Learning Strategies for Erecting Horizontal Objects via Half-Grasping to Aid Subsequent Tasks

In robotic assembly environments, the difficulty of subsequent tasks depends on how and where the target object is grasped, such as in peg-in-hole or bolting/screwing scenarios. Therefore, adjusting the pose of the target object is often necessary before achieving the desired configuration. Conventional wisdom dictates a grasping position distant from the center of mass of an object, resulting in the rotational displacement near the grasp point, as a failure. However, in this study, we redefine such scenarios as "half-grasping" and intentionally leverage them to execute the task of erecting horizontally lying objects. In this research, we introduce the concept of uprighting objects using half-grasping. We implemented this approach in the virtual environment using NVIDIA Isaac Sim and trained manipulation strategies through deep reinforcement learning. Using the trained policy, an object erection task was performed in a virtual environment, achieving a success rate of over 98.2% thus confirming the feasibility of the proposal.

1. INTRODUCTION

In robotic assembly environments, the difficulty in performing subsequent tasks can vary depending on how and where the target object is grasped. However, objects are not always conveniently positioned for robotic manipulation. Therefore, poses often require adjustment before an object can be grasped in the desired configuration. To address these challenges, researchers have developed task-oriented grasping methods, to reorient objects to their nominal poses using in-hand manipulation [1], [2].