Logo
ScienceStack
Table of Contents
Fetching ...
Sign In

Retargeting Matters: General Motion Retargeting for Humanoid Motion Tracking

Joao Pedro Araujo, Yanjie Ze, Pei Xu, Jiajun Wu, C. Karen Liu

TL;DR

The paper addresses the embodiment gap in humanoid learning by evaluating how retargeting quality affects motion-tracking policies trained with BeyondMimic. It introduces General Motion Retargeting (GMR), a non-uniform scaling and two-stage IK method that reduces artifacts like foot sliding and ground penetration, and compares it to PHC, ProtoMotions, and Unitree retargets. Using a diverse LAFAN1 subset, it shows that retargeting artifacts can significantly impact policy robustness, with GMR delivering near-Unitree fidelity and higher policy success than open-source methods. The work highlights the practical importance of retargeting quality for reliable, reward-light policy learning and outlines directions for broader datasets and cross-robot applicability.

Abstract

Humanoid motion tracking policies are central to building teleoperation pipelines and hierarchical controllers, yet they face a fundamental challenge: the embodiment gap between humans and humanoid robots. Current approaches address this gap by retargeting human motion data to humanoid embodiments and then training reinforcement learning (RL) policies to imitate these reference trajectories. However, artifacts introduced during retargeting, such as foot sliding, self-penetration, and physically infeasible motion are often left in the reference trajectories for the RL policy to correct. While prior work has demonstrated motion tracking abilities, they often require extensive reward engineering and domain randomization to succeed. In this paper, we systematically evaluate how retargeting quality affects policy performance when excessive reward tuning is suppressed. To address issues that we identify with existing retargeting methods, we propose a new retargeting method, General Motion Retargeting (GMR). We evaluate GMR alongside two open-source retargeters, PHC and ProtoMotions, as well as with a high-quality closed-source dataset from Unitree. Using BeyondMimic for policy training, we isolate retargeting effects without reward tuning. Our experiments on a diverse subset of the LAFAN1 dataset reveal that while most motions can be tracked, artifacts in retargeted data significantly reduce policy robustness, particularly for dynamic or long sequences. GMR consistently outperforms existing open-source methods in both tracking performance and faithfulness to the source motion, achieving perceptual fidelity and policy success rates close to the closed-source baseline. Website: https://jaraujo98.github.io/retargeting_matters. Code: https://github.com/YanjieZe/GMR.

Retargeting Matters: General Motion Retargeting for Humanoid Motion Tracking

TL;DR

The paper addresses the embodiment gap in humanoid learning by evaluating how retargeting quality affects motion-tracking policies trained with BeyondMimic. It introduces General Motion Retargeting (GMR), a non-uniform scaling and two-stage IK method that reduces artifacts like foot sliding and ground penetration, and compares it to PHC, ProtoMotions, and Unitree retargets. Using a diverse LAFAN1 subset, it shows that retargeting artifacts can significantly impact policy robustness, with GMR delivering near-Unitree fidelity and higher policy success than open-source methods. The work highlights the practical importance of retargeting quality for reliable, reward-light policy learning and outlines directions for broader datasets and cross-robot applicability.

Abstract

Humanoid motion tracking policies are central to building teleoperation pipelines and hierarchical controllers, yet they face a fundamental challenge: the embodiment gap between humans and humanoid robots. Current approaches address this gap by retargeting human motion data to humanoid embodiments and then training reinforcement learning (RL) policies to imitate these reference trajectories. However, artifacts introduced during retargeting, such as foot sliding, self-penetration, and physically infeasible motion are often left in the reference trajectories for the RL policy to correct. While prior work has demonstrated motion tracking abilities, they often require extensive reward engineering and domain randomization to succeed. In this paper, we systematically evaluate how retargeting quality affects policy performance when excessive reward tuning is suppressed. To address issues that we identify with existing retargeting methods, we propose a new retargeting method, General Motion Retargeting (GMR). We evaluate GMR alongside two open-source retargeters, PHC and ProtoMotions, as well as with a high-quality closed-source dataset from Unitree. Using BeyondMimic for policy training, we isolate retargeting effects without reward tuning. Our experiments on a diverse subset of the LAFAN1 dataset reveal that while most motions can be tracked, artifacts in retargeted data significantly reduce policy robustness, particularly for dynamic or long sequences. GMR consistently outperforms existing open-source methods in both tracking performance and faithfulness to the source motion, achieving perceptual fidelity and policy success rates close to the closed-source baseline. Website: https://jaraujo98.github.io/retargeting_matters. Code: https://github.com/YanjieZe/GMR.

Paper Structure

This paper contains 14 sections, 6 equations, 4 figures, 3 tables.

Table of Contents

  1. Introduction
  2. Related Work
  3. Evaluation Method
  4. Retargeting Methods
  5. Data Processing
  6. Metrics
  7. User Study Evaluation
  8. General Motion Retargeting
  9. Evaluation Results
  10. Impact of retargeting method in policy performance
  11. Impact of retargeting artifacts
  12. User study
  13. First reference frame
  14. Conclusion

Figures (4)

  • Figure 1: For the user study, participants were shown videos of the reference motion (a), and asked to choose which retarget video (b) was more similar to it.
  • Figure 2: General Motion Retargeting (GMR) Pipeline.
  • Figure 3: Example artifacts found in the retargeted references with low success rates.
  • Figure 4: User study ($N=20$) results for comparing GMR to other retargets in terms of faithfulness to the source motion. The bars represent the percentage of responses.