The article introduces a new approach that combines tactile estimation and control for in-hand object manipulation. This method integrates measurements from robot kinematics and an image-based tactile sensor to estimate and control the pose of a grasped object. The approach uses a discrete pose estimator and a continuous pose estimator-controller to manipulate the pose of the grasped object accurately. The method has been tested on diverse objects and configurations, and it has been successful in achieving manipulation objectives. This approach can be beneficial for tasks requiring precise manipulation in scenarios where visual perception is limited.
Publication date: 19 Jan 2024
Project Page: https://sites.google.com/view/texterity
Paper: https://arxiv.org/pdf/2401.10230
