huang, Siqi Huang, Ruanjun Li, Yuanzhi Liang, Jiawei Shao, Yiliang Song, Zihan Wang, Cheng Yuan, et al. Ai flow: Perspectives, scenarios, and approaches.Vicinagearth, 3(1):1, 2026

    work page 2026

  3. [4]

    Quadrupedal robot skateboard mounting via reverse cur- riculum learning.arXiv preprint arXiv:2505.06561, 2025

    Danil Belov, Artem Erkhov, Elizaveta Pestova, Ilya Osokin, Dzmitry Tsetserukou, and Pavel Osinenko. Quadrupedal robot skateboard mounting via reverse cur- riculum learning.arXiv preprint arXiv:2505.06561, 2025

    work page arXiv 2025

  4. [5]

    Homie: Humanoid loco-manipulation with isomorphic exoskeleton cockpit.arXiv preprint arXiv:2502.13013, 2025

    Qingwei Ben, Feiyu Jia, Jia Zeng, Junting Dong, Dahua Lin, and Jiangmiao Pang. Homie: Humanoid loco- manipulation with isomorphic exoskeleton cockpit.arXiv preprint arXiv:2502.13013, 2025

    work page arXiv 2025

  5. [6]

    Humanplus: Hu- manoid shadowing and imitation from humans.arXiv preprint arXiv:2406.10454, 2024

    Zipeng Fu, Qingqing Zhao, Qi Wu, Gordon Wet- zstein, and Chelsea Finn. Humanplus: Humanoid shad- owing and imitation from humans.arXiv preprint arXiv:2406.10454, 2024

    work page arXiv 2024

  6. [7]

    Skater: Synthesized kinematics for advanced traversing efficiency on a humanoid robot via roller skate swizzles.arXiv preprint arXiv:2601.04948, 2026

    Junchi Gu, Feiyang Yuan, Weize Shi, Tianchen Huang, Haopeng Zhang, Xiaohu Zhang, Yu Wang, Wei Gao, and Shiwu Zhang. Skater: Synthesized kinematics for advanced traversing efficiency on a humanoid robot via roller skate swizzles.arXiv preprint arXiv:2601.04948, 2026

    work page arXiv 2026

  7. [8]

    Humanoid-gym: Reinforcement learning for humanoid robot with zero-shot sim2real transfer.arXiv preprint arXiv:2404.05695, 2024

    Xinyang Gu, Yen-Jen Wang, and Jianyu Chen. Humanoid-gym: Reinforcement learning for humanoid robot with zero-shot sim2real transfer.arXiv preprint arXiv:2404.05695, 2024

    work page arXiv 2024

  8. [9]

    Advancing humanoid locomotion: Mastering challenging terrains with denoising world model learning.arXiv preprint arXiv:2408.14472, 2024

    Xinyang Gu, Yen-Jen Wang, Xiang Zhu, Chengming Shi, Yanjiang Guo, Yichen Liu, and Jianyu Chen. Advancing humanoid locomotion: Mastering challenging terrains with denoising world model learning.arXiv preprint arXiv:2408.14472, 2024

    work page arXiv 2024

  9. [10]

    Karen Liu, Abder- rahmane Kheddar, Xue Bin Peng, Yuke Zhu, Guanya Shi, Quan Nguyen, Gordon Cheng, Huijun Gao, and Ye Zhao

    Zhaoyuan Gu, Junheng Li, Wenlan Shen, Wenhao Yu, Zhaoming Xie, Stephen McCrory, Xianyi Cheng, Ab- dulaziz Shamsah, Robert Griffin, C. Karen Liu, Abder- rahmane Kheddar, Xue Bin Peng, Yuke Zhu, Guanya Shi, Quan Nguyen, Gordon Cheng, Huijun Gao, and Ye Zhao. Humanoid locomotion and manipulation: Current progress and challenges in control, planning, and lear...

    work page arXiv 2025

  10. [11]

    Kungfubot2: Learn- ing versatile motion skills for humanoid whole-body control.arXiv preprint arXiv:2509.16638, 2025

    Jinrui Han, Weiji Xie, Jiakun Zheng, Jiyuan Shi, Weinan Zhang, Ting Xiao, and Chenjia Bai. Kungfubot2: Learn- ing versatile motion skills for humanoid whole-body control.arXiv preprint arXiv:2509.16638, 2025

    work page arXiv 2025

  11. [12]

    Omnih2o: Universal and dexterous human-to-humanoid whole-body teleoperation and learning.arXiv preprint arXiv:2406.08858, 2024

    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 arXiv 2024

  12. [13]

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

    Tairan He, Jiawei Gao, Wenli Xiao, Yuanhang Zhang, Zi Wang, Jiashun Wang, Zhengyi Luo, Guanqi He, Nikhil Sobanbab, Chaoyi Pan, et al. Asap: Aligning simula- tion and real-world physics for learning agile humanoid whole-body skills.arXiv preprint arXiv:2502.01143, 2025

    work page arXiv 2025

  13. [14]

    Learning-based quadruped robot framework for locomotion on dynamic rigid platforms

    Kaihui Huang, Heming Feng, Wei Meng, Tianqi Wei, and Tianjiang Hu. Learning-based quadruped robot framework for locomotion on dynamic rigid platforms. In 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 17746–17753. IEEE, 2025

    work page 2025

  14. [15]

    Learning humanoid standing-up control across diverse postures.arXiv preprint arXiv:2502.08378, 2025

    Tao Huang, Junli Ren, Huayi Wang, Zirui Wang, Qingwei Ben, Muning Wen, Xiao Chen, Jianan Li, and Jiangmiao Pang. Learning humanoid standing- up control across diverse postures.arXiv preprint arXiv:2502.08378, 2025

    work page arXiv 2025

  15. [16]

    Efficient learning of stand-up motion for humanoid robots with bilateral symmetry

    Heejin Jeong and Daniel D Lee. Efficient learning of stand-up motion for humanoid robots with bilateral symmetry. In2016 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS), pages 1544–

    work page

  16. [17]

    Mathematical model of the skateboard

    Alexander S Kuleshov. Mathematical model of the skateboard. InISBS-Conference Proceedings Archive, 2006

    work page 2006

  17. [18]

    Hold my beer: Learning gentle humanoid locomotion and end-effector stabilization control, 2025

    Yitang Li, Yuanhang Zhang, Wenli Xiao, Chaoyi Pan, Haoyang Weng, Guanqi He, Tairan He, and Guanya Shi. Hold my beer: Learning gentle humanoid locomotion and end-effector stabilization control, 2025. URL https: //arxiv.org/abs/2505.24198

    work page arXiv 2025

  18. [19]

    BeyondMimic: From Motion Tracking to Versatile Humanoid Control via Guided Diffusion

    Qiayuan Liao, Takara E Truong, Xiaoyu Huang, Guy Tevet, Koushil Sreenath, and C Karen Liu. Be- yondmimic: From motion tracking to versatile hu- manoid control via guided diffusion.arXiv preprint arXiv:2508.08241, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  19. [20]

    Discrete-time hybrid automata learning: Legged locomotion meets skateboarding.arXiv preprint arXiv:2503.01842, 2025

    Hang Liu, Sangli Teng, Ben Liu, Wei Zhang, and Maani Ghaffari. Discrete-time hybrid automata learning: Legged locomotion meets skateboarding.arXiv preprint arXiv:2503.01842, 2025

    work page arXiv 2025

  20. [21]

    Learning hu- manoid locomotion with perceptive internal model

    Junfeng Long, Junli Ren, Moji Shi, Zirui Wang, Tao Huang, Ping Luo, and Jiangmiao Pang. Learning hu- manoid locomotion with perceptive internal model. In 2025 IEEE International Conference on Robotics and Automation (ICRA), pages 9997–10003. IEEE, 2025

    work page 2025

  21. [22]

    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

    work page internal anchor Pith review Pith/arXiv arXiv 2021

  22. [23]

    Online planning and control of physics- based skateboarding animation

    Elias Mikkola. Online planning and control of physics- based skateboarding animation. 2024

    work page 2024

  23. [24]

    Isaac Lab: A GPU-Accelerated Simulation Framework for Multi-Modal Robot Learning

    Mayank Mittal, Pascal Roth, James Tigue, Antoine Richard, Octi Zhang, Peter Du, Antonio Serrano-Mu ˜noz, Xinjie Yao, Ren ´e Zurbr ¨ugg, Nikita Rudin, et al. Isaac lab: A gpu-accelerated simulation framework for multi- modal robot learning.arXiv preprint arXiv:2511.04831, 2025

    work page internal anchor Pith review Pith/arXiv arXiv 2025

  24. [25]

    Prentice hall, 2010

    Katsuhiko Ogata.Modern control engineering. Prentice hall, 2010

    work page 2010

  25. [27]

    Deepmimic: Example-guided deep re- inforcement learning of physics-based character skills

    Xue Bin Peng, Pieter Abbeel, Sergey Levine, and Michiel Van de Panne. Deepmimic: Example-guided deep re- inforcement learning of physics-based character skills. ACM Transactions On Graphics (TOG), 37(4):1–14, 2018

    work page 2018

  26. [28]

    Amp: Adversarial motion priors for stylized physics-based character control.ACM Transac- tions on Graphics (ToG), 40(4):1–20, 2021

    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

    work page 2021

  27. [29]

    Real-world hu- manoid locomotion with reinforcement learning.Science Robotics, 9(89):eadi9579, 2024

    Ilija Radosavovic, Tete Xiao, Bike Zhang, Trevor Darrell, Jitendra Malik, and Koushil Sreenath. Real-world hu- manoid locomotion with reinforcement learning.Science Robotics, 9(89):eadi9579, 2024

    work page 2024

  28. [30]

    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

  29. [31]

    Adversarial locomotion and motion imitation for humanoid policy learning.arXiv preprint arXiv:2504.14305, 2025

    Jiyuan Shi, Xinzhe Liu, Dewei Wang, Ouyang Lu, S ¨oren Schwertfeger, Chi Zhang, Fuchun Sun, Chenjia Bai, and Xuelong Li. Adversarial locomotion and motion imitation for humanoid policy learning.arXiv preprint arXiv:2504.14305, 2025

    work page arXiv 2025

  30. [32]

    Real- time skating motion control of humanoid robots for acceleration and balancing

    Noriaki Takasugi, Kunio Kojima, Shu nichi Nozawa, Yohei Kakiuchi, Kei Okada, and Masayuk i Inaba. Real- time skating motion control of humanoid robots for acceleration and balancing. In2016 IEEE/RSJ Inter- national Conference on Intell igent Robots and Systems (IROS), pages 1356–1363. IEEE, 2016

    work page 2016

  31. [33]

    Noriaki Takasugi, Kunio Kojima, Shu nichi Nozawa, Fu- mihito Sugai, Kakiuchi Yohei, Kei Okada, and Masayuki Inaba. Extended three-dimensional walking and skating motion generation for multiple noncoplanar contacts with anisotropic friction: Application to walk and skateboard and roller skate.IEEE Robotics and Automation Letters, 4(1):9–16, 2018

    work page 2018

  32. [34]

    Mujoco: A physics en- gine for model-based control, in: 2012 IEEE/RSJ International Con- ference on Intelligent Robots and Systems, IEEE

    Emanuel Todorov, Tom Erez, and Yuval Tassa. Mu- joco: A physics engine for model-based control. In 2012 IEEE/RSJ International Conference on Intelligent Robots and Systems, pages 5026–5033. IEEE, 2012. doi: 10.1109/IROS.2012.6386109

    work page doi:10.1109/iros.2012.6386109 2012

  33. [35]

    Humanoid robot G1 Humanoid Robot Functions Humanoid Robot Price — Unitree Robotics,

    Unitree Robotics. Humanoid robot G1 Humanoid Robot Functions Humanoid Robot Price — Unitree Robotics,

    work page

  34. [36]

    https://www

    URL https://www.unitree.com/g1/. https://www. unitree.com/g1/

    work page

  35. [37]

    Skateboard: a human controlled non-holonomic syste m

    Balazs Varszegi, Denes Takacs, and Gabo r Stepan. Skateboard: a human controlled non-holonomic syste m. InInternational Design Engineering Technical Confe rences and Computers and Information in Engineering Conferenc e, volume 57168, page V006T10A066. Amer- ican Society of Mechanical Engineers, 2015

    work page 2015

  36. [38]

    More: Mixture of residual experts for humanoid life- like gaits learning on complex terrains.arXiv preprint arXiv:2506.08840, 2025

    Dewei Wang, Xinmiao Wang, Xinzhe Liu, Jiyuan Shi, Yingnan Zhao, Chenjia Bai, and Xuelong Li. More: Mixture of residual experts for humanoid life- like gaits learning on complex terrains.arXiv preprint arXiv:2506.08840, 2025

    work page arXiv 2025

  37. [39]

    Physhsi: Towards a real-world generalizable and natural humanoid-scene interaction system.arXiv preprint arXiv:2510.11072, 2025

    Huayi Wang, Wentao Zhang, Runyi Yu, Tao Huang, Junli Ren, Feiyu Jia, Zirui Wang, Xiaojie Niu, Xiao Chen, Jiahe Chen, et al. Physhsi: Towards a real-world gener- alizable and natural humanoid-scene interaction system. arXiv preprint arXiv:2510.11072, 2025

    work page arXiv 2025

  38. [40]

    Hdmi: Learning interactive humanoid whole-body control from human videos.arXiv preprint arXiv:2509.16757, 2025

    Haoyang Weng, Yitang Li, Nikhil Sobanbabu, Zihan Wang, Zhengyi Luo, Tairan He, Deva Ramanan, and Guanya Shi. Hdmi: Learning interactive humanoid whole-body control from human videos.arXiv preprint arXiv:2509.16757, 2025

    work page arXiv 2025

  39. [41]

    Humanoid whole-body locomotion on narrow terrain via dynamic balance and reinforcement learning.arXiv preprint arXiv:2502.17219, 2025

    Weiji Xie, Chenjia Bai, Jiyuan Shi, Junkai Yang, Yun- fei Ge, Weinan Zhang, and Xuelong Li. Humanoid whole-body locomotion on narrow terrain via dynamic balance and reinforcement learning.arXiv preprint arXiv:2502.17219, 2025

    work page arXiv 2025

  40. [42]

    Kungfubot: Physics-based humanoid whole- body control for learning highly-dynamic skills.arXiv preprint arXiv:2506.12851, 2025

    Weiji Xie, Jinrui Han, Jiakun Zheng, Huanyu Li, Xinzhe Liu, Jiyuan Shi, Weinan Zhang, Chenjia Bai, and Xue- long Li. Kungfubot: Physics-based humanoid whole- body control for learning highly-dynamic skills.arXiv preprint arXiv:2506.12851, 2025

    work page arXiv 2025

  41. [43]

    Optimization based dynamic skateboarding of quadrupedal robot

    Zhe Xu, Mohamed Al-Khulaqui, Hanxin Ma, Jiajun Wang, Quanbin Xin, Yangwei You, Mingliang Zhou, Diyun Xiang, and Shiwu Zhang. Optimization based dynamic skateboarding of quadrupedal robot. In2024 IEEE International Conference on Robotics and Automa- tion (ICRA), pages 8058–8064. IEEE, 2024

    work page 2024

  42. [44]

    Omniretarget: Interaction-preserving data gen- eration for humanoid whole-body loco-manipulation and scene interaction.arXiv preprint arXiv:2509.26633, 2025

    Lujie Yang, Xiaoyu Huang, Zhen Wu, Angjoo Kanazawa, Pieter Abbeel, Carmelo Sferrazza, C Karen Liu, Rocky Duan, and Guanya Shi. Omniretarget: Interaction- preserving data generation for humanoid whole-body loco-manipulation and scene interaction.arXiv preprint arXiv:2509.26633, 2025

    work page arXiv 2025

  43. [45]

    An overview on vehicle dynamics.International Journal of Dynamics and Control, 1(4):385–395, 2013

    Shaopu Yang, Yongjie Lu, and Shaohua Li. An overview on vehicle dynamics.International Journal of Dynamics and Control, 1(4):385–395, 2013

    work page 2013

  44. [46]

    mjlab: A lightweight framework for gpu-accelerated robot learn- ing, 2026

    Kevin Zakka, Qiayuan Liao, Brent Yi, Louis Le Lay, Koushil Sreenath, and Pieter Abbeel. mjlab: A lightweight framework for gpu-accelerated robot learn- ing, 2026. URL https://arxiv.org/abs/2601.22074

    work page arXiv 2026

  45. [47]

    Twist: Teleoperated whole-body imitation system.arXiv preprint arXiv:2505.02833, 2025

    Yanjie Ze, Zixuan Chen, Jo ˜ao Pedro Ara ´ujo, Zi ang Cao, Xue Bin Peng, Jiajun Wu, and C. Karen Liu. Twist: Teleoperated whole-body imitation system.arXiv preprint arXiv:2505.02833, 2025

    work page arXiv 2025

  46. [48]

    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, 2025. URL https://arxiv.org/abs/2511.02832

    work page arXiv 2025

  47. [50]

    Wococo: Learning whole-body humanoid control with sequential contacts.arXiv preprint arXiv:2406.06005, 2024

    Chong Zhang, Wenli Xiao, Tairan He, and Guanya Shi. Wococo: Learning whole-body humanoid control with sequential contacts, 2024. URL https://arxiv.org/abs/ 2406.06005

    work page arXiv 2024

  48. [51]

    Falcon: Learning force-adaptive humanoid loco-manipulation.arXiv preprint arXiv:2505.06776, 2025

    Yuanhang Zhang, Yifu Yuan, Prajwal Gurunath, Ishita Gupta, Shayegan Omidshafiei, Ali-akbar Agha-mohammadi, Marcell Vazquez-Chanlatte, Liam Pedersen, Tairan He, and Guanya Shi. Falcon: Learning force-adaptive humanoid loco-manipulation.arXiv preprint arXiv:2505.06776, 2025

    work page arXiv 2025

  49. [52]

    Humanoid parkour learning.arXiv preprint arXiv:2406.10759, 2024

    Ziwen Zhuang, Shenzhe Yao, and Hang Zhao. Humanoid parkour learning.arXiv preprint arXiv:2406.10759, 2024. APPENDIX A. Derivation of Skateboard Equality Constraint This appendix derives the kinematic relationship between the deck tilt angleγand the wheel axle steering angleσfor a skateboard truck with fixed rake angleλ. The derivation follows a two-stage ...

    work page arXiv 2024