Detection of slip during object grasping and manipulation plays a vital role in object handling. Existing solutions largely depend on visual information to devise a strategy for grasping. Nonetheless, in order to achieve proficiency akin to humans and achieve consistent grasping and manipulation of unfamiliar objects, the incorporation of artificial tactile sensing has become a necessity in robotic systems. In this work, we propose a novel physics-informed, data-driven method to detect slip continuously in real time. The GelSight Mini, an optical tactile sensor, is mounted on custom grippers to acquire tactile readings. Our work leverages the inhomogeneity of tactile sensor readings during slip events to develop distinctive features and formulates slip detection as a classification problem. To evaluate our approach, we test multiple data-driven models on 10 common objects under different loading conditions, textures, and materials. Our results show that the best classification algorithm achieves an average accuracy of 99\%. We demonstrate the application of this work in a dynamic robotic manipulation task in which real-time slip detection and prevention algorithm is implemented.

翻译：在物体抓取与操作过程中，滑移检测对物体操控起着至关重要的作用。现有解决方案主要依赖视觉信息来制定抓取策略。然而，要实现类人水平的灵巧操作并持续可靠地抓取不熟悉物体，人工触觉传感已成为机器人系统的必备功能。本文提出了一种基于物理信息的数据驱动方法，用于实时连续检测滑移。我们将光学触觉传感器GelSight Mini安装在定制夹爪上采集触觉数据。本研究利用滑移事件中触觉传感器读数的非均匀性构建差异化特征，并将滑移检测转化为分类问题。为评估该方法，我们在10种常见物体上测试了多种数据驱动模型，涵盖不同负载条件、纹理和材料。结果表明，最优分类算法的平均准确率达到99%。我们还展示了该方法在动态机器人操作任务中的应用，实现了实时滑移检测与预防算法。