Image analysis – Applications – Robotics
Reexamination Certificate
1999-07-06
2002-12-03
Tran, Phuoc (Department: 2621)
Image analysis
Applications
Robotics
C382S151000, C382S209000, C382S217000, C382S286000
Reexamination Certificate
active
06490369
ABSTRACT:
BACKGROUND OF THE INVENTION
1) Technical Field
The subject invention relates to a robot manipulator having a vision system for viewing and identifying a part.
2) Description of the Prior Art
Robot manipulators are well known in the art and are frequently used in many industrial applications. Vision systems are utilized on certain types of robots for assisting the robot in recognizing a location and orientation of a part and identifying the shape of the part. These vision systems typically include a camera and a computer. The identification is typically accomplished by comparing the image of the part with an image on a stored template. Once the part is matched with the template, the relative location, orientation, and size of the part can then be determined. After the location and identification of the part are known, the robot may proceed with a preprogrammed operation, such as gripping and moving the part.
The identification and location of a part and its profile and/or features can be a difficult process. Prior art systems do not take into account the effects of image distortion and perspective variation when the part shifts and/or rotates on a viewing surface. In other words, a camera can skew and distort the image of the part, especially when the part rotates into different orientations. Many prior art systems use a template comparison based on gray scale normalized correlation to determine the location of the part.
Gray scale normalized correlation is a process which takes each pixel or point of a template and compares it with each pixel or point of the actual part being viewed. Using gray scale normalized correlation, however, is computationally expensive and sometimes computationally prohibitive. Gray scale normalized correlation is even more expensive in a situation where the orientation of the object can change. For example, if a 64×64 image is compared to a 16×16 template, then at least 1 million pixel comparing operations would have to be performed to compare the image to the template. This operation is expensive and can take an appreciable amount of time to perform.
Accordingly, there is a need for a vision system which can operate on a distorted image of the part to provide identification and location information without using computationally expensive algorithms.
SUMMARY OF THE INVENTION AND ADVANTAGES
A method of viewing and identifying a known part for a robot manipulator using a vision system having a computer and a camera with known viewing geometry and optics to view the part. The method comprises the steps of: storing a template of the known part in the computer with identification parameters to identify the part; determining the relative position of the camera in relation to the part; storing the relative position information of the camera in relation to the part; viewing the part through the camera and creating a distorted image of the part resulting from the relative position of the camera and the viewing geometry and optics of the camera; using the stored relative position information of the camera in relation to the part and the known viewing geometry and optics of the camera to convert the distorted image into a true image of the part which removes any distortions from the distorted image; determining the true parameters of the true image of the part; and comparing the true parameters of the true image of the part to the identification parameters of the stored template to allow the computer to identify the part such that the robot may accurately perform a desired operation on the part.
Accordingly, the subject invention provides a vision system which eliminates any distortions and perspective effects from an imaged picture of the part before the picture of the part is compared to the template. By removing the distortions before performing the remaining calculations, the true image of the part is determined and identified by a relatively inexpensive means.
REFERENCES:
patent: 4731860 (1988-03-01), Wahl
patent: 5177563 (1993-01-01), Everett et al.
patent: 5463697 (1995-10-01), Toda et al.
patent: 5491759 (1996-02-01), Nagao et al.
patent: 5499306 (1996-03-01), Sasaki et al.
patent: 5579444 (1996-11-01), Dalziel et al.
patent: 5651075 (1997-07-01), Frazier et al.
patent: 5946425 (1999-08-01), Bove et al.
patent: 6064749 (2000-05-01), Hirota et al.
Vision-based object registration for real-time image overlay (1995), author M. Uenohara and T.Kanade the robotics institute, Carnegie Mellon University).
Bayat Ali
Fanuc Robotics North America
Howard & Howard
Tran Phuoc
LandOfFree
Method of viewing and identifying a part for a robot... does not yet have a rating. At this time, there are no reviews or comments for this patent.
If you have personal experience with Method of viewing and identifying a part for a robot..., we encourage you to share that experience with our LandOfFree.com community. Your opinion is very important and Method of viewing and identifying a part for a robot... will most certainly appreciate the feedback.
Profile ID: LFUS-PAI-O-2991581