S2O: Static to Openable Enhancement for Articulated 3D Objects

2024-09-27

arXiv.org

AI Summary

Existing large-scale 3D datasets lack diverse, interactive, articulable objects and suffer from high manual annotation costs. To address this, we introduce the novel “Static-to-Operable” (S2O) generation task: given a static 3D model, jointly detect movable parts, regress joint parameters, model motion trajectories, and complete internal geometry via implicit fields to produce physically drivable articulated 3D objects. We construct the first large-scale S2O benchmark dataset and propose a unified end-to-end framework integrating part segmentation, differentiable joint optimization, and geometric prior constraints. Experiments on our dataset show that the best-performing method achieves a 68.3% simulation success rate for opening/closing motions—significantly outperforming all baselines. This work establishes a new paradigm for robotic manipulation and embodied AI by enabling scalable, physics-aware articulation synthesis from static geometry.

Technology Category

• Intelligent Robots: Manipulation
• Computer Vision: 3D Computer Vision
• Search and Optimization: Sampling/Simulation-based Search

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Semantics and Knowledge: Data modeling to support human-machine intelligence, including LLMs agents, intelligent system behavior, explanations, and user-friendly interactions

Abstract

Despite much progress in large 3D datasets there are currently few interactive 3D object datasets, and their scale is limited due to the manual effort required in their construction. We introduce the static to openable (S2O) task which creates interactive articulated 3D objects from static counterparts through openable part detection, motion prediction, and interior geometry completion. We formulate a unified framework to tackle this task, and curate a challenging dataset of openable 3D objects that serves as a test bed for systematic evaluation. Our experiments benchmark methods from prior work, extended and improved methods, and simple yet effective heuristics for the S2O task. We find that turning static 3D objects into interactively openable counterparts is possible but that all methods struggle to generalize to realistic settings of the task, and we highlight promising future work directions. Our work enables efficient creation of interactive 3D objects for robotic manipulation and embodied AI tasks.

Problem

• Enhance static 3D objects to interactive openable versions
• Detect openable parts and predict motion for 3D objects
• Complete interior geometry for articulated 3D object creation

Innovation

• Openable part detection for 3D objects
• Motion prediction in static 3D models
• Interior geometry completion for interactivity