Honors College Thesis
 

Improve the Grasping Performance by Analyzing Target Objects with Computer Vision and Deep Learning Algorithm

Public Deposited

Contenu téléchargeable

Télécharger le fichier PDF
https://ir.library.oregonstate.edu/concern/honors_college_theses/fn107586p

Descriptions

Attribute NameValues
Creator
Abstract
  • Recently there has been a large amount of research into basic human movement as a baseline for robotic motion. Robotic grasping is a challenging problem for a number of reasons. One of which is that it is difficult to accurately know the interaction between the hand and the object, especially for tasks where the arm is moving and as a result a fixed camera is not convenient. To alleviate the difficulty, there are lots of studies that focus on planning ideal pre-grasp pose by mimicking human actions on robots. This thesis developed and validated computer vision and deep learning framework to provide object detection, object size, and location based on the camera and ultrasonic sensor. In this thesis, there are two main methodology. First, the deep learning framework, are called ‘YOLOv3’. It allows us to detect objects with the real-world image training dataset. Second, this thesis presents the utilization of OpenCV for computer vision because it allows measuring the size of detected objects and distance from the camera to the object. After tests and trials, we found some problems like the increment error range by camera calibration and the optimization of the fixed camera position. However, our methods can accurately determine the location of the object in XYZ space relative to the hand accurate within about 7% of error rate. Future work can extend towards using other deep learning frameworks and change segmentation methods with other vision sensors like depth camera and LIDAR. Although the transition method could require high computational cost, we expect it will improve the accuracy of object detection with more objects.
Contributor
Resource Type
Date Issued
Degree Level
Degree Name
Degree Field
Degree Grantor
Commencement Year
Advisor
Committee Member
Non-Academic Affiliation
Déclaration de droits
Publisher
Peer Reviewed
Language

Des relations

Parents:

This work has no parents.

Dans Collection:

Articles

La vignette Titre Date de téléchargement Visibilité actes
file details JunhyeokJeong2021.pdf 2021-05-13 Public Télécharger