Browse Prior Art Database

Machine Vision Calibration Technique for Vision-Aided Robotics

IP.com Disclosure Number: IPCOM000061315D
Original Publication Date: 1986-Jul-01
Included in the Prior Art Database: 2005-Mar-09
Document File: 3 page(s) / 16K

Publishing Venue

IBM

Related People

Colson, JC: AUTHOR

Abstract

A method of machine vision calibration is described for machine vision systems whose camera field of view resides in the working envelope of the manipulator. The resulting calibration technique is easy to use and adds no additional cost to the automation workcell. Machine vision applications in the past have typically been implemented as stand-alone devices. This required calibration fixtures to be made which were costly and prone to error if not used correctly. For industrial automation applications of machine vision, a high priority must be placed on ease of use and low cost. For robotic processes that use machine vision to determine position and orientation information, some form of calibration must be performed.

This text was extracted from a PDF file.
This is the abbreviated version, containing approximately 33% of the total text.

Page 1 of 3

Machine Vision Calibration Technique for Vision-Aided Robotics

A method of machine vision calibration is described for machine vision systems whose camera field of view resides in the working envelope of the manipulator. The resulting calibration technique is easy to use and adds no additional cost to the automation workcell. Machine vision applications in the past have typically been implemented as stand-alone devices. This required calibration fixtures to be made which were costly and prone to error if not used correctly. For industrial automation applications of machine vision, a high priority must be placed on ease of use and low cost. For robotic processes that use machine vision to determine position and orientation information, some form of calibration must be performed. The result of the calibration describes a relationship between a coordinate system fixed in the field of view of the camera and a coordinate system which is appropriate for the manipulator. This information will subsequently be used to transform position and orientation information that is determined in the camera's coordinate system to the coordinate system that is appropriate for the manipulator. When using vision information in conjunction with the manipulator for goal information, this type of information is necessary for all camera-mounting locations, whether the camera is mounted rigidly in space or mounted on the manipulator itself. The technique for determining this information is described below. Machine vision calibration has historically been performed using precision fixtures which are both expensive to generate and delicate to handle. Both of these attributes are nemeses on the manufacturing floor; therefore, a calibration technique which was both inexpensive to implement and required a minimum of hardware, and was easy to use was developed. There are two distinct cases for machine vision calibration for manipulator guidance. These cases are 1) camera fixed in space and 2) camera mounted to a link of the manipulator. Instead of using a precise calibration fixture, which must be placed in a known position and orientation, fixed camera calibration requires that the general position of two objects be determined by both the camera and the manipulator. This is accomplished by first allowing the user to define a particular object in an image. Then, several images are acquired and the center of the object, in pixels, is determined statistically. The user is then prompted to define a second object in the same field of view of the camera and, again, statistics are used to determine its center location, in pixels, in the field of view. Next, the user is prompted to guide the manipulator so that a particular feature of the first object that was defined is a known distance from an arbitrary point on the manipulator. This manipulator position is recorded. The user is then prompted to guide the manipulator so that a particular feature of the second object, w...