Context-aware Learned Mesh-based Simulation via Trajectory-Level Meta-Learning

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Dahlinger, Philipp, Freymuth, Niklas, Hoang, Tai, Würth, Tobias, Volpp, Michael, Kärger, Luise, Neumann, Gerhard
Format: Preprint
Published: 2025
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866915741959192576
author Dahlinger, Philipp
Freymuth, Niklas
Hoang, Tai
Würth, Tobias
Volpp, Michael
Kärger, Luise
Neumann, Gerhard
author_facet Dahlinger, Philipp
Freymuth, Niklas
Hoang, Tai
Würth, Tobias
Volpp, Michael
Kärger, Luise
Neumann, Gerhard
contents Simulating object deformations is a critical challenge across many scientific domains, including robotics, manufacturing, and structural mechanics. Learned Graph Network Simulators (GNSs) offer a promising alternative to traditional mesh-based physics simulators. Their speed and inherent differentiability make them particularly well suited for applications that require fast and accurate simulations, such as robotic manipulation or manufacturing optimization. However, existing learned simulators typically rely on single-step observations, which limits their ability to exploit temporal context. Without this information, these models fail to infer, e.g., material properties. Further, they rely on auto-regressive rollouts, which quickly accumulate error for long trajectories. We instead frame mesh-based simulation as a trajectory-level meta-learning problem. Using Conditional Neural Processes, our method enables rapid adaptation to new simulation scenarios from limited initial data while capturing their latent simulation properties. We utilize movement primitives to directly predict fast, stable and accurate simulations from a single model call. The resulting approach, Movement-primitive Meta-MeshGraphNet (M3GN), provides higher simulation accuracy at a fraction of the runtime cost compared to state-of-the-art GNSs across several tasks.
format Preprint
id arxiv_https___arxiv_org_abs_2511_05234
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Context-aware Learned Mesh-based Simulation via Trajectory-Level Meta-Learning
Dahlinger, Philipp
Freymuth, Niklas
Hoang, Tai
Würth, Tobias
Volpp, Michael
Kärger, Luise
Neumann, Gerhard
Robotics
Machine Learning
Simulating object deformations is a critical challenge across many scientific domains, including robotics, manufacturing, and structural mechanics. Learned Graph Network Simulators (GNSs) offer a promising alternative to traditional mesh-based physics simulators. Their speed and inherent differentiability make them particularly well suited for applications that require fast and accurate simulations, such as robotic manipulation or manufacturing optimization. However, existing learned simulators typically rely on single-step observations, which limits their ability to exploit temporal context. Without this information, these models fail to infer, e.g., material properties. Further, they rely on auto-regressive rollouts, which quickly accumulate error for long trajectories. We instead frame mesh-based simulation as a trajectory-level meta-learning problem. Using Conditional Neural Processes, our method enables rapid adaptation to new simulation scenarios from limited initial data while capturing their latent simulation properties. We utilize movement primitives to directly predict fast, stable and accurate simulations from a single model call. The resulting approach, Movement-primitive Meta-MeshGraphNet (M3GN), provides higher simulation accuracy at a fraction of the runtime cost compared to state-of-the-art GNSs across several tasks.
title Context-aware Learned Mesh-based Simulation via Trajectory-Level Meta-Learning
topic Robotics
Machine Learning
url https://arxiv.org/abs/2511.05234