Identifying Abnormalities in Indoor Environments using Obstacle Detection Programming

Jennings, Nicholas (2021) Identifying Abnormalities in Indoor Environments using Obstacle Detection Programming. [USQ Project]

Text (Project)
JENNINGS Nicholas dissertation_redacted.pdf

Download (4MB) | Preview


In autonomous mechatronic systems, object detection is a vital functionality which allows the system to process information about it’s surrounding environment and interpret it in order to act without the need for human interference. In order to procure information regarding it’s environment, a system must be equipped with sensors appropriate for the conditions in which it is operating, which are also capable of perceiving the required types of object for the current application. Despite the numerous existing systems tailored to the environment, in applications requiring autonomous navigation in an indoor environment, little research has been conducted observing the accuracy of object detection when applied to small-scale low-lying obstacles. Of the common forms of sensors employed in object detection applications, being ultrasonic, LiDAR and both monocular and stereo vision camera configurations, the sensors system most appropriate for the detection of small-scale low-lying obstacles is a stereo vision camera. A sample stereo vision system was created and subsequently evaluated in a series of tests in order to determine the accuracy of it’s depth estimation when directed at applicable obstacles likely to be present within an indoor environment. The sample system was prepared with hardware consisting of a pair of identical webcams fixed a set distance apart, and software developed utilising MATLAB, governing the calibration of the stereo camera, the image segmentation and the depth estimation of identified obstacles. First tested within a controlled environment, the system was determined to have a baseline error of less than 4% when detecting more traditional obstacles. When applied to the analysis of small-scale low-lying obstacles, factors such as the size of flat obstacles, the angle at which they were orientated and their positioning within the camera’s visible area were determined to impact the accuracy of the depth estimation.

Statistics for USQ ePrint 51823
Statistics for this ePrint Item
Item Type: USQ Project
Item Status: Live Archive
Faculty/School / Institute/Centre: Historic - Faculty of Health, Engineering and Sciences - School of Mechanical and Electrical Engineering (1 Jul 2013 - 31 Dec 2021)
Supervisors: Low, Tobias
Qualification: Bachelor of Engineering (Honours) (Mechatronic)
Date Deposited: 03 Jan 2023 02:37
Last Modified: 26 Jun 2023 01:31
Uncontrolled Keywords: autonomous, mechatronic, indoor, obstacle, detection, sensors, LiDAR, stereo vision

Actions (login required)

View Item Archive Repository Staff Only