Konrad, Anna
(2023)
Learning to grasp unknown objects in domestic
environments.
PhD thesis, National University of Ireland Maynooth.
Abstract
For robots to become valuable assistants in real-world scenarios, they must effectively
manipulate unknown objects in complex environments, such as homes with
multiple objects on shelves or desks. A critical aspect of manipulating objects is
grasping them with the robot’s end-effector. The key challenge for grasping an
unknown object lies in determining how to position and orientate the end-effector
relative to the object. While prior research has predominantly focused on tabletop
scenarios with fixed camera-workspace transforms, in this thesis, we extend
robotic grasping to domestic settings.
This thesis makes three principal contributions in this area: (i) extending a depthimage-
based grasp quality prediction method to accommodate 6-DoF grasps and
versatile camera poses, (ii) creating a depth-image-based grasp proposal method
tailored for 6-DoF grasps and unknown camera-workspace transforms, and (iii) applying
such a grasp proposal method in complex, domestic environments. For application
in domestic environments, we develop a simulation pipeline specifically
designed for training and evaluating grasp proposal models in domestic settings.
Quantitative experiments demonstrate competitive or superior performance of our
approach when compared with widely used techniques across a number of challenging
simulated experiments. Additionally, real-world experiments with a mobile
manipulator demonstrate successful grasping of unknown objects positioned
on tables and shelves in unknown poses. We make our code and pre-trained models
available for fellow researchers, facilitating the direct application of our models
when attempting to grasp unknown objects with mobile manipulators in domestic
environments.
Item Type: |
Thesis
(PhD)
|
Keywords: |
Learning; grasp unknown objects; domestic environments; |
Academic Unit: |
Faculty of Science and Engineering > Research Institutes > Hamilton Institute |
Item ID: |
18342 |
Depositing User: |
IR eTheses
|
Date Deposited: |
02 Apr 2024 10:44 |
URI: |
|
Use Licence: |
This item is available under a Creative Commons Attribution Non Commercial Share Alike Licence (CC BY-NC-SA). Details of this licence are available
here |
Repository Staff Only(login required)
|
Item control page |
Downloads per month over past year
Origin of downloads