Gerrit Jordaan

Automatic Guided Vehicles (AGVs) are navigated utilising multiple types of sensors for detecting the environment. In this investigation such sensors are replaced and/or minimized by the use of a single omnidirectional camera picture stream. An area of interest is extracted, and by using image processing the vehicle is navigated on a set path. Reconfigurability is added to the route layout by signs incorporated in the navigation process. The result is the possible manipulation of a number of AGVs, each on its own designated colour-signed path. This route is reconfigurable by the operator with no programming alteration or intervention. A low resolution camera and a Matlab^® software development platform are utilised. The use of Matlab^® lends itself to speedy evaluation and implementation of image processing options on the AGV, but its functioning in such an environment needs to be assessed.

Three-dimensional vision and photography has been used in many disciplines for a long time. However, modern technological developments and increasing levels of industrial automation are making increasing demands for the development and implementation of new, innovative support systems for, inter alia, quality assurance and component identification purposes. Three-dimensional photography is often the preferred way of addressing this need. The system as described represents an experimental photographic system for the acquisition of three-dimensional images of components moving along a conveyor belt system as part of an automated material-handling system in a pseudo-industrial environment. The photographic system consists of only one digital camera and a scanning laser beam. Preliminary results indicate that fairly accurate images can be acquired using such a system, whilst good rendering can result from careful manipulation of the data of photographed components.

There is a quest in the field of robotics for an automated dynamic gripper with characteristics as close as possible to that of the human hand. In this paper an automated robot gripper is described that achieves preset levels of gripping force for a certain task. It utilises dynamic gripping through the use of piezoelectric sensors. Analysis of the output pulse from the piezoelectric film indicates the velocity of gripping, deceleration and the total force the gripper is inducing on the gripped object. Piezoelectric film is chosen for its superior sensitivity and structural simplicity over the typical metal-foil strain gauges. The paper shows that, compared to three dimensional analysis, object handling using dynamic fingertip force sensing is a cheaper and more accurate solution.

This paper gives an overview of a material-handling facility incorporating machine vision - as a component of a research platform for the investigation of automated industrial processes. The aim with the research initiative is the development of relevant skills and expertise in automated material-handling installations with SCADA facilities in a pseudo-industrial environment