Enhancing Glass Surface Reconstruction via Depth Prior for Robot Navigation

TL;DR

This paper presents a training-free method that uses depth foundation models and RANSAC alignment to improve indoor robot navigation around glass surfaces, recovering accurate depth and scale.

Contribution

It introduces a novel framework leveraging depth foundation models as structural priors with RANSAC for robust fusion, and provides a new RGB-D dataset with ground truth for glass regions.

Findings

Outperforms state-of-the-art methods under severe depth corruption.

Recovers accurate metric scale in glass-affected indoor environments.

Demonstrates robustness and effectiveness through extensive experiments.

Abstract

Indoor robot navigation is often compromised by glass surfaces, which severely corrupt depth sensor measurements. While foundation models like Depth Anything 3 provide excellent geometric priors, they lack an absolute metric scale. We propose a training-free framework that leverages depth foundation models as a structural prior, employing a robust local RANSAC-based alignment to fuse it with raw sensor depth. This naturally avoids contamination from erroneous glass measurements and recovers an accurate metric scale. Furthermore, we introduce \ti{GlassRecon}, a novel RGB-D dataset with geometrically derived ground truth for glass regions. Extensive experiments demonstrate that our approach consistently outperforms state-of-the-art baselines, especially under severe sensor depth corruption. The dataset and related code will be released at https://github.com/jarvisyjw/GlassRecon.