Stiffness estimation and finger-object impact detection with a robotic gripper using intrinsic sensors

Abstract

Robotics applications grow impressively year after year. While robots are subject to considerable technological and behavioral improvements, robotic grippers, often acting as end-effectors in a vast range of industrial scenarios, still suffer from limited performance. In this article, specific algorithms that may improve the performance of robotic grippers are proposed. The algorithms allow the implementation of functions that are not available on commercial end-effectors, namely: (1) to safely grip items regardless the size and hardness at high finger velocity (up to 16.5 mm/s) and (2) to estimate stiffness in a wide range (up to ca. 92 N/mm). As a key advantage, the above functions were implemented resorting to the sole intrinsic sensors, i.e. sensors included in the gripper itself as per construction. Moreover, no mechanical adjustments, such as dampening elements or flexible fingers, are required. Experiments on two different setups, involving bench trials and validation on a fully integrated robotic platform, demonstrate a clear enhancement of gripper performance when exploiting the proposed algorithms.