Evaluation of Monocular SLAM Systems on High-Altitude Nadir UAV Footage

Gašper Spagnolo; Matej Dobrevski; Danijel Skočaj

cs.ROAug 19, 2026

Evaluation of Monocular SLAM Systems on High-Altitude Nadir UAV Footage

Authors: Gašper Spagnolo, Matej Dobrevski, Danijel Skočaj

Organizations: Faculty of Computer and Information Science, University of Ljubljana, Večna pot 113, Ljubljana, Slovenia

Abstract

Aerial nadir video combines weak geometric constraints with severe perceptual aliasing, making it a difficult regime for monocular SLAM. We benchmark five monocular SLAM systems on local UAV flights, synthetic city-scale imagery, and long-range aerial sequences. To isolate visual performance, we provide no inertial or GNSS aiding. Performance varies strongly with environment and trajectory scale: MASt3R-SLAM achieves the lowest mean horizontal MAE on the five DJI flights (0.53% of reference path length), whereas no system consistently preserves global trajectory shape on the long GES and ALTO sequences. Overall, DROID-SLAM performs best, averaging 2.88% of reference path length across completed runs. Vertical position remains poor, and large-area trajectories remain highly distorted despite loop-closure capability. Current monocular SLAM methods are by themselves therefore insufficient for reliable visual-only aerial navigation.

Figures & tables

Figure 1: FRI sequence of our DJI dataset: nadir frames composited over the flight area, with the flight trajectory shown in green.

Figure 2: Representative aligned trajectories for two contrasting sequences from each dataset. Note the different axis scales: DJI and MARS-LVIG sequences span hundreds of meters, whereas GES and ALTO span several kilometers.

DJI
Sequence	DPVO	DROID	MASt3R	ORB3	VGGT2
FRI-110m	10.47	0.16	0.70	0.22	1.09
Šempeter-1	4.54	0.20	1.03	2.25	0.39
Šempeter-2	8.71	1.92	0.34	7.74	8.38
Šempeter-3	1.96	0.25	0.29	0.21	11.13
Šempeter-4	3.09	0.37	0.29	1.57	10.74

Table 1: Per-sequence horizontal MAE divided by reference path length (%). The best result in each row is shown in bold. “–” marks a failed run.

Figure 3: Representative loop-closure failures on GES. In (a,b), ground truth and image overlap confirm true revisits that VGGT-SLAM 2.0 never proposes for attention verification. In (c,d), MASt3R-SLAM accepts spatially inconsistent pairs. Both are sequential candidates, as indicated by their small frame-index gaps. $Q$ and $R$ denote query and retrieved frames, $\Delta f$ their frame-index separation, and $d_{\mathrm{GT}}$ their ground-truth distance. For MASt3R-SLAM, $C$ is the candidate identifier and $m$ the geometric match fraction relative to its threshold. Red borders mark erroneous outcomes.

	DPVO	DROID	MASt3R	ORB3	VGGT2
DJI	5.75 $\pm$ 3.67	0.58 $\pm$ 0.75	0.53 $\pm$ 0.33	2.40 $\pm$ 3.11	6.34 $\pm$ 5.23
GES	1.85 $\pm$ 1.98	1.82 $\pm$ 0.20	3.71 $\pm$ 2.34	4.20 $\pm$ 1.39	3.89 $\pm$ 1.30
MARS-LVIG	2.24 $\pm$ 1.75	2.38 $\pm$ 2.32	4.23 $\pm$ 2.23	3.09 $\pm$ 2.32	3.08 $\pm$ 1.59
ALTO	8.57 $\pm$ 11.02	8.03 $\pm$ 9.85	18.36 $\pm$ 10.11	6.61 $\pm$ 7.29	8.18 $\pm$ 2.79
ALL	3.98 $\pm$ 5.15	2.88 $\pm$ 4.62	5.24 $\pm$ 6.59	3.67 $\pm$ 3.66	4.71 $\pm$ 3.36

Table 2: Horizontal MAE divided by reference path length (%), reported as mean

\pm

sample SD over completed runs.

Visual SLAM

Visual simultaneous localization and mapping (SLAM) estimates camera motion while building maps from images for navigation and scene understanding.

See all 108 papers

Robust Visual SLAM for UAV Navigation in GPS-Denied and Degraded Environments: A Multi-Paradigm Evaluation and Deployment Study
May 5, 2026Prasoon Kumar, Akshay Deepak, Sandeep KumarUAV Navigation GNSS-Denied Localization
National Institute of Technology Patna, Bihar, India · Central Research Laboratory, Bharat Electronics Ltd., Ghaziabad, Uttar Pradesh, India
AERO-VIS: Asynchronous Event-based Real-time Onboard Visual-Inertial SLAM
May 8, 2026Yannick Burkhardt, Sebastián Barbas Laina, Simon Boche +2Event-Based Vision UAV Navigation
Mobile Robotics Lab, ETH Zürich · Technical University of Munich · Munich Center for Machine Lerning (MCML) +1
PanoAir: A Panoramic Visual-Inertial SLAM with Cross-Time Real-World UAV Dataset
Apr 1, 2026Yiyang Wu, Xiaohu Zhang, Yanjin Du +5Visual-Inertial SLAM UAV Navigation
School of Aeronautics and Astronautics, Sun Yat-sen University, Guangzhou 510275, China.

May 5, 2026J/K move · Enter open · S save

Robust Visual SLAM for UAV Navigation in GPS-Denied and Degraded Environments: A Multi-Paradigm Evaluation and Deployment Study

Prasoon Kumar, Akshay Deepak, Sandeep Kumar

National Institute of Technology Patna, Bihar, India · Central Research Laboratory, Bharat Electronics Ltd., Ghaziabad, Uttar Pradesh, India

UAV Navigation GNSS-Denied Localization Visual SLAM

Open paper AskPDF arXiv

Reliable localization in GPS-denied, visually degraded environments is critical for autonomous UAV opera- tions. This paper presents a systematic comparative evaluation of five V-SLAM systems ORB-SLAM3, DPVO, DROID-SLAM, DUSt3R, and MASt3R spanning classical, deep learning, recurrent, and Vision Transformer (ViT) paradigms. Experiments are conducted on curated sequences from four public benchmarks (TUM RGB-D, EuRoC MAV, UMA-VI, SubT-MRS) and a custom monocular indoor dataset under five controlled degradation conditions (normal, low light, dust haze, motion blur, and combined), with sub-millimeter Vicon ground truth. Results show that ORB-SLAM3 fails critically under severe degradation (62.4% overall TSR; 0% under dense haze), while learning-based methods remain robust: MASt3R achieves the lowest degraded ATE (0.027 m) and DUSt3R the highest tracking success (96.5%). DPVO offers the best efficiency robustness trade-off (18.6 FPS, 3.1 GB GPU memory, 86.1% TSR), making it the preferred choice for memory-constrained embedded platforms. Embedded deployment analysis across NVIDIA Jetson platforms provides actionable guidelines for SLAM selection under SWaP-constrained UAV scenarios.

Abstract

Visual SLAM

Explore similar work

Robust Visual SLAM for UAV Navigation in GPS-Denied and Degraded Environments: A Multi-Paradigm Evaluation and Deployment Study

AERO-VIS: Asynchronous Event-based Real-time Onboard Visual-Inertial SLAM

PanoAir: A Panoramic Visual-Inertial SLAM with Cross-Time Real-World UAV Dataset