A PROGRAMMING ENVIRONMENT FOR SENSOR GUIDED ASSEMBLY (ROBOTICS)

ROBERT JAMES SAFRANEK, Purdue University

Abstract

This report examines the hardware and software requirements for a flexible manufacturing cell designed for small batch assembly. The cell configuration could include one or more manipulators, along with extensive sensory support. Types of sensory input that should be supported include 2D and 3D vision. The programming environment is designed to be used as a research tool for sensor guided assembly. One of the major goals of the system design is to make manipulators usable by researchers who do not have an extensive background in robotics. Until recently, most robot programming languages have focused almost exclusively on manipulation capabilities while providing little or no support for the integration of sensor systems. This taxonomy isolates these functions into independent modules. These modules consist of a Supervisor, Motion Control Unit, Sensory Processor, Global Knowledge Base, and Current World State. Communication between modules is via an extensible set of "messages." A configuration of this type allows each module to be implemented in the most appropriate language/hardware combination. To show the applicability of this methodology, a Motion Controller for the Cincinnati Milacron T3-726 industrial robot has been implemented. This module has been successfully utilized for several experiments in the Robot Vision Lab. Next, a means of representing and executing manipulator plans is presented. the approach allows for sensory feedback to guide the course of execution. An example of message flow through the entire system is also given at this time. A major problem in model based systems, such as this, is confirming the validity of the systems internal model. The design of an object position verification system to support the Current World State is described. This system utilizes an evidential feature detector utilizing the Dempster-Schafer theory of inexact reasoning. To support this work, the Binary Frame of Discernment is introduced and several of its properties outlined. In addition, the systems performance is examined in relation to more traditional reasoning and feature extraction methods.

Degree

Ph.D.

Subject Area

Electrical engineering

Off-Campus Purdue Users:
To access this dissertation, please log in to our
proxy server
.

Share

COinS