arXiv 2409.204732026-05-25cs.RO

数据驱动的触觉传感器配置优化以实现高效灵巧操作

Data-Driven Optimization of Tactile Sensor Configurations for Efficient Dexterous Manipulation

ShanghaiTech University, School of Information Science and Technology（上海科技大学信息科学与技术学院）；University of Alberta（阿尔伯塔大学）；Oklahoma State University（俄克拉荷马州立大学）；University of Colorado Denver, Department of Computer Science and Engineering（科罗拉多大学丹佛分校计算机科学与工程系）；Department of Robotics and Mechatronics Engineering, Kennesaw State University（凯斯西储大学机器人与机电工程系）

Haoran Guo, Haoyang Wang, Zhengxiong Li, He Bai, Lingfeng Tao

备注：: This work has been submitted to the ICRA for possible publication

阅读 PDF

arXiv 页面 ↗HTML 原文 ↗

AI总结：

提出两阶段框架量化触觉传感器对深度强化学习策略的贡献，将Shadow Hand传感器从92个减少至14个仍保持90%以上性能，并发现中指传感器具有负贡献。

摘要：

触觉感知对于基于学习的灵巧操作至关重要，但传感器放置的原则性指导仍然缺乏。虽然密集传感器阵列提供丰富的接触反馈，但它们带来显著的硬件成本，甚至可能通过引入冗余或冲突输入而降低策略性能。本文提出了第一个系统框架，用于量化单个触觉传感器对深度强化学习（DRL）策略性能的贡献。我们提出了一种两阶段方法：粗粒度经验剪枝阶段将Shadow Hand上的传感器数量从92个减少到21个，同时保留93%的任务性能；随后是细粒度主动学习阶段，结合高斯过程回归（GPR）与Lasso回归对每个剩余传感器的功能重要性进行排序。我们的分析揭示，拇指、无名指和小指上的传感器主导操作性能，而中指传感器表现出负贡献——主动降低策略学习。跨三个操作任务（方块、鸡蛋和笔）的消融研究证实，14个传感器的配置保留了全阵列90%以上的性能。在两个新物体上的零样本迁移实验以及在Allegro和Leap Hand上的跨平台验证进一步表明，识别出的重要性排序在任务和机器人形态之间具有泛化性。这些发现建立了量化部署指南，使从业者能够选择具有可预测性能权衡的成本效益传感器配置。

英文摘要：

Tactile sensing is critical for learning-based dexterous manipulation, yet principled guidelines for sensor placement remain largely absent. While dense sensor arrays provide rich contact feedback, they impose significant hardware costs and can even degrade policy performance by introducing redundant or conflicting inputs. This paper presents the first systematic framework for quantifying the contribution of individual tactile sensors to deep reinforcement learning (DRL) policy performance. We propose a two-stage approach: a coarse empirical pruning phase that reduces the sensor count on the Shadow Hand from 92 to 21 while retaining 93\% task performance, followed by a fine-grained active learning phase that combines Gaussian Process Regression (GPR) with Lasso regression to rank the functional importance of each remaining sensor. Our analysis reveals that sensors on the thumb, ring finger, and little finger dominate manipulation performance, while middle-finger sensors exhibit negative contributions -- actively degrading policy learning. Ablation studies across three manipulation tasks (block, egg, and pen) confirm that a 14-sensor configuration preserves over 90\% of the full-array performance. Zero-shot transfer experiments on two novel objects and cross-platform validation on the Allegro and Leap Hand further demonstrate that the identified importance rankings generalize across tasks and robot morphologies. These findings establish quantitative deployment guidelines that enable practitioners to select cost-effective sensor configurations with predictable performance trade-offs.

机构、总结、中文摘要均由AI分析而来，可能存在错误，请以论文正文为准。