Guiding Diffusion-Based Articulated Object Generation by Partial Point Cloud Alignment and Physical Plausibility Constraints

2025-08-01

AI Summary

This work addresses the challenge of generating physically plausible articulated objects guided by partial point clouds. We propose PhysNAP—the first generative method that jointly embeds geometric alignment and physical constraints into the reverse diffusion process. PhysNAP represents part geometry using signed distance functions (SDFs) and jointly optimizes point-cloud alignment loss, non-penetration constraints, and mobility constraints during denoising—ensuring geometric fidelity, collision-free configurations, and kinematically valid articulation. Crucially, it enables end-to-end, category-conditioned modeling of movable parts without post-processing. Experiments on PartNet-Mobility demonstrate that PhysNAP significantly improves physical plausibility (+28.6% constraint satisfaction rate) while maintaining high-fidelity reconstruction quality, outperforming physics-agnostic baselines.

Technology Category

• Computer Vision: Low Level & Physics-based Vision
• Natural Language Processing: Generation
• Intelligent Robots: Motion and Path Planning

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Search and Retrieval-Augmented AI: Agentic search
• User Modeling, Personalization and Recommendation: Fairness-aware retrieval and ranking

Abstract

Articulated objects are an important type of interactable objects in everyday environments. In this paper, we propose PhysNAP, a novel diffusion model-based approach for generating articulated objects that aligns them with partial point clouds and improves their physical plausibility. The model represents part shapes by signed distance functions (SDFs). We guide the reverse diffusion process using a point cloud alignment loss computed using the predicted SDFs. Additionally, we impose non-penetration and mobility constraints based on the part SDFs for guiding the model to generate more physically plausible objects. We also make our diffusion approach category-aware to further improve point cloud alignment if category information is available. We evaluate the generative ability and constraint consistency of samples generated with PhysNAP using the PartNet-Mobility dataset. We also compare it with an unguided baseline diffusion model and demonstrate that PhysNAP can improve constraint consistency and provides a tradeoff with generative ability.

Problem

• Generates articulated objects aligning with partial point clouds
• Improves physical plausibility via non-penetration and mobility constraints
• Enhances category-aware point cloud alignment using diffusion models

Innovation

• Diffusion model guided by point cloud alignment
• SDF-based physical plausibility constraints
• Category-aware diffusion for improved alignment