# Spectral Skills：宇树G1人形机器人运动跟踪误差降低62%

> 原文链接：https://robotics.hellobike.com/papers/2026-09-29-2609-37677
> 频道：papers | 更新日期：2026-09-29 | 标签：auto, arxiv
> 核心摘要：研究提出Spectral Skills潜在表征，通过预测式学习紧凑编码短运动片段。在29自由度宇树G1上验证，控制器全局跟踪误差较SOTA降低62%，支持技能链式组合与正交叠加生成新行为，并实现语言规划器硬件部署。

# Spectral Skills：宇树G1人形机器人运动跟踪误差降低62%

::: v-pre
> 本文由 AI 根据公开内容整理，可能有误，请以原文为准：[Learning Expressive and Compositional Motion Representation via Spectral Skills](https://arxiv.org/abs/2609.37677)（arXiv 2609.37677）

## 核心内容

针对人形机器人基础模型分层架构中规划器与控制器接口难以兼顾执行精度与可预测性的痛点，研究提出Spectral Skills潜在运动表征。该方法通过预测后续运动而非重构输入进行表征学习，在29自由度宇树G1人形机器人实机验证中，使条件控制器的全局跟踪误差相比当前最优方法降低62%，显著提升了底层运动执行的保真度。

Spectral Skills紧凑编码短运动片段，允许冻结控制器直接链接独立编码的技能而无需额外过渡策略，并通过向兼容基础技能添加正交方向来组合训练集中未见的新行为。系统已在Unitree G1硬件上完成轨迹跟踪、技能链接、行为组合及语言条件规划器的全流程演示，验证了该表征在具身智能分层控制中的通用性与零样本泛化能力。

## 关键看点

- **跟踪性能**：在29-DoF宇树G1上，基于Spectral Skills的控制器全局跟踪误差较SOTA降低62%
- **技能组合**：支持向基础技能添加正交方向以生成训练集未见的新行为，无需重训模型
- **无缝链接**：冻结控制器可直接链式执行独立编码技能，无需设计单独的过渡策略
- **系统验证**：在Unitree G1硬件完成语言规划器驱动下的跟踪、链接与组合全栈实测

## 工程与产业观察

Spectral Skills为人形机器人分层控制提供了高信息密度的中间层接口范式。其预测式表征学习解决了传统自编码器重构目标与下游控制任务不对齐的问题，62%的误差降幅证明了潜在空间结构对物理执行精度的决定性影响。正交组合机制为具身大模型在不增加参数量的前提下扩展技能库提供了可复用的工程路径。

## 英文摘要 (Abstract)

Robotic foundation models offer a promising path toward general-purpose humanoid robot control, often through hierarchical architectures. However, their effectiveness depends on the command interface between the planner and the controller, which must support accurate execution while remaining easy to predict, and ideally allow new behaviors to be composed from prior ones. In this work, we introduce spectral skills, a latent representation of this interface that meets these requirements through predictive representation learning. By design, spectral skills compactly encode short motion segments and are learned by predicting subsequent motion rather than reconstructing the encoder input. On a 29-DoF humanoid, a controller conditioned on spectral skills reduces global tracking error by 62\% relative to the state of the art. The same frozen controller chains independently encoded skills without a separate transition policy. It also composes new behaviors by adding orthogonal directions to any compatible base skill, producing combinations unseen in the training data. We demonstrate tracking, chaining, and composition, as well as control through a language-conditioned planner, on Unitree G1 hardware. Project page: https://spectral-skill.github.io

作者：Feiyang Wu, Chenxiao Gao, Chen Yang, Ye Zhao, Bo Dai, Anqi Wu
:::

## 原始参考来源 (Sources)

- [arXiv 2609.37677：Learning Expressive and Compositional Motion Representation via Spectral Skills](https://arxiv.org/abs/2609.37677)