.png){kind=icon}
.png){kind=icon}
Return
SVE: Stable variational encoder for end-to-end monocular navigation of unmanned aerial vehicles
.png){kind=icon}
OA
AI
X
J
Z
G
DOI：10.1016/j.eswa.2026.133887.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
• Introduces the stable variational encoder for end-to-end monocular UAV navigation. • Replaces unstable VAE sampling with a trainable deviation embedding for stability. • Uses a decoder-only architecture, applying variation directly to original images. • Achieves superior navigation in high-difficulty scenarios and diverse weather. • Validates practical sim-to-real transfer through real-world experiments.
Keywords:
Unmanned aerial vehicle (UAV)
Aerial navigation
Stable variational encoder (SVE)
Variational autoencoder (VAE)
Imitation learning (IL)
Journal
IF:
7.5
Papers:
2.9W
Citations:
10.2W
Organization
No organization information available
Cited Papers