期刊
ACM TRANSACTIONS ON GRAPHICS
卷 41, 期 4, 页码 -出版社
ASSOC COMPUTING MACHINERY
DOI: 10.1145/3528223.3530067
关键词
Character Animation; Physics-based Simulation and Control; Motion Capture; Reinforcement Learning; Variational Autoencoder; Behavior Cloning
This paper presents an algorithm for building physics-based controllers for physically simulated characters with multiple degrees of freedom. The controllers, learned using conditional VAEs, can generate diverse and plausible human motions without conditioning on specific goals. They are robust enough to solve complex downstream tasks efficiently.
High-quality motion capture datasets are now publicly available, and researchers have used them to create kinematics-based controllers that can generate plausible and diverse human motions without conditioning on specific goals (i.e., a task-agnostic generative model). In this paper, we present an algorithm to build such controllers for physically simulated characters having many degrees of freedom. Our physics-based controllers are learned by using conditional VAEs, which can perform a variety of behaviors that are similar to motions in the training dataset. The controllers are robust enough to generate more than a few minutes of motion without conditioning on specific goals and to allow many complex downstream tasks to be solved efficiently. To show the effectiveness of our method, we demonstrate controllers learned from several different motion capture databases and use them to solve a number of downstream tasks that are challenging to learn controllers that generate natural-looking motions from scratch. We also perform ablation studies to demonstrate the importance of the elements of the algorithm.
作者
我是这篇论文的作者
点击您的名字以认领此论文并将其添加到您的个人资料中。
推荐
暂无数据