Please use this identifier to cite or link to this item: https://hdl.handle.net/10356/18778
Title: Vision system for object recognition and staircase climbing
Authors: Tran, Quang Hieu.
Keywords: DRNTU::Engineering::Mechanical engineering::Robots
Issue Date: 2008
Abstract: The ability of a robot to see is important, especially for a mobile robot. On such robots, the vision sensor provides much more information than other types of sensor. Associated with the vision sensor is the image processing that is required before useful information can be derived. Computer vision can be used to accomplish various types of tasks, such as localization and mapping, obstacle avoidance, object detection, tracking, etc. In January 2007, the Defense Science & Technology Agency (DSTA) announced a robot competition, the TechX Challenge. To win the grand prize of 1 million Singapore dollars, the robot has to autonomously navigate throughout different types of environment and execute various types of tasks. The conditions and tasks for the robot are what can be expected by an autonomous mobile robot working in an urban environment. Among the competition tasks are patterns or targets recognition and staircase climbing. This dissertation focuses on building a working vision system for a mobile robot to accomplish two tasks: to recognize flat (2-dimensional) pattern or solid (3-dimensional) object that the system has been trained with, and to aid the robot to overcome a straight staircase. The reason for building a vision system is obvious for the first task, as patterns recognition is difficult or impossible without the help of vision sensors. While the second task – staircase climbing – can often be accomplished using other types of sensors, such as bumper sensor or range finders, an additional requirement is that there could be no wall on the two sides of the staircase. To overcome this, a dedicated sensor specially designed for detecting the two sides of the staircase is required. Another option is to take advantage of the existing vision sensor to position the robot and to climb the staircase.
URI: http://hdl.handle.net/10356/18778
Fulltext Permission: restricted
Fulltext Availability: With Fulltext
Appears in Collections:MAE Theses

Files in This Item:
File Description SizeFormat 
TranQuangHieu08.pdf
  Restricted Access
11.31 MBAdobe PDFView/Open

Page view(s) 10

343
checked on Oct 24, 2020

Download(s) 10

27
checked on Oct 24, 2020

Google ScholarTM

Check

Items in DR-NTU are protected by copyright, with all rights reserved, unless otherwise indicated.