Enhancing Glass Surface Reconstruction via Depth Prior for Robot Navigation
Jiamin Zheng, Jingwen Yu, Guangcheng Chen, Hong Zhang
TLDR
A training-free framework improves robot navigation by fusing depth foundation models with raw sensor depth to reconstruct glass surfaces accurately.
Key contributions
- Proposes a training-free framework for robust glass surface reconstruction.
- Leverages depth foundation models as a structural prior for geometric accuracy.
- Fuses prior with raw sensor depth using robust local RANSAC-based alignment.
- Introduces GlassRecon, a novel RGB-D dataset with geometrically derived glass ground truth.
Why it matters
Indoor robot navigation is often compromised by glass surfaces corrupting depth measurements. This paper offers a robust, training-free solution that accurately reconstructs glass, significantly improving robot perception and safety in complex environments. It also provides a valuable new dataset for further research.
Original 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.
📬 Weekly AI Paper Digest
Get the top 10 AI/ML arXiv papers from the week — summarized, scored, and delivered to your inbox every Monday.