One-Shot Real-to-Sim via End-to-End Differentiable Simulation and Rendering

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Zhu, Yifan, Xiang, Tianyi, Dollar, Aaron, Pan, Zherong
Natura: Preprint
Pubblicazione: 2024
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866918016488308736
author Zhu, Yifan
Xiang, Tianyi
Dollar, Aaron
Pan, Zherong
author_facet Zhu, Yifan
Xiang, Tianyi
Dollar, Aaron
Pan, Zherong
contents Identifying predictive world models for robots in novel environments from sparse online observations is essential for robot task planning and execution in novel environments. However, existing methods that leverage differentiable programming to identify world models are incapable of jointly optimizing the geometry, appearance, and physical properties of the scene. In this work, we introduce a novel rigid object representation that allows the joint identification of these properties. Our method employs a novel differentiable point-based geometry representation coupled with a grid-based appearance field, which allows differentiable object collision detection and rendering. Combined with a differentiable physical simulator, we achieve end-to-end optimization of world models, given the sparse visual and tactile observations of a physical motion sequence. Through a series of world model identification tasks in simulated and real environments, we show that our method can learn both simulation- and rendering-ready world models from only one robot action sequence. The code and additional videos are available at our project website: https://tianyi20.github.io/rigid-world-model.github.io/
format Preprint
id arxiv_https___arxiv_org_abs_2412_00259
institution arXiv
publishDate 2024
record_format arxiv
spellingShingle One-Shot Real-to-Sim via End-to-End Differentiable Simulation and Rendering
Zhu, Yifan
Xiang, Tianyi
Dollar, Aaron
Pan, Zherong
Robotics
Computer Vision and Pattern Recognition
Graphics
Identifying predictive world models for robots in novel environments from sparse online observations is essential for robot task planning and execution in novel environments. However, existing methods that leverage differentiable programming to identify world models are incapable of jointly optimizing the geometry, appearance, and physical properties of the scene. In this work, we introduce a novel rigid object representation that allows the joint identification of these properties. Our method employs a novel differentiable point-based geometry representation coupled with a grid-based appearance field, which allows differentiable object collision detection and rendering. Combined with a differentiable physical simulator, we achieve end-to-end optimization of world models, given the sparse visual and tactile observations of a physical motion sequence. Through a series of world model identification tasks in simulated and real environments, we show that our method can learn both simulation- and rendering-ready world models from only one robot action sequence. The code and additional videos are available at our project website: https://tianyi20.github.io/rigid-world-model.github.io/
title One-Shot Real-to-Sim via End-to-End Differentiable Simulation and Rendering
topic Robotics
Computer Vision and Pattern Recognition
Graphics
url https://arxiv.org/abs/2412.00259