HSImul3R: Physics-in-the-Loop Reconstruction of Simulation-Ready Human-Scene Interactions

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Cao, Yukang, Xie, Haozhe, Hong, Fangzhou, Zhuo, Long, Chen, Zhaoxi, Pan, Liang, Liu, Ziwei
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915867530362880
author Cao, Yukang
Xie, Haozhe
Hong, Fangzhou
Zhuo, Long
Chen, Zhaoxi
Pan, Liang
Liu, Ziwei
author_facet Cao, Yukang
Xie, Haozhe
Hong, Fangzhou
Zhuo, Long
Chen, Zhaoxi
Pan, Liang
Liu, Ziwei
contents We present HSImul3R, a unified framework for simulation-ready 3D reconstruction of human-scene interactions (HSI) from casual captures, including sparse-view images and monocular videos. Existing methods suffer from a perception-simulation gap: visually plausible reconstructions often violate physical constraints, leading to instability in physics engines and failure in embodied AI applications. To bridge this gap, we introduce a physically-grounded bi-directional optimization pipeline that treats the physics simulator as an active supervisor to jointly refine human dynamics and scene geometry. In the forward direction, we employ Scene-targeted Reinforcement Learning to optimize human motion under dual supervision of motion fidelity and contact stability. In the reverse direction, we propose Direct Simulation Reward Optimization, which leverages simulation feedback on gravitational stability and interaction success to refine scene geometry. We further present HSIBench, a new benchmark with diverse objects and interaction scenarios. Extensive experiments demonstrate that HSImul3R produces the first stable, simulation-ready HSI reconstructions and can be directly deployed to real-world humanoid robots.
format Preprint
id arxiv_https___arxiv_org_abs_2603_15612
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle HSImul3R: Physics-in-the-Loop Reconstruction of Simulation-Ready Human-Scene Interactions
Cao, Yukang
Xie, Haozhe
Hong, Fangzhou
Zhuo, Long
Chen, Zhaoxi
Pan, Liang
Liu, Ziwei
Computer Vision and Pattern Recognition
Robotics
We present HSImul3R, a unified framework for simulation-ready 3D reconstruction of human-scene interactions (HSI) from casual captures, including sparse-view images and monocular videos. Existing methods suffer from a perception-simulation gap: visually plausible reconstructions often violate physical constraints, leading to instability in physics engines and failure in embodied AI applications. To bridge this gap, we introduce a physically-grounded bi-directional optimization pipeline that treats the physics simulator as an active supervisor to jointly refine human dynamics and scene geometry. In the forward direction, we employ Scene-targeted Reinforcement Learning to optimize human motion under dual supervision of motion fidelity and contact stability. In the reverse direction, we propose Direct Simulation Reward Optimization, which leverages simulation feedback on gravitational stability and interaction success to refine scene geometry. We further present HSIBench, a new benchmark with diverse objects and interaction scenarios. Extensive experiments demonstrate that HSImul3R produces the first stable, simulation-ready HSI reconstructions and can be directly deployed to real-world humanoid robots.
title HSImul3R: Physics-in-the-Loop Reconstruction of Simulation-Ready Human-Scene Interactions
topic Computer Vision and Pattern Recognition
Robotics
url https://arxiv.org/abs/2603.15612