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

Seoyoung Lee; Shaekh Mohammad Shithil; Durgakant Pushp; Lantao Liu; Zhangyang Wang

cs.ROMar 15, 2026

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

Authors: Seoyoung Lee, Shaekh Mohammad Shithil, Durgakant Pushp, Lantao Liu, Zhangyang Wang

Organizations: The University of Texas at Austin, USA. · Indiana University, Bloomington, USA.

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 requiring LiDAR-based dense mapping, we introduce an embodied motion prior that recovers metric scale by aligning predicted camera motion with onboard platform egomotion. From the metrically grounded reconstruction, we construct a confidence-aware geometric-semantic traversability map and evaluate candidate deployment zones under reachability constraints. Experiments on a tethered UAV-UGV platform demonstrate reliable deployment-zone identification in hidden space scenarios.

Figures & tables

Fig. 1 : Overview of proposed pipeline. From a UAV-captured top-down RGB video, VGGT predicts camera parameters and depth, which yield dense 3D geometry via depth unprojection. In parallel, adaptive segmentation generates temporally consistent instance masks that are lifted into 3D to produce a geometry-aligned semantic reconstruction. Onboard egomotion recovers metric scale for the reconstruction through motion-consistent alignment, enabling deployment-relevant target extraction for traversability analysis and deployment planning.

Fig. 2 : Integrated aerial–ground robotic system . The DJI Matrice M600 hexacopter carries a compact tracked ground robot and integrates a Livox Mid-360 LiDAR, RealSense camera, and Jetson Xavier NX compute module. FAST-LIO provides egomotion, but LiDAR point clouds are not used for reconstruction or traversability.

Model	ATE $\downarrow$	RPE trans $\downarrow$	RPE rot $\downarrow$
LSM / CIPHER	0.225	0.362	2.320
Easi3R	0.481	0.422	2.901
Fast3R	0.361	0.377	1.938
MASt3R	0.207	0.358	2.030
Ours	0.222	0.338	1.920
Ours + Relative Motion	0.194	0.310	1.878

TABLE I: Comparison of Multi-view Pose Estimation.

Model	Peak GPU Memory (GB) $\downarrow$	Inference Time (sec) $\downarrow$
LSM / CIPHER	28.89	526.47
Easi3R	33.26	233.61
Fast3R	19.24	7.60
MASt3R	24.42	15.20
Ours	14.09	3.37
Ours + Relative Motion	16.41	5.28

TABLE II: Comparison of Average Inference Efficiency. Runtime was measured with one RTX A6000 GPU.

Fig. 3 : RGB-based geometric and semantic 3D reconstruction from purely top-down aerial views at increasing UAV elevations : a culvert entrance (left) and a vent system occluded by surrounding structures (right). For each elevation, RGB reconstructions are shown above their corresponding semantic reconstructions for both top-down and ground-level views. Ground-level views also include the camera trajectory. Despite limited viewpoint parallax for overhead observations and higher altitudes, our method maintains geometrically consistent ground-level renderings and accurately segments deployment-relevant targets across a wide range of flight altitudes from RGB imagery. Semantic mask granularity can be adjusted to capture structures at different spatial scales as required.

Fig. 4 : Visual Results of Traversability mapping. From left to right: RGB-based 3D reconstruction, semantic segmentation, geometric traversability map, and geometric–semantic fusion result. The geometric map assigns high scores to planar regions but partially misclassifies rocky areas due to smoothed geometry. The fusion map assigns low traversability to terrain classes such as rocks and structural obstacles, producing a more conservative and deployment-consistent representation.

Model	mACC $\uparrow$	aAcc $\uparrow$	ROC AUC $\uparrow$	MSE $\downarrow$
Geometric	0.779	0.841	0.926	0.113
Fusion	0.958	0.938	0.948	0.042

TABLE III: Comparison of Traversability Accuracy.

Fig. 5 : Deployment location selection and trajectory generation. High traversability regions (green) indicate safe deployment areas, while red regions denote unsafe terrain. Candidate deployment locations (blue) are identified and feasible trajectories are generated from the UAV to the deployment zone.

Robotics

Robotics studies the design and control of machines that perceive, move, and act in physical environments.

See all 751 papers

MiniUGV $_2$ : A Compact UAV-Deployable Tracked Ground Vehicle with Manipulation Capabilities
Mar 1, 2026Durgakant Pushp, Swapnil Kalhapure, Shaekh Mohammad Shithil +1Robotics Mobile Robotics
Luddy School of Informatics, Computing, and Engineering at Indiana University, Bloomington, IN 47408, USA. · S. Kalhapure is an Independent Researcher at the time of work.
Towards Reliable Aerial Ground Vehicle Collaboration: An Integrated Planning and Autonomy Framework for Field Deployment
Jul 7, 2026Md Safwan Mondal, Luca Russo, James D. Humann +2Vehicle Routing Multi-Robot Coordination
Department of Mechanical and Industrial Engineering, University of Illinois Chicago, IL, 60607 USA · DEVCOM Army Research Laboratory, Minneapolis, Minnesota, USA · DEVCOM Army Research Laboratory, Aberdeen Proving Grounds, Aberdeen, MD 21005 USA
FC-Vision: Real-Time Visibility-Aware Replanning for Occlusion-Free Aerial Target Structure Scanning in Unknown Environments
Feb 14, 2026Chen Feng, Yang Xu, Shaojie ShenNext-Best-View Planning UAV Navigation
Dept. of Electronic and Computer Engineering, The Hong Kong University of Science and Technology, Hong Kong, China.

Mar 1, 2026J/K move · Enter open · S save

MiniUGV $_2$ : A Compact UAV-Deployable Tracked Ground Vehicle with Manipulation Capabilities

Durgakant Pushp, Swapnil Kalhapure, Shaekh Mohammad Shithil, Lantao Liu

Luddy School of Informatics, Computing, and Engineering at Indiana University, Bloomington, IN 47408, USA. · S. Kalhapure is an Independent Researcher at the time of work.

Robotics Mobile Robotics Mobile Manipulation Aerial Robotics

Open paper AskPDF arXiv

Exploring and inspecting Hidden Spaces, defined as environments whose entrances are accessible only to aerial robots but remain unexplored due to geometric constraints, limited flight time, and communication loss, remains a major challenge. We present miniUGV

_2

, a compact UAV-deployable tracked ground vehicle that extends UAV capabilities into confined environments. The system introduces dual articulated arms, integrated LiDAR and depth sensing, and modular electronics for enhanced autonomy. A novel tether module with an electro-permanent magnetic head enables safe deployment, retrieval, and optional detachment, thereby overcoming prior entanglement issues. Experiments demonstrate robust terrain navigation, self-righting, and manipulation of objects up to 3.5 kg, validating miniUGV

_2

as a versatile platform for hybrid aerial-ground robotics.

Abstract

Robotics

Explore similar work

MiniUGV2_22​: A Compact UAV-Deployable Tracked Ground Vehicle with Manipulation Capabilities

Towards Reliable Aerial Ground Vehicle Collaboration: An Integrated Planning and Autonomy Framework for Field Deployment

FC-Vision: Real-Time Visibility-Aware Replanning for Occlusion-Free Aerial Target Structure Scanning in Unknown Environments

MiniUGV $_2$ : A Compact UAV-Deployable Tracked Ground Vehicle with Manipulation Capabilities