HERE: Hierarchical Active Exploration of Radiance Field with Epistemic Uncertainty Minimization

TL;DR

The paper introduces HERE, an active 3D scene reconstruction framework using neural radiance fields and epistemic uncertainty to guide camera exploration, achieving more complete and efficient scene mapping.

Contribution

It proposes a hierarchical active exploration method leveraging epistemic uncertainty quantification for improved 3D scene reconstruction with neural radiance fields.

Findings

Higher reconstruction completeness than previous methods.

Effective identification of unexplored scene regions.

Validated on simulated scenes and real-world hardware.

Abstract

We present HERE, an active 3D scene reconstruction framework based on neural radiance fields, enabling high-fidelity implicit mapping. Our approach centers around an active learning strategy for camera trajectory generation, driven by accurate identification of unseen regions, which supports efficient data acquisition and precise scene reconstruction. The key to our approach is epistemic uncertainty quantification based on evidential deep learning, which directly captures data insufficiency and exhibits a strong correlation with reconstruction errors. This allows our framework to more reliably identify unexplored or poorly reconstructed regions compared to existing methods, leading to more informed and targeted exploration. Additionally, we design a hierarchical exploration strategy that leverages learned epistemic uncertainty, where local planning extracts target viewpoints from high-uncertainty voxels based on visibility for trajectory generation, and global planning uses uncertainty to guide large-scale coverage for efficient and comprehensive reconstruction. The effectiveness of the proposed method in active 3D reconstruction is demonstrated by achieving higher reconstruction completeness compared to previous approaches on photorealistic simulated scenes across varying scales, while a hardware demonstration further validates its real-world applicability. Project page: https://taekbum.github.io/here/