Please use this identifier to cite or link to this item:
|Title:||Flattening and folding of crumpled clothes using robotic arms and stereovision||Authors:||Aditya, Sundaresan||Keywords:||Engineering::Mechanical engineering::Robots
Engineering::Computer science and engineering::Computing methodologies::Image processing and computer vision
|Issue Date:||2022||Publisher:||Nanyang Technological University||Source:||Aditya, S. (2022). Flattening and folding of crumpled clothes using robotic arms and stereovision. Final Year Project (FYP), Nanyang Technological University, Singapore. https://hdl.handle.net/10356/159154||Project:||C145||Abstract:||This report presents the design, development, and implementation of a theoretical framework, software architecture, and algorithms to autonomously bring a towel that is initially in a randomly crumpled state to a final folded state, using a stereocamera and industrial robotic arm. Unlike standard rigid bodies that may be recognized and manipulated by robots, a towel is a textile object that has unusual geometries that cannot be approximated by any standard geometric shapes to an acceptable extent. Further, towels are not rigid, and any manipulation of the towel completely changes the geometry of the towel completely. This report considers these factors that are unique to textile manipulation, and proposes methods to deal with the unusual geometries, deformabilites etc. of textile objects. In particular, a method to isolate local areas on a towel for feature recognition without the need for any modeling or representation of the towel is proposed; Motions used in manipulating the towel are designed to avoid contact with the towel where not necessary, as well as take advantage of particular textile properties such as stretching under the influence of gravity. A specialized gripper design is also proposed. Methods proposed in this report are novel in that this is the first time this particular task has been achieved with the use of only a single robotic arm, and relying purely on depth data, without any usage of 2D computer vision techniques. This enables these techniques to be replicated even with the use of hardware such as LiDAR scanners, which do not have 2D image capture capabilities. This report focuses specifically on stereovision and depth map based feature detection and localization including such features as ridges, occlusions, folds, highest points, and corners.||URI:||https://hdl.handle.net/10356/159154||Fulltext Permission:||restricted||Fulltext Availability:||With Fulltext|
|Appears in Collections:||MAE Student Reports (FYP/IA/PA/PI)|
Updated on Jun 24, 2022
Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.