Saved in:
Bibliographic Details
Main Authors: Tan, Jing, Xu, Weisheng, Jiang, Xiangrui, Zhang, Jiaxi, Yang, Kun, Wu, Kai, Xiong, Jiaqi, Chen, Shiting, Li, Yangfan, Feng, Yixiao, Fang, Yuetong, Zou, Yujia, Song, Yiqun, Xu, Renjing
Format: Preprint
Published: 2026
Subjects:
Online Access:https://arxiv.org/abs/2603.01294
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866918364283142144
author Tan, Jing
Xu, Weisheng
Jiang, Xiangrui
Zhang, Jiaxi
Yang, Kun
Wu, Kai
Xiong, Jiaqi
Chen, Shiting
Li, Yangfan
Feng, Yixiao
Fang, Yuetong
Zou, Yujia
Song, Yiqun
Xu, Renjing
author_facet Tan, Jing
Xu, Weisheng
Jiang, Xiangrui
Zhang, Jiaxi
Yang, Kun
Wu, Kai
Xiong, Jiaqi
Chen, Shiting
Li, Yangfan
Feng, Yixiao
Fang, Yuetong
Zou, Yujia
Song, Yiqun
Xu, Renjing
contents Learning motion priors for physics-based humanoid control is an active research topic. Existing approaches mainly include variational autoencoders (VAE) and adversarial motion priors (AMP). VAE introduces information loss, and random latent sampling may sometimes produce invalid behaviors. AMP suffers from mode collapse and struggles to capture diverse motion skills. We present the Spherical Latent Motion Prior (SLMP), a two-stage method for learning motion priors. In the first stage, we train a high-quality motion tracking controller. In the second stage, we distill the tracking controller into a spherical latent space. A combination of distillation, a discriminator, and a discriminator-guided local semantic consistency constraint shapes a structured latent action space, allowing stable random sampling without information loss. To evaluate SLMP, we collect a two-hour human combat motion capture dataset and show that SLMP preserves fine motion detail without information loss, and random sampling yields semantically valid and stable behaviors. When applied to a two-agent physics-based combat task, SLMP produces human-like and physically plausible combat behaviors only using simple rule-based rewards. Furthermore, SLMP generalizes across different humanoid robot morphologies, demonstrating its transferability beyond a single simulated avatar.
format Preprint
id arxiv_https___arxiv_org_abs_2603_01294
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Spherical Latent Motion Prior for Physics-Based Simulated Humanoid Control
Tan, Jing
Xu, Weisheng
Jiang, Xiangrui
Zhang, Jiaxi
Yang, Kun
Wu, Kai
Xiong, Jiaqi
Chen, Shiting
Li, Yangfan
Feng, Yixiao
Fang, Yuetong
Zou, Yujia
Song, Yiqun
Xu, Renjing
Robotics
Learning motion priors for physics-based humanoid control is an active research topic. Existing approaches mainly include variational autoencoders (VAE) and adversarial motion priors (AMP). VAE introduces information loss, and random latent sampling may sometimes produce invalid behaviors. AMP suffers from mode collapse and struggles to capture diverse motion skills. We present the Spherical Latent Motion Prior (SLMP), a two-stage method for learning motion priors. In the first stage, we train a high-quality motion tracking controller. In the second stage, we distill the tracking controller into a spherical latent space. A combination of distillation, a discriminator, and a discriminator-guided local semantic consistency constraint shapes a structured latent action space, allowing stable random sampling without information loss. To evaluate SLMP, we collect a two-hour human combat motion capture dataset and show that SLMP preserves fine motion detail without information loss, and random sampling yields semantically valid and stable behaviors. When applied to a two-agent physics-based combat task, SLMP produces human-like and physically plausible combat behaviors only using simple rule-based rewards. Furthermore, SLMP generalizes across different humanoid robot morphologies, demonstrating its transferability beyond a single simulated avatar.
title Spherical Latent Motion Prior for Physics-Based Simulated Humanoid Control
topic Robotics
url https://arxiv.org/abs/2603.01294