Search papers, labs, and topics across Lattice.
This paper introduces Physics-Guided Residual Dynamics (PGRD), a hybrid framework that enhances the simulation of deformable objects by integrating a physics-based spring-mass simulator with a neural network that corrects its predictions. The method employs a velocity-based formulation for stability and utilizes a sliding-window transformer to effectively capture temporal dependencies in the dynamics of deformable objects. Experimental results demonstrate that PGRD outperforms both traditional physics-based and purely learning-based methods in accuracy across various real-world scenarios, while also showcasing its applicability in manipulation planning and interactive simulation tasks.
PGRD achieves superior accuracy in deformable object simulation by seamlessly blending physics-based principles with advanced neural corrections.
Simulating deformable objects is essential for a wide range of robotic manipulation applications, yet accurately predicting their dynamics remains challenging. We propose Physics-Guided Residual Dynamics (PGRD), a hybrid simulation framework that combines the advantages of physics-based and learning-based approaches. Specifically, PGRD combines an optimizable spring-mass simulator as a backbone with a learned neural network that predicts residual corrections to the physics-based predictions. We adopt a velocity-based formulation to ensure stable simulation and a sliding-window transformer architecture to capture temporal dependencies. We show that PGRD produces more accurate results than both purely physics-based and learning-based methods on a set of diverse real-world deformable objects. We further demonstrate the utility of PGRD in two applications: manipulation planning via Model Predictive Control, including a language-conditioned setting with a generated goal image; and interactive simulation via action-conditioned video prediction by 3D Gaussian Splatting.
