Seeing Where to Deploy: Metric RGB-Based Traversability Analysis for Aerial-to-Ground Hidden Space Inspection

2026-03-15

AI Summary

This work addresses the challenges of region selection from an aerial perspective in covert air-to-ground spatial inspection, where scale ambiguity, reconstruction uncertainty, and terrain semantic understanding pose significant difficulties. The authors propose a geometry-semantic traversability analysis framework that relies solely on RGB imagery. By integrating multi-view dense reconstruction with temporally consistent semantic segmentation and incorporating embodied motion priors, the method recovers metric scale and constructs a confidence-aware 3D semantic map. This approach enables effective fusion of geometric and semantic information without LiDAR, facilitating deployment area assessment under reachability constraints. Experimental validation on a tethered UAV–ground robot platform demonstrates its capability to reliably identify feasible deployment regions in concealed environments.

Technology Category

• Computer Vision: Remote Sensing / Geospatial AI
• Intelligent Robots: Localization, Mapping, and Navigation
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Semantics and Knowledge: Methods to enhance, augment, integrate or synergize semantic models such as knowledge graphs and LLMs
• Security and Privacy: Large-scale security measurements
• Search and Retrieval-Augmented AI: Web query analysis, representation and understanding

Abstract

Inspection of confined infrastructure such as culverts often requires accessing hidden spaces whose entrances are reachable primarily from elevated viewpoints. Aerial-ground cooperation enables a UAV to deploy a compact UGV for interior exploration, but selecting a suitable deployment region from aerial observations requires metric terrain reasoning involving scale ambiguity, reconstruction uncertainty, and terrain semantics. We present a metric RGB-based geometric-semantic reconstruction and traversability analysis framework for aerial-to-ground hidden space inspection. A feed-forward multi-view RGB reconstruction backbone produces dense geometry, while temporally consistent semantic segmentation yields a 3D semantic map. To enable deployment-relevant measurements without LiDAR-based dense mapping, we introduce an embodied motion prior that recovers metric scale by enforcing consistency between predicted camera motion and onboard platform egomotion. From the metrically grounded reconstruction, we construct a confidence-aware geometric-semantic traversability map and evaluate candidate deployment zones under explicit reachability constraints. Experiments on a tethered UAV-UGV platform demonstrate reliable deployment-zone identification in hidden space scenarios.

Problem

• traversability analysis
• aerial-to-ground deployment
• metric reconstruction
• hidden space inspection
• RGB-based perception

Innovation

• metric reconstruction
• embodied motion prior
• geometric-semantic traversability
• aerial-to-ground inspection
• RGB-based mapping