|
|
|
ASSISTED VISUAL SERVOING BY MEANS OF STRUCTURED LIGHTAuthor: PAGÉS MARCO JORDI. Year: 2005. University: GIRONA [ www.udg.es]. Place of defense: UNIVERSIDD DE GIRONA. Place of preparation: UNIVERSIDAD DE GIRONA. Summary: This paper deals with the combination of visual inspection with the light structured to perform positioning robot connection objects. The visual inspection is a technique that allows widespread currently perform robotics using a control loop based on visual information. The most common configuration is to attach a camera at the terminal element of a robotic arm or a mobile robot. The objective of the task is defined from the perceived image that the camera once the robot is in the desired position regarding the subject. Then, when the robot is moved to another position or the object is moved, the loop-based control vision is able to return to the robot to the desired position relative. To make a loop based on vision is necessary that the object of interest containing visual elements. Therefore, the classic visual inspection is not capable of dealing with objects slightly textured, with uniform appearance, or complex objects in which it is difficult to extract visual elements in a robust and fast. The solution we are proposing in this thesis is to use structured light to project visual elements in objects regardless of their appearance. The structured light is a technique estero-visión active in a light emitting (now more flexible option is a video-proyector) projected a pattern of light known. Then, noting the object lit by a camera, it is possible to obtain the correlation between image and pattern, and then you can get information 3D from reconstruction. The thesis presents a detailed study of the light-encoded structured, analyzing the various existing patterns. Different results show the performance of each type of pattern within the typical application of structured light: the 3D reconstruction of the object lit. It has made a contribution in this field with the presentation of a new pattern that improvement in terms of the existing resolution (number correlation) and accuracy. After this study, the combination of projection patterns of structured light with the visual inspection is properly studied. Specifically, two possible configurations have been addressed. In the first, a vído-proyector static located next to a robotic arm is used to illuminate the work area. Thus, when projecting a pattern of dots with color-coded visual elements are obtained in order to position the robot regarding the illuminated object, regardless of the appearance of the latter. In this sense, this is the first time that this type of configuration has been used for visual inspection. Secondly, has been studied using a projector attached to the camera's robotic arm. In this case, a pattern of light based on laser technology has been designed. This sensor allows the robot to position the camera parallel to a flat object using the projected points. In this case, they proved that the structured light arriving at a robust control law to calibration errors in an analytical way i pilot.
ASSITED VISUAL SERVOING BY MEANS OF STRUCTURED LIGHTAuthor: PAGÉS MARCO JORDI. Year: 2005. University: GIRONA [ www.udg.es]. Place of defense: UNIVERSIDAD DE GIRONA. Place of preparation: UNIVERSIDAD DE GIRONA. Summary: This paper deals with the combination of visual inspection with the light structured to perform positioning robot connection objects. The visual inspection is a technique that allows widespread currently perform robotics using a control loop roast in visual information. The most common configuration is to attach a camera at the terminal element of a robotic arm or even mobile robot. The objective of the task is defined from the perceived image that the camera once the robot are in the desired position regarding the subject. Then, when the robot is moved to another position or the object is moved, the loop-based control vision is able to return to the robot to the desired position relative. To make a loop based on vision is necessary that the object of interest containing visual elements. Therefore, the classic visual inspection is not capable of dealing with objects slightly textured, with uniform appearance, or complex objects in which it is difficult to extract visual elements in a robust and fast. The solution we are proposing in this thesis is to use structured light to project visual elements in objects regardless of their appearance. The structured light is a technique estereo-visión active in a light emitting (now more flexible option is a video-proyector) projected a pattern of light known. Then, noting the object lit by a camera, it is possible to obtain the correlation between image and pattern, and then you can get correlation between the image and pattern, and then you can get information 3D from reconstruction. The thesis presents a detailed study of the light-encoded structured, analyzing the various existing patterns. Different patterns exist. Different results show the performance of each type of pattern within the typical application of structured light: the 3D reconstruction of the object lit. It has made a contribution in this field with the presentation of a new pattern that improvement in terms of the existing resolution (number correlation) and accuracy. After this study, the combination of projection patterns of structured light with the visual inspection is properly studied. Specifically, two possible configurations have been addressed. In the first, a video-proyector static located next to a robotic arm is used to illuminate the work area. Thus, when projecting a pattern of dots with color-coded visual elements are obtained in order to position the robot regarding the illuminated object, regardless of the appearance of the latter. In this sense, this is the first time that this type of configuration has been used for visual inspection. Secondly, has been studied using a projector attached to the camera's robotic arm. In this case, a pattern of light based on laser technology has been designed. This sensor allows the robot to position the camera parallel to a flat object using the projected points. In this case, they proved that the structured light arriving at a robust control law to calibration errors in an analytical way i pilot.
|
|
|