AI Paper Atlas

Research map / Probabilistic and scientific modeling / Dynamical systems

Extending Lagrangian and Hamiltonian Neural Networks with Differentiable Contact Models

Yaofeng Desmond Zhong, Biswadip Dey, Amit Chakraborty

NeurIPS 2021 · Probabilistic and scientific modeling › Dynamical systems · approach: Physics-informed dynamics learning · 6 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 regularization R=εI (CMr) to the contact model, vs unregularized CM (ablation)CM-CD-CLNN · BP5-erelative L1 trajectory error: 2 orders of magnitudeHurt
add regularization (CMr) vs CM (ablation)CM-CD-CLNN · Ropelearned friction μ (true 0.000): 0.026 -> 0.037Hurt
add regularization (CMr) vs CM (ablation)CM-CD-CLNN · Gyrolearned restitution e_P (true 0.800): 0.800 -> 0.892Hurt
add regularization (CMr) vs CM (ablation)CM-CD-CLNN · CP3-elearned friction μ (true 0.000): 0.000 -> 0.002Hurt
add regularization (CMr) vs CM (ablation)CM-CD-CLNN · CP3-e and Gyro-eenergy conservationHurt
use smaller regularization ε (e.g. 0.001) in CMr (ablation)CMr · contact property learningaccuracy of learned contact propertiesImproved
make ε a learnable parameter in CMr (ablation)CMr · contact property learningaccuracy of learned contact propertiesNo clear effect
replace CLNN with CHNN backboneCM contact model · 8 dynamics tasksrelative L1 trajectory errorNo clear effect
proposed contact model vs MLP/IN baselinesMLP-CD-CLNN, IN-CP-CLNN, IN-CP-SP · 8 dynamics tasks, 5-step predictionrelative L1 trajectory errorImproved
proposed contact model vs baselines on long trajectoriesbaselines · 50-step test trajectoriesrelative L1 errorImproved
train on LCP-generated dataCM · Gyro (3D)learned restitution e_P (true 0.800): 0.800 -> 0.822Hurt
train on noisy dataCM · dynamics tasksperformanceNo clear effect

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