Vision-Based Flexure Gripper Configuration Estimation for Visuo-Tactile Sensing

Benelux meeting March 2025

Marissa Jonker, Jan de Jong, Wesley Roozing and Wouter Hakvoort

Introduction

As the need for automation in agriculture is rising, the demand for suitable grippers to grasp delicate crops grows. Sensing the grasping forces is needed to carefully manipulate crops. However, sensing without contact sensors is desired to upkeep cleanability, thus we aim to do this remotely via vision. For this, we need to know the configuration of the gripper and contact points first. Current research on visually obtaining mechanism configuration do not account for occlusions, or do so by adding an additional camera [2]. We propose a method to visually estimate the configuration of an underactuated gripper, while accounting for occlusions due to e.g. leaves, by tracking a redundant set of markers on the gripper.

Open the full PDF