DynActiveGS: Active Gaussian Splatting for Dynamic Scene Reconstruction

2026-08-02

AI Summary

This work addresses the challenge of degraded scene reconstruction quality in dynamic environments, where motion interference complicates the distinction between unobserved static regions and unreliable dynamic areas. The authors propose a dynamic-aware active reconstruction framework based on 3D Gaussian splatting that explicitly decomposes uncertainty into structural and motion-induced components, enabling precise identification of both region types. Integrating online uncertainty prediction, uncertainty-weighted optimization, dynamic-aware view selection, and motion-constrained path planning, the method forms a closed-loop, robust exploration strategy. Experiments demonstrate that the approach significantly outperforms existing active reconstruction techniques across multiple dynamic benchmarks, consistently improving reconstruction accuracy, completeness, rendering quality, and exploration efficiency.

Abstract

We present DynActiveGS, a dynamic-aware active reconstruction framework based on 3D Gaussian Splatting (3DGS) for autonomous exploration in dynamic environments. The framework incrementally reconstructs a 3D Gaussian scene representation while suppressing motion-corrupted observations through online uncertainty prediction and uncertainty-weighted Gaussian optimization. A key component of DynActiveGS is the explicit decomposition of uncertainty into structural uncertainty and motion-induced uncertainty, which enables the system to distinguish under-reconstructed static regions from dynamically unreliable areas. Based on these uncertainty fields, DynActiveGS performs dynamic-aware viewpoint selection and dynamic-constrained path planning to favor informative yet stable observations during exploration. The resulting system forms a unified closed-loop pipeline for robust active reconstruction in dynamic scenes. Extensive experiments on challenging dynamic benchmarks demonstrate consistent improvements over existing active reconstruction baselines in reconstruction accuracy, completeness, rendering quality, and exploration efficiency.

Problem

• dynamic scene reconstruction
• active reconstruction
• 3D Gaussian Splatting
• motion-induced uncertainty
• autonomous exploration

Innovation

• Dynamic-aware reconstruction
• 3D Gaussian Splatting
• Uncertainty decomposition
• Active exploration
• Motion-robust optimization