Real-Time Tracing Of A Weld Line Using Artificial Neural Networks

Abstract

Robotic Manipulators are becoming increasingly popular nowadays with applications in almost every industry and production line. It is difficult but essential to create a common algorithm for the different types of manipulators present in todays market so that automation can be achieved at a faster rate. This paper aims to present a real time implementation of a method to control a Tal Brabo! Robotic manipulator to move along a given weld line in order to be utilized in factories for increasing production capacity and decreasing production time. The controller used here is provided by Trio, whose ActiveX component is interfaced to MATLAB. Images were captured to identify weld lines in every possible alignment to find points of interest and the neural network was trained in order to follow a given weld line once the work-piece was placed on the work-table.