Hand Posture Generation for Object Manipulation Using Reinforcement Learning with Physics Simulation

View at publisher Summarise with VIRAT
Authors: Ikuma Kuribayashi, Taishi Iriyama, Takashi Komuro
Cited by: 0
References: 11

Summary

No summary yet. VIRAT can read this paper and write a plain-language summary of its aim, methods, key findings and limitations.

Summarise with VIRAT Free to try with your Science Catalogs account.

Abstract

In this study, we propose a method for generating hand postures that match the shape of virtual objects in systems where users manipulate virtual objects using physical objects. In order to reflect the user's real-time manipulation of physical objects, the proposed method uses the hand postures as well as the positions and orientations of physical objects as input. Based on this input, the proposed method generates hand postures using physics simulation and reinforcement learning. This enables the generation of hand postures that match the shape of virtual objects while maintaining physical balance and reflecting the user's manipulation of physical objects. In experiments, we confirmed that the proposed method can stably generate physically plausible manipulation postures in real time.

Fields of Study & Subjects

Interactive and Immersive Displays Teleoperation and Haptic Systems Hand Gesture Recognition Systems

Publication details

Journal / source
IEEE International Conference on Artificial Intelligence and eXtended and Virtual Reality (AIxVR)
Pages
198–202
Publication date
26 January 2026
Language
EN