Article 2025 · 0 atıf

Deep learning enhanced monocular visual odometry: Advancements in fusion mechanisms and training strategies

Özet

Recent advances in deep learning have revolutionized robotic applications such as 3D mapping, visual navigation and autonomous control. Monocular Visual Odometry (MVO) represents a critical advancement in autonomous systems, particularly drones, utilizing single-camera setups to navigate complex environments effectively. This review explores MVO's evolution from traditional methods to its integration with cutting edge technologies like deep learning and semantic understanding. In this study, we explore the latest training strategies, innovations in model architecture, and advanced fusion techniques used in hybrid models that combine depth and semantic information. A comprehensive literature review traces the evolution of MVO techniques, highlighting key datasets and performance metrics. Section 2 outlines the problem, while Section 3 reviews the studies, charting the evolution of MVO techniques predating the advent of deep learning. Section 4 details the methodology, focusing on cutting-edge training strategies, advancements in architectural designs, and fusion techniques in hybrid models integrating depth and semantic information. Finally, Section 5 summarizes findings, discusses implications, and suggests future research directions.

Anahtar Kelimeler

Monocular visual odometry Deep learning Unsupervised learning Visual navigation Hybrid models Synthesis mechanism Methodology review DEPTH LOCALIZATION

WoS Kategorileri

Computer Science, Artificial Intelligence Computer Science, Software Engineering Computer Science, Theory & Methods Engineering, Electrical & Electronic Optics

Künye

Dergi: IMAGE AND VISION COMPUTING
Yıl: 2025
Cilt: 162
ISSN: 0262-8856
eISSN: 1872-8138
DOI: 10.1016/j.imavis.2025.105732
WoS UID: WOS:001583966700001

Yayını aç (DOI)