Learning Equivariant Neural-Augmented Object Dynamics From Few Interactions

2026-05-04

AI Summary

This work addresses the challenge of efficiently learning physically plausible long-term dynamics of both rigid and deformable objects under limited real interaction data. The authors propose PIEGraph, a novel framework that uniquely integrates an analytical spring-mass physical model with an equivariant graph neural network. By leveraging the symmetry properties of particle systems, PIEGraph constructs object representations and introduces a symmetry-aware action encoding mechanism. This approach substantially improves data efficiency and long-horizon prediction stability. Extensive experiments demonstrate that PIEGraph achieves high-fidelity dynamics prediction and reliable manipulation planning across diverse object categories—including ropes, fabrics, plush toys, and rigid bodies—on both simulated and real robotic platforms, outperforming state-of-the-art methods.

Abstract

Learning data-efficient object dynamics models for robotic manipulation remains challenging, especially for deformable objects. A popular approach is to model objects as sets of 3D particles and learn their motion using graph neural networks. In practice, this is not enough to maintain physical feasibility over long horizons and may require large amounts of interaction data to learn. We introduce PIEGraph, a novel approach to combining analytical physics and data-driven models to capture object dynamics for both rigid and deformable bodies using limited real-world interaction data. PIEGraph consists of two components: (1) a \textbf{P}hysically \textbf{I}nformed particle-based analytical model (implemented as a spring--mass system) to enforce physically feasible motion, and (2) an \textbf{E}quivariant \textbf{Graph} Neural Network with a novel action representation that exploits symmetries in particle interactions to guide the analytical model. We evaluate PIEGraph in simulation and on robot hardware for reorientation and repositioning tasks with ropes, cloth, stuffed animals and rigid objects. We show that our method enables accurate dynamics prediction and reliable downstream robotic manipulation planning, which outperforms state of the art baselines.

Problem

• deformable objects
• data-efficient dynamics
• physical feasibility
• robotic manipulation
• object dynamics

Innovation

• equivariant graph neural network
• physics-informed modeling
• data-efficient dynamics learning
• deformable object manipulation
• particle-based simulation