ArticFlow: Generative Simulation of Articulated Mechanisms

2025-11-21

AI Summary

Existing generative models excel at static 3D shape synthesis but struggle with joint modeling of deformable articulated structures—i.e., simultaneous morphology generation and motion simulation—due to challenges in modeling motion-dependent deformations and scarcity of annotated motion-geometry paired data. This paper proposes a two-stage flow-matching framework: Stage I learns motion-conditioned latent flows to encode motion priors; Stage II couples point-wise flows to enable explicit, controllable point-cloud generation under articulated motion. By performing joint interpolation in the shared latent space, the method unifies shape priors and dynamic responses. It supports cross-category generalization and co-synthesis of novel morphologies with corresponding motions, integrating generative capability with neural simulation. Evaluated on the MuJoCo Menagerie benchmark, our approach significantly improves both kinematic accuracy and geometric fidelity over dedicated physics simulators and static point-cloud generators.

Technology Category

• Natural Language Processing: Code Generation / Program Synthesis from Natural Language
• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: Motion & Tracking

Application Category

• User Modeling, Personalization and Recommendation: User modeling and simulation for interactive and conversational systems
• Web Mining and Content Analysis: Web data generation and simulation
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems

Abstract

Recent advances in generative models have produced strong results for static 3D shapes, whereas articulated 3D generation remains challenging due to action-dependent deformations and limited datasets. We introduce ArticFlow, a two-stage flow matching framework that learns a controllable velocity field from noise to target point sets under explicit action control. ArticFlow couples (i) a latent flow that transports noise to a shape-prior code and (ii) a point flow that transports points conditioned on the action and the shape prior, enabling a single model to represent diverse articulated categories and generalize across actions. On MuJoCo Menagerie, ArticFlow functions both as a generative model and as a neural simulator: it predicts action-conditioned kinematics from a compact prior and synthesizes novel morphologies via latent interpolation. Compared with object-specific simulators and an action-conditioned variant of static point-cloud generators, ArticFlow achieves higher kinematic accuracy and better shape quality. Results show that action-conditioned flow matching is a practical route to controllable and high-quality articulated mechanism generation.

Problem

• Generating articulated 3D shapes with action-dependent deformations
• Learning controllable velocity fields for articulated mechanism simulation
• Overcoming limited datasets for diverse articulated category generation

Innovation

• Two-stage flow matching framework for articulated mechanisms
• Latent and point flows enable action-conditioned generation
• Functions as both generative model and neural simulator