Utilizing Motion Matching with Deep Reinforcement Learning for Target Location Tasks

Eurographics 2024 Short Papers
Jeongmin Lee1,2 Taesoo Kwon2 Hyunju Shin1* Yoonsang Lee2*
1 Samsung Electronics Co., Seoul R&D Campus
2 Hanyang University, Department of Computer Science
* Co-corresponding authors
Publisher arXiv Slides (PDF) Slides (PPTX)

Video


Abstract

We present an approach using deep reinforcement learning (DRL) to directly generate motion matching queries for long-term tasks, particularly targeting the reaching of specific locations. By integrating motion matching and DRL, our method demonstrates the rapid learning of policies for target location tasks within minutes on a standard desktop, employing a simple reward design. Additionally, we propose a unique hit reward and obstacle curriculum scheme to enhance policy learning in environments with moving obstacles.

Paper

Publisher: page, paper
arXiv: page, paper

Presentation

Eurographics 2024 Presentation Slides: pdf (1.5MB), pptx (289.5MB)