Learning based robotic grasping

Nowadays, with the rapid development of artificial intelligence, machine learning has made great strides in making computers behave more intelligently. In this context, machine learning has been applied to robots to make them work in a more reasonable way. This has enabled rapid advances in robotic...

Full description

Saved in:
Bibliographic Details
Main Author: Wang, Chongyu
Other Authors: Cheah Chien Chern
Format: Thesis-Master by Coursework
Language:English
Published: Nanyang Technological University 2022
Subjects:
Online Access:https://hdl.handle.net/10356/158933
Tags: Add Tag
No Tags, Be the first to tag this record!
Institution: Nanyang Technological University
Language: English
id sg-ntu-dr.10356-158933
record_format dspace
spelling sg-ntu-dr.10356-1589332023-07-04T17:47:54Z Learning based robotic grasping Wang, Chongyu Cheah Chien Chern School of Electrical and Electronic Engineering ECCCheah@ntu.edu.sg Engineering::Electrical and electronic engineering Nowadays, with the rapid development of artificial intelligence, machine learning has made great strides in making computers behave more intelligently. In this context, machine learning has been applied to robots to make them work in a more reasonable way. This has enabled rapid advances in robotic work planning. At the same time, with the transformation and upgrading of manufacturing, intelligent industrial robots with vision systems are widely used in modern factories. In order to deploy robots rapidly in flexible manufacturing and make robots operate objects accurately, this dissertation studied a robot that applies machine learning to robot object detecting. The content and results of this report mainly include the following aspects: Firstly, this report presents a grasping robot that is able to detect objects accurately by using a model trained by YOLO (You only look once). In order to make the robot grasp objects while avoiding the sharp points on their edge, YOLO is also used to train the model to detect the sharp points on objects' edges. Training results show that the mean average precision of detecting different objects in the lab can reach as high as 91.8\%. Secondly, this report presents a method to calculate the position of the robot tool for grasping objects. The experimental results show that the robot is able to grasp the objects accurately and steadily with the detecting outcomes and robot motion commands. Master of Science (Computer Control and Automation) 2022-06-02T11:50:29Z 2022-06-02T11:50:29Z 2022 Thesis-Master by Coursework Wang, C. (2022). Learning based robotic grasping. Master's thesis, Nanyang Technological University, Singapore. https://hdl.handle.net/10356/158933 https://hdl.handle.net/10356/158933 en application/pdf Nanyang Technological University
institution Nanyang Technological University
building NTU Library
continent Asia
country Singapore
Singapore
content_provider NTU Library
collection DR-NTU
language English
topic Engineering::Electrical and electronic engineering
spellingShingle Engineering::Electrical and electronic engineering
Wang, Chongyu
Learning based robotic grasping
description Nowadays, with the rapid development of artificial intelligence, machine learning has made great strides in making computers behave more intelligently. In this context, machine learning has been applied to robots to make them work in a more reasonable way. This has enabled rapid advances in robotic work planning. At the same time, with the transformation and upgrading of manufacturing, intelligent industrial robots with vision systems are widely used in modern factories. In order to deploy robots rapidly in flexible manufacturing and make robots operate objects accurately, this dissertation studied a robot that applies machine learning to robot object detecting. The content and results of this report mainly include the following aspects: Firstly, this report presents a grasping robot that is able to detect objects accurately by using a model trained by YOLO (You only look once). In order to make the robot grasp objects while avoiding the sharp points on their edge, YOLO is also used to train the model to detect the sharp points on objects' edges. Training results show that the mean average precision of detecting different objects in the lab can reach as high as 91.8\%. Secondly, this report presents a method to calculate the position of the robot tool for grasping objects. The experimental results show that the robot is able to grasp the objects accurately and steadily with the detecting outcomes and robot motion commands.
author2 Cheah Chien Chern
author_facet Cheah Chien Chern
Wang, Chongyu
format Thesis-Master by Coursework
author Wang, Chongyu
author_sort Wang, Chongyu
title Learning based robotic grasping
title_short Learning based robotic grasping
title_full Learning based robotic grasping
title_fullStr Learning based robotic grasping
title_full_unstemmed Learning based robotic grasping
title_sort learning based robotic grasping
publisher Nanyang Technological University
publishDate 2022
url https://hdl.handle.net/10356/158933
_version_ 1772827168247644160