Title:
Deep Supervised Light Weight Networks for Visual Odometry Estimation

Author:

Supervisor:

AP Ts Ir Dr Irraivan Elamvazuthi Dr Lila Iznita Izhar

Department:

Civil & Environmental Engineering

Year:

2024

URL:

Abstract:

Visual odometry involves the computation of an autonomous agent's pose by analyzing images. Utilizing deep learning to estimate visual odometry enhances robustness against noise. Training neural netwo...
View more

Multimedia Content:

01

Methodology

1. Research Design The study follows an experimental research design that integrates supervised deep learning techniques with lightweight convolutional neural networks (CNNs) to improve visual odomet...
View more
02

Citation Analysis

1. Sensors (MDPI) Scope: Covers sensor technologies, computer vision, robotics, and autonomous systems. Relevance: Your work on lightweight deep learning networks for visual odometry fits well und...
View more
03

Impact

1. Societal Impact • Advancement in Autonomous Mobility: The thesis contributes to safer and more efficient autonomous vehicles by improving how visual odometry systems estimate camera motion in r...
View more
04

Benefits to Society

1. Societal Benefits Safer Autonomous Systems: The proposed lightweight visual odometry networks enhance the accuracy and stability of autonomous vehicles and drones, leading to safer navigation i...
View more
05

Applications

1. Autonomous Vehicles and Driver Assistance Systems Application: Real-time visual odometry for localization and navigation. Use Case: Self-driving cars and advanced driver-assistance systems (ADA...
View more
06

Infographic

Summary of this research
Infographic Image
07

Patents / Products

1. Lightweight Visual Odometry Neural Engine Product Type: Deep learning software module or embedded AI chip architecture Use Case: Integrate into autonomous vehicles, drones, or robots to perform...
View more
08

Experts

1. Raúl Mur-Artal Institution: Technical University of Munich (TUM), Germany Contribution: Creator of ORB-SLAM and co-developer of ORB-SLAM2, two of the most influential open-source visual odometr...
View more