REVIEW 7 cited by
Physical Design using Differentiable Learned Simulators
Not yet reviewed by Pith; the record is open.
This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.
SPECIMEN: schema-true, not a live event
T0 review · schema-true
One-sentence machine reading of the paper's core claim.
pith:XXXXXXXX · record.json · timestamp
read the original abstract
Designing physical artifacts that serve a purpose - such as tools and other functional structures - is central to engineering as well as everyday human behavior. Though automating design has tremendous promise, general-purpose methods do not yet exist. Here we explore a simple, fast, and robust approach to inverse design which combines learned forward simulators based on graph neural networks with gradient-based design optimization. Our approach solves high-dimensional problems with complex physical dynamics, including designing surfaces and tools to manipulate fluid flows and optimizing the shape of an airfoil to minimize drag. This framework produces high-quality designs by propagating gradients through trajectories of hundreds of steps, even when using models that were pre-trained for single-step predictions on data substantially different from the design tasks. In our fluid manipulation tasks, the resulting designs outperformed those found by sampling-based optimization techniques. In airfoil design, they matched the quality of those obtained with a specialized solver. Our results suggest that despite some remaining challenges, machine learning-based simulators are maturing to the point where they can support general-purpose design optimization across a variety of domains.
Forward citations
Cited by 7 Pith papers
-
Transformer Transformer: A Unified Model for Motion-Conditioned Robot Co-design
A single diffusion transformer trains on tokenized robot bodies and motions to generate and optimize robot designs for unseen rewards and trajectories, outpacing evolutionary search in speed and often in reward.
-
Hybrid Physics-Machine Learning Models for Quantitative Electron Diffraction Refinements
A differentiable physics-machine-learning hybrid jointly fits structural parameters and neural-network corrections for experimental effects in 3D electron diffraction refinement.
-
VLMgineer: Vision Language Models as Robotic Toolsmiths
VLMgineer combines VLM-generated URDF tool designs with evolutionary search to co-design tools and action plans, outperforming human-specified and existing tools on a new simulated manipulation benchmark.
-
RobotSmith: Generative Robotic Tool Design for Acquisition of Complex Manipulation Skills
RobotSmith autonomously designs, 3D-prints, and uses task-specific tools for robotic manipulation, raising task success from 2.8% (no tool) to 50% in simulation.
-
Dynamical Data for More Efficient and Generalizable Learning: A Case Study in Disordered Elastic Networks
A graph neural network trained only on compression trajectories of disordered elastic networks infers Poisson's ratio and generalizes to networks with Poisson's ratios outside the training range.
-
Accelerating PDE-Constrained Optimization by the Derivative of Neural Operators
Reference neural operators with a Virtual-Fourier layer learn solution derivatives and a hybrid solver feedback loop accelerates PDE-constrained optimization.
-
Inverse Design with Dynamic Mode Decomposition
Fitting a low-rank, linear-in-parameter DMD model by least squares gives a fast surrogate for inverse design and optimization of dynamical systems.
Discussion (0). Continue with ORCID to comment.