AI Paper Atlas

Research map / Probabilistic and scientific modeling / Dynamical systems

Learning Particle Dynamics for Manipulating Rigid Bodies, Deformable Objects, and Fluids

Yunzhu Li, Jiajun Wu, Russ Tedrake, Joshua B. Tenenbaum, Antonio Torralba

ICLR 2019 · Probabilistic and scientific modeling › Dynamical systems · approach: Physics-informed dynamics learning · 169 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add hierarchical modeling (ablation)DPI-Net · BoxBathrollout MSE (×10^-2): 2.64 $\pm$ 0.69 -> 2.03 $\pm$ 0.41Improved
add hierarchical modeling (ablation)DPI-Net · RiceGriprollout MSE (×10^-2): 0.29 $\pm$ 0.13 -> 0.13 $\pm$ 0.07Improved
add hierarchical modeling (ablation)DPI-Net · FluidFallrollout MSE (×10^-2): 0.15 $\pm$ 0.03 -> 0.15 $\pm$ 0.03No clear effect
add hierarchical modeling (ablation)DPI-Net · FluidShakerollout MSE (×10^-2): 1.89 $\pm$ 0.36 -> 1.89 $\pm$ 0.36No clear effect
use dynamic interaction graphs instead of HRN's graph approachparticle dynamics modeling · FluidShakerollout MSE (×10^-2): 3.58 $\pm$ 0.77 -> 1.89 $\pm$ 0.36Improved
use state-specific motion predictors instead of a unified predictor (ablation)DPI-Net on BoxBath · BoxBathrollout MSEImproved
use fewer hierarchy roots (ablation)DPI-Net on RiceGrip · RiceGriprollout MSEImproved
increase propagation steps (ablation)DPI-Net on RiceGrip · RiceGriprollout MSEMixed
increase neighborhood size (ablation)DPI-Net on RiceGrip · RiceGriprollout MSEImproved
increase neighborhood size (ablation)DPI-Net on RiceGrip · RiceGripforward-step runtime: 30ms for d=0.04, 33ms for d=0.08, and 40ms for d=0.12Hurt
add online system identificationRiceGrip control · RiceGripChamfer distanceImproved
use model-based search and trajectory optimization instead of model-free RLFluidShake control · FluidShaketraining time steps: 600K time steps vs more than 10 million time stepsImproved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy