# DexMimic：价值引导重置分配实现通用灵巧手物体跟踪策略

> 原文链接：https://robotics.hellobike.com/papers/2026-10-15-2610-19089
> 频道：papers | 更新日期：2026-10-15 | 标签：auto, arxiv
> 核心摘要：针对灵巧操作技能复用难题，提出DexMimic通用策略与价值引导重置分配机制。实验验证了训练效率提升、零样本泛化能力及真机部署有效性，并开源涵盖训练、微调与泛化的全链路基准测试集。

# DexMimic：价值引导重置分配实现通用灵巧手物体跟踪策略

::: v-pre
> 本文由 AI 根据公开内容整理，可能有误，请以原文为准：[DexMimic: A Generalist Dexterous Hand-Object Tracking Policy via Value-Guided Reset Allocation](https://arxiv.org/abs/2610.19089)（arXiv 2610.19089）

## 核心内容

针对现有基于物理的灵巧操作方法仅能训练单一轨迹专家策略、难以规模化复用技能的痛点，研究团队提出DexMimic通用灵巧手-物体跟踪策略。该方法通过联合训练多样化人手-物体交互（HOI）轨迹，解决了多任务学习中的初始化分布难题。论文不仅发布了包含训练、微调和泛化测试的完整Benchmark，还在真实世界灵巧操作平台上完成了算法验证，证实了其在未见轨迹与物体上的零样本泛化潜力及大数据集下的性能增益。

核心技术采用价值引导重置分配（Value-Guided Reset Allocation）机制，在线自适应调整初始化分布，优先采样欠训练阶段并选择高预期回报的重置状态，从而显著提升联合训练效率与跟踪成功率。相比传统方法，该架构支持从大规模HOI数据集中高效提取通用特征，实现了更快的下游任务微调速度。系统已在实体机器人平台完成部署验证，证明了仿真训练策略向真实硬件迁移的有效性，为具身智能灵巧操作提供了可复现的工程范式。

## 关键看点

- **核心算法**：提出Value-Guided Reset Allocation机制，在线优化重置状态分布以聚焦欠训练阶段，提升多轨迹联合训练收敛效率
- **泛化能力**：验证了对未见轨迹与新物体的零样本泛化潜力，且扩大训练数据集可进一步提升模型效率与泛化上限
- **工程验证**：在真实世界灵巧操作平台完成部署验证，确认了仿真策略到实体硬件的有效迁移与实际操作性能
- **开源基准**：发布涵盖训练、微调及泛化测试的全链路Benchmark，填补了通用灵巧手策略标准化评测工具的空白

## 工程与产业观察

DexMimic通过价值引导重置机制突破了灵巧操作多任务学习的瓶颈，其真机验证与开源Benchmark为具身智能提供了关键基础设施。该工作表明，构建高质量HOI数据集与高效的课程学习策略是实现通用灵巧操作的核心路径，对解决Sim-to-Real迁移难题具有重要工程参考价值。

## 英文摘要 (Abstract)

Achieving human-like dexterous manipulation remains a central challenge in robotics. Learning from human hand-object interaction (HOI) datasets offers a promising pathway to diverse dexterous skills, yet existing physics-based approaches train specialist policies for individual trajectories, thereby limiting skill reuse at scale. We introduce DexMimic, a generalist dexterous hand-object tracking policy jointly trained across diverse HOI trajectories. Efficient joint training requires allocating each rollout to both an undertrained phase and a starting state that supports continued tracking. DexMimic realizes this insight through Value-Guided Reset Allocation, which adapts the initialization distribution online to prioritize difficult phases and select reset states with higher expected tracking returns. We further introduce a benchmark spanning training, fine-tuning, and generalization. Experiments show improved tracking success and training efficiency, faster fine-tuning, and the potential for zero-shot generalization to unseen trajectories and objects; larger training sets further improve efficiency and generalization. Finally, we validate DexMimic on a real-world dexterous manipulation platform.

作者：Zhexi Luo, Zhixuan Xu, Rui Yang, Zeyu Xiao, Yiwen Hou, Lin Shao
:::

## 原始参考来源 (Sources)

- [arXiv 2610.19089：DexMimic: A Generalist Dexterous Hand-Object Tracking Policy via Value-Guided Reset Allocation](https://arxiv.org/abs/2610.19089)