Learning 3D Particle-based Simulators from RGB-D Videos

TL;DR

This paper introduces Visual Particle Dynamics (VPD), a method that learns 3D particle-based simulators directly from RGB-D videos, enabling realistic scene modeling, editing, and long-term prediction without privileged information.

Contribution

VPD is the first end-to-end framework that learns 3D particle representations, dynamics, and rendering from RGB-D videos without requiring ground truth physics data.

Findings

VPD accurately predicts long-term scene evolution.

VPD enables scene editing and view synthesis.

VPD outperforms existing 2D video prediction models.

Abstract

Realistic simulation is critical for applications ranging from robotics to animation. Traditional analytic simulators sometimes struggle to capture sufficiently realistic simulation which can lead to problems including the well known "sim-to-real" gap in robotics. Learned simulators have emerged as an alternative for better capturing real-world physical dynamics, but require access to privileged ground truth physics information such as precise object geometry or particle tracks. Here we propose a method for learning simulators directly from observations. Visual Particle Dynamics (VPD) jointly learns a latent particle-based representation of 3D scenes, a neural simulator of the latent particle dynamics, and a renderer that can produce images of the scene from arbitrary views. VPD learns end to end from posed RGB-D videos and does not require access to privileged information. Unlike existing 2D video prediction models, we show that VPD's 3D structure enables scene editing and long-term predictions. These results pave the way for downstream applications ranging from video editing to robotic planning.