USP Theses Collection


Menu



A B C D-E F G H-J K L M N O-P Q-R S T U-V W-Y
close this section of the library Sharan, Roneel Vikash.


View the PDF document A vision-based pick-and-place robot.
Author:Sharan, Roneel Vikash.
Institution: University of the South Pacific.
Award: M.Sc.
Date: 2006.
Call No.: pac In Process
BRN: 1047546
Copyright:Under 10% of this thesis may be copied without the authors written permission

Abstract: Pick-and-place robots have their applications limited in a system where it is fed with all the information regarding its target. In other words, the robot is given a complete specification of each motion for a manipulation. This work considers the approach taken in automating the manipulation process of an in-house designed pick-and-place robot. It reports the studies of flexible manufacturing concepts using a combination of vision and motion. The arm links of the five degree-of-freedom (DOF) pick-and-place robot are actuated using a combination of stepper and DC motors while a force sensor and a limit switch act as feedback devices. Arm links are controlled based on calibration with respect to the controlling motors while the sensors handle the opening and closing of the gripper. In addition, a graphical user interface (GUI) acts as the interface between the user and the system such that the work-piece to be manipulated can be specified. A PC is used for vision processing and motion planning of the robot. Based on the planned motion, the PC then communicates through the enhanced parallel port (EPP) to a PIC microcontroller which in turn controls the robot in a sequential manner. The error of the joints controlled by stepper motors is found to be constant, unlike the linear relationship for the DC motor, with respect to the angular movement. Maximum error was obtained for the elbow joint, at -2.45°, in one of its two operation regions. Also, the absolute maximum error of the end-of-arm-tool (EOAT) of the robot around its operation area is (4.24, 4.06, 3.60) mm for the x,y, and z directions respectively. The development of a vision system is also presented that automates the work-piece recognition and localization process for the pick-and-place robot. Work-pieces, placed on the two-dimensional (2D) work-plane of the robot, are differentiated with respect to their shape and color. The location of a work-piece is then utilized to pick it and then placing it a predefined position. Two methods were tried each for shape and color recognition. These are the geometric features of roundness and radius ratio, combined to form a 2D feature vector, and corner detection for shape recognition and the RGB (Red, Green, and Blue) and the HSV (Hue, Saturation, and Value) color spaces for color recognition. The latter methods for each were utilized in the final vision system since they produced total success in the recognition process. This work has been carried out as part of development of a smart flexible manufacturing system. The pick-and-place robot is primarily employed for transfer of work-pieces to/from automatic guided vehicles, used for carrying the work-pieces around in the work-cell, from/to the different work-stations, mainly for drilling and milling. Rather than considering different methods of implementation of the whole system for this work, in-house developed technologies over the past years has been preferred. This includes the choice of materials in the mechanical and electronics design of the robot along with the overall architecture of the system such as parallel communication between the master and slave units. Similarly, preference has been given to an in-house designed vision system rather than an off the shelf package so that full knowledge can be gained about its operation. In the case of shape and color recognition of the work-pieces, the more natural methods have been preferred such as the number of corners for shape recognition and the HSV color space for color recognition. These are discussed in detail in the following sections.
Disclaimer & Copyright  l  Contact Us  l  
© Copyright 2004 - 2018. All Rights Reserved.
USP Library
The University of the South Pacific
Laucala Campus, Suva, Fiji
Tel: +679 323 1000