Faib, Semyon (2023) Depth estimation from a monocular camera and neural network PRE - Research Project, ENSTA.

Full text not available from this repository.

Abstract

In the modern age of robotics, precise perception of the environment stands as a cornerstone for safe and effective operations. This work delves deep into the integration of the MonoDepth2 model with the Robot Operating System (ROS) framework, aiming to imbue robots with the capability to discern depth from single images. Utilizing Clearpath Gazebo Worlds, diverse simulated environments acted as the backdrop for testing and refining this integration. While initial results are promising, they also highlight the disparities between images from common databases and those captured by a mobile robot, emphasizing the need for model fine-tuning and optimization. This exploration serves as a foundation, revealing pathways for future advancements in robotic depth perception.

Item Type:Thesis (PRE - Research Project)
Subjects:Information and Communication Sciences and Technologies
ID Code:9725
Deposited By:Semyon FAIB
Deposited On:28 nov. 2024 15:03
Dernière modification:28 nov. 2024 15:03

Repository Staff Only: item control page