Robust Autonomous Manipulation of Articulated Objects

Abstract

This research aims to develop a unified framework for robust autonomous manipulation of articulated objects, like doors and switches, under uncertain environmental conditions. We integrate perception, modeling, planning, and control to enhance manipulation capabilities. Our approach focuses on providing high-resolution, interaction-rich scene information, incorporating modeling uncertainties, and creating adaptive real- time interaction plans. This integrated framework will enable autonomous agents to perform complex tasks in real-world scenarios, advancing the field of autonomous manipulation in robotics.