超越模仿：从人际示范中学习全身人-人形机器人交互
Beyond Mimicry: Learning Whole-Body Human-Humanoid Interaction from Human-Human Demonstrations
PAIR 的接触中心两阶段重定向思路可迁移到任意跨形态动作迁移（如人手到机械手、人体到虚拟人），其核心是显式保留接触语义而非只对齐关节轨迹；D-STAR 的「何时动/何处动」解耦加扩散检测头结构，为视觉-动作联合建模提供了一种可复用的分层策略设计。
原文依据 PAIR (Physics-Aware Interaction Retargeting), a contact-centric, two-stage pipeline that preserves contact semantics across morphology differences
摘要译文
由模型翻译 · 术语统一 · 请对照原文核对关键结论使人形机器人能够与人类进行物理交互是一个关键前沿，但高质量的人-人形机器人交互（HHoI）数据稀缺阻碍了进展。利用丰富的人际交互（HHI）数据是一种可扩展的替代方案，但我们首先证明标准的重定向方法会因破坏关键接触而失效。为此我们提出 PAIR（物理感知交互重定向），一个以接触为中心的两阶段流程，在形态差异之间保留接触语义，从而生成物理一致的 HHoI 数据。然而，这一高质量数据暴露出第二个问题：常规模仿学习策略仅模仿轨迹，缺乏交互理解。因此我们引入 D-STAR（解耦时空动作推理器），一种将「何时行动」与「何处行动」解耦的分层策略。在 D-STAR 中，相位注意力（何时）与多尺度空间模块（何处）由扩散检测头融合，产生超越模仿的同步全身行为。通过解耦这些推理流，模型能够学习鲁棒的时序相位而不被空间噪声干扰，从而实现响应迅速、同步的协作。我们通过大量严谨的仿真验证了该框架，相比基线方法取得显著性能提升，并给出了一条从 HHI 数据学习复杂全身交互的完整有效流程。
英文原文摘要 1457 字符
Enabling humanoid robots to physically interact with humans is a critical frontier, but progress is hindered by the scarcity of high-quality Human-Humanoid Interaction (HHoI) data. While leveraging abundant Human-Human Interaction (HHI) data presents a scalable alternative, we first demonstrate that standard retargeting fails by breaking the essential contacts. We address this with PAIR (Physics-Aware Interaction Retargeting), a contact-centric, two-stage pipeline that preserves contact semantics across morphology differences to generate physically consistent HHoI data.This high-quality data, however, exposes a second failure: conventional imitation learning policies merely mimic trajectories and lack interactive understanding. We therefore introduce D-STAR (Decoupled Spatio-Temporal Action Reasoner), a hierarchical policy that disentangles when to act from where to act. In D-STAR, Phase Attention (when) and a Multi-Scale Spatial module (where) are fused by the diffusion head to produce synchronized whole-body behaviors beyond mimicry.By decoupling these reasoning streams, our model learns robust temporal phases without being distracted by spatial noise, leading to responsive, synchronized collaboration. We validate our framework through extensive and rigorous simulations, demonstrating significant performance gains over baseline approaches and a complete, effective pipeline for learning complex whole-body interactions from HHI data.