Document Type

Thesis

Date of Award

1-31-1988

Degree Name

Master of Science in Mechanical Engineering - (M.S.)

Department

Mechanical Engineering

First Advisor

Rajesh N. Dave

Second Advisor

Bernard Koplik

Third Advisor

Rong-Yaw Chen

Abstract

There has been a continued. interest in utilizing machine vision systems for robot guidance applications. Commercial vision systems can be programmed to give information about the parts in the robot work area and this information can be used by the robot computer to guide the robot. In most applications, however, the vision information cannot be easily integrated with the robot control language. This thesis considers implementation of part identification routines directly into a robot guidance language. This makes all the complicated machine vision algorithms transparent to the user and allows him to treat higher level vision functions as an extension of the robot control routines.

First, a textual robot guidance language is developed for a small educational robot by creating robot motion functions in 'C' language. For the purpose of guidance through vision, a set of routines is developed to extract information from a video camera. The vision system used here is a low cost micro-computer based unit with a video camera, a digitizer and a frame buffer. The image captured by the video camera is digitized and stored in the frame buffer. Low level processing is applied and the image is segmented, and the edges of the objects are detected. Various features such as perimeter, area, number of holes as well as the features of the holes can be extracted from the image using the algorithms described here. These features are used as a database to identify parts.

Different objects can be 'taught' to the system by first placing examples of these objects into the image field. The system automatically creates a feature set for each object and a database is created for all the objects. Once all the parts are taught to the system, it is able to recognize any part placed in the field as long as the parts are not overlapping. Now the user can access location and orientation of the parts through simple 'C' function calls and thus very easily implement the vision information into robot guidance language, which is also composed of functions written in 'C' language.

Share

COinS
 
 

To view the content in your browser, please download Adobe Reader or, alternately,
you may Download the file to your hard drive.

NOTE: The latest versions of Adobe Reader do not support viewing PDF files within Firefox on Mac OS and if you are using a modern (Intel) Mac, there is no official plugin for viewing PDF files within the browser window.