Infinite Mobility: Scalable High-Fidelity Synthesis of Articulated Objects via Procedural Generation

2025-03-17

AI Summary

This work addresses the critical challenge of generating large-scale, high-fidelity articulated objects (e.g., hinge- and slide-based mechanisms) required for embodied AI. We propose the first parameterized, physics-aware generative framework specifically designed for articulated structures. Unlike prior methods constrained by limited data volume, noisy annotations, or low simulation fidelity, our approach integrates geometric constraint solving, rigid-body dynamics modeling, parametric CAD primitives, and differentiable rendering optimization to achieve photorealistic, editable, and scalable procedural generation. Quantitative evaluation and user studies demonstrate significant improvements over state-of-the-art methods. Moreover, synthetic data generated by our framework substantially enhances the performance of downstream 3D generative models—validating its efficacy and practical utility as a high-quality synthetic data source.

Technology Category

• Humans and AI: Game Design — Procedural Content Generation & Storytelling
• Intelligent Robots: Embodied AI
• Natural Language Processing: Code Generation / Program Synthesis from Natural Language

Application Category

• Economics, Online Markets and Human Computation: Economic ramifications for generative AI infrastructure and applications
• Social Networks and Social Media: Generative AI / large language models and their impact on social systems
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Large-scale articulated objects with high quality are desperately needed for multiple tasks related to embodied AI. Most existing methods for creating articulated objects are either data-driven or simulation based, which are limited by the scale and quality of the training data or the fidelity and heavy labour of the simulation. In this paper, we propose Infinite Mobility, a novel method for synthesizing high-fidelity articulated objects through procedural generation. User study and quantitative evaluation demonstrate that our method can produce results that excel current state-of-the-art methods and are comparable to human-annotated datasets in both physics property and mesh quality. Furthermore, we show that our synthetic data can be used as training data for generative models, enabling next-step scaling up. Code is available at https://github.com/Intern-Nexus/Infinite-Mobility

Problem

• Generates high-fidelity articulated objects for AI tasks
• Overcomes limitations of data-driven and simulation-based methods
• Produces synthetic data for training generative models

Innovation

• Procedural generation for high-fidelity articulated objects
• Synthetic data for training generative models
• Exceeds state-of-the-art in physics and mesh quality