Pushing Dexterity and Safety for In-Hand Object Pose Reaching That Works Remarkably Well

📄 arXiv: 2610.37652v1 📥 PDF

作者: Ian Chuang, Hiya Shah, Vamsi Chilamkur, Iman Soltani, Claire Tomlin

分类: cs.RO, cs.LG, eess.SY

发布日期: 2026-10-27

备注: 26 pages, 30 figures, 14 tables. Project page: https://ian-chuang.github.io/dexsafe/

🔗 代码/项目: PROJECT_PAGE


💡 一句话要点

提出一种新方法以提升多指机器人手的灵巧性与安全性

🎯 匹配领域: 支柱一:机器人控制 (Robot Control)

关键词: 多指机器人手 灵巧操作 安全过滤器 物体姿态到达 自适应摩擦课程 零样本迁移 抓取数据集

📋 核心要点

  1. 现有的多指机器人手在执行复杂的抓取任务时,往往面临安全性不足的问题,导致物体掉落现象频繁。
  2. 本文提出了一种新的训练策略,通过多样的重置和目标生成抓取数据集,并结合自适应摩擦课程,提升了机器人手的灵巧性和安全性。
  3. 在真实机器人评估中,策略在4800个未见目标上达成了95.2%的目标,且安全过滤器将物体掉落率降低了75%。

📝 摘要(中文)

多指机器人手的灵巧操作是机器人技术的下一个前沿,但现有的学习手部技能仅限于简单的抓取或基本的重复动作,如旋转物体。本文研究了在手内物体姿态到达的任务,要求实现多样且复杂的抓取。我们提出了一种在模拟环境中训练LEAP手的策略,利用生成的抓取数据集和自适应摩擦课程等多种见解。通过精确的系统识别,这些策略能够零样本迁移到真实手上,并表现出卓越的能力和鲁棒性。我们还增加了一个可调节保守性的学习残差安全过滤器,显著减少了物体掉落的情况。

🔬 方法详解

问题定义:本文旨在解决多指机器人手在执行复杂的抓取任务时,安全性不足和物体掉落的问题。现有方法通常仅能处理简单的抓取或重复动作,难以应对多样化的抓取需求。

核心思路:论文提出了一种在模拟环境中训练多指手的策略,利用生成的抓取数据集和自适应摩擦课程,使得机器人能够在多种目标和手腕姿态下灵活操作。

技术框架:整体架构包括数据生成、策略训练和安全过滤三个主要模块。首先,通过生成抓取数据集实现多样的重置和目标;其次,训练策略以实现高效的物体姿态到达;最后,应用安全过滤器以减少物体掉落。

关键创新:最重要的创新在于提出了一种无须人类数据的训练方法,能够实现策略的零样本迁移,并且引入了可调节的安全过滤器,显著提升了操作的安全性。

关键设计:在训练过程中,采用了多样的目标设置和自适应摩擦课程,确保策略在不同条件下的鲁棒性。同时,安全过滤器的设计允许根据需求调整保守性,以平衡性能与安全性。

🖼️ 关键图片

fig_0
fig_1
fig_2

📊 实验亮点

在真实机器人评估中,策略在4800个未见目标上达成了95.2%的目标,且安全过滤器将物体掉落率降低了75%。此外,机器人能够在任意姿态下旋转灯泡,并在快速运动和扰动下保持稳定,展现出极强的适应能力。

🎯 应用场景

该研究的潜在应用领域包括服务机器人、工业自动化和家庭助理等场景,能够提升机器人在复杂环境中的操作能力和安全性。未来,该方法有望推动机器人技术在更广泛领域的应用,尤其是在需要高灵活性和安全性的任务中。

📄 摘要(原文)

Dexterous manipulation with multi-fingered hands is the next frontier of robotics, yet learned hand skills remain limited to simple grasps or basic repeated motions, such as rotating an object. As robots attempt more demanding in-hand maneuvers, safety, such as not dropping the object, becomes critical, yet it is rarely measured. We study in-hand object pose reaching, a task that demands diverse and difficult grasps: moving an object to almost any position and orientation within reach of the fingers, at any wrist orientation, through long sequences of consecutive goals. We present a recipe for training such policies for a LEAP hand in simulation without human data, built on a range of insights, from diverse resets and goals via generated grasp datasets to an adaptive friction curriculum. With careful system identification, the policies transfer zero-shot to the real hand, using either motion capture or a single wrist-mounted depth camera, and prove remarkably capable and robust. On top of the task policy, we add a learned residual safety filter with adjustable conservativeness, letting us tune the balance between performance and safety and greatly reduce object drops. In 13.3 hours of real-robot evaluation over 4,800 unseen goals across 4 objects and diverse wrist orientations, the state-based policy reaches 95.2% of goals within 2 cm and 0.4 rad, and the filter cuts its drops by 75% while still reaching 91.5% of goals. The policies can also spin a lightbulb held in an arbitrary pose, reach out-of-distribution goals, withstand fast arm motion and perturbations, and, needing only a wrist camera, serve as portable skills such as screwing in a lightbulb. We release the full recipe as a platform for studying performance, reliability, and safety in high-dimensional dexterous systems. Videos and more are available at https://ian-chuang.github.io/dexsafe/.