Author: Sai Juturu
Juturu, Sai, 2022 Investigating Robotic Designs to address the ever-increasing space debris, Flinders University, College of Science and Engineering
Terms of Use: This electronic version is (or will be) made publicly available by Flinders University in accordance with its open access policy for student theses. Copyright in this thesis remains with the author. You may use this material for uses permitted under the Copyright Act 1968. If you are the owner of any included third party copyright material and/or you believe that any material has been made available without permission of the copyright owner please contact copyright@flinders.edu.au with the details.
Since the dawn of space exploration in 1957, numerous structures such satellites and rockets, have been situated in Earth’s orbit and space. Due to a lack of a system that deals with the disposal of now defunct structures (or their collision resultant fragments), the amount of debris in space is estimated to have reached the Kessler benchmark. Thus, there is a need to conceptualize systems to improve the safety of all personnel and property in space. The thesis aims to evaluate the viability of image processing and machine learning techniques in the field of debris identification. The kernel-based program (image processing) was capable of detecting space debris, however the quality of the output was degraded due to preset baseline assumptions. Other factors that determined this program’s inability to appropriately function in space was its dependency on color/pixel values; variables that cannot be replicated in actual space, more so since this code is tested upon clips from the movie "Gravity". The project then progresses to validate the applicability of machine learning in similar scenarios. Since the model’s training and testing phases were comprised of only 200 images (along with being synthetic), the precision of 0.6 and recall of 0.67 indicate that the model developed for this thesis is not realistically feasible. However, the research did indicate that machine learning with better resources could be a potential solution working towards space debris detection. Subsequently, the thesis also dives into the field of robotic design, 3D modelling a hypothetical robot, named the WOMBAT, that would be theoretically capable of capturing/collecting space debris through the equipped net/sheet. Finally, the discussion gains a marketing perspective on introducing such a product, and the resistance it could receive due to its potential ability to gain access to inter-country security and surveillance data.
Keywords: satellites, rockets, space, robot, kernel-based program, image processing, machine learning, precision, recall, space debris
Subject: Engineering thesis
Thesis type: Masters
Completed: 2022
School: College of Science and Engineering
Supervisor: Mr Joseph Pepe Velasquez