PhysGS: Bayesian-Inferred Gaussian Splatting for Physical Property Estimation

2025-11-23

AI Summary

Existing 3D reconstruction methods model only geometry and appearance, failing to infer critical physical properties—such as friction, stiffness, hardness, and material composition—thereby limiting safe robotic interaction. This work introduces the first framework unifying geometric reconstruction, uncertainty quantification, and dense physical property estimation by integrating Bayesian inference with 3D Gaussian Splatting (3DGS). We innovatively incorporate vision-language priors and explicitly distinguish aleatoric from epistemic uncertainty to enable iterative refinement of physical property estimates. Evaluated on multiple real-world datasets, our method improves physical property estimation accuracy by 22.8%, reduces Shore hardness estimation error by 61.2%, and decreases dynamic friction error by 18.1%, significantly outperforming deterministic baselines.

Technology Category

• Intelligent Robots: State Estimation
• Computer Vision: Low Level & Physics-based Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs
• Search and Retrieval-Augmented AI: Web learning to rank, online learning, and counterfactual learning for ranking
• Security and Privacy: Data transparency and provenance

Abstract

Understanding physical properties such as friction, stiffness, hardness, and material composition is essential for enabling robots to interact safely and effectively with their surroundings. However, existing 3D reconstruction methods focus on geometry and appearance and cannot infer these underlying physical properties. We present PhysGS, a Bayesian-inferred extension of 3D Gaussian Splatting that estimates dense, per-point physical properties from visual cues and vision--language priors. We formulate property estimation as Bayesian inference over Gaussian splats, where material and property beliefs are iteratively refined as new observations arrive. PhysGS also models aleatoric and epistemic uncertainties, enabling uncertainty-aware object and scene interpretation. Across object-scale (ABO-500), indoor, and outdoor real-world datasets, PhysGS improves accuracy of the mass estimation by up to 22.8%, reduces Shore hardness error by up to 61.2%, and lowers kinetic friction error by up to 18.1% compared to deterministic baselines. Our results demonstrate that PhysGS unifies 3D reconstruction, uncertainty modeling, and physical reasoning in a single, spatially continuous framework for dense physical property estimation. Additional results are available at https://samchopra2003.github.io/physgs.

Problem

• Estimating dense physical properties from visual cues
• Unifying 3D reconstruction with uncertainty-aware physical reasoning
• Improving accuracy of mass, hardness and friction estimation

Innovation

• Bayesian-inferred Gaussian Splatting for physical properties
• Models aleatoric and epistemic uncertainties explicitly
• Unifies reconstruction and physical reasoning in framework