JPH0364806B2 - - Google Patents

Description

[Detailed Description of the Invention] [Technical Field of the Invention] The present invention provides a method for easily obtaining posture information of a measurement object from distance information of the measurement object obtained using a light projection means such as a laser range finder. This invention relates to a posture measuring device that can be obtained at high speed.

[Technical background of the invention and its problems]

In a robot that performs work such as grasping and moving an object, it is important to accurately grasp the posture in which the object to be grasped is placed, that is, the posture of the object. In particular, when a robot is used to pick up so-called bulk parts, it is very important to accurately grasp the orientation of each part.

Conventionally, as a means for accurately grasping (measuring) the attitude of such an object, a laser range finder that can directly obtain distance information of each part of the object as its coordinate information has been widely used.

Conventionally, all the coordinate information forming the object is obtained using a laser range finder, and the posture information of the object is obtained from the correlation of these coordinate information. Therefore, an enormous amount of calculation processing is required to obtain posture information of an object from these coordinate information. For this reason, it has been difficult to obtain posture information of an object at high speed and understand its posture. Moreover, as the orientation information of the object is treated as a large volume of coordinate information as described above, a large-capacity storage device is required to write this information, and data necessary for controlling the robot can be extracted from this information. There were some problems, such as the processing required to do so was quite complicated.

[Purpose of the invention]

The present invention has been made in consideration of these circumstances, and its purpose is to provide a highly practical system that can easily and quickly grasp the posture of an object and effectively utilize it for robot control, etc. An object of the present invention is to provide a posture measuring device.

[Summary of the invention]

The present invention uses a light projection means such as a laser range finder to sequentially obtain distance information regarding an object to be measured as line information at predetermined scanning intervals, and calculates a line segment indicating the object to be measured in each line information. For example, information on the length and position of each line segment is extracted, and the connection relationship between each scanning line of the information on these line segments is extracted, for example, as a set of line segments that are mutually related between each line. Find as,
The attitude of the object to be measured is determined as an equation expressing a straight line connecting each line segment of this set.

〔Effect of the invention〕

Thus, according to the present invention, while inputting distance information regarding a target object sequentially at predetermined scanning intervals, line segments indicating the target object in the line information are determined, and correlations are calculated from correlations across each line of these line segments. A certain number of line segments is regarded as a set of line segments indicating the same target object, and the straight line connecting each line segment of this set of line segments expresses the pose of the target object as a linear equation, so the pose of the target object can be expressed as a straight line equation. Information can be sought quickly. Moreover, each time new line information is input, the line segment information extracted from the line information is added to the set of already determined line segments and the linear equation is sequentially corrected. It becomes possible to determine the posture of an object with high precision.

In addition, since the attitude of the target object is expressed as a linear equation, it is possible to describe the attitude of the target object using a small-capacity storage device, and the attitude information can be used directly for robot control, etc. A great practical effect can be achieved, such as:

[Embodiment of the Invention] Hereinafter, an embodiment of the present invention will be described with reference to the drawings.

FIG. 1 is a schematic configuration diagram of the embodiment device, FIGS. 2 and 3 are diagrams showing the processing concept thereof, and FIG. 4 is a diagram showing an example of the processing sequence in the embodiment device.

In this embodiment, various industrial parts, which often have cylindrical (column) shaped parts, are exemplified as objects to be measured, and a case will be described using as an example a case where the objects are gripped by a robot.

The laser range finder 1 irradiates a measurement target object with laser light and inputs the distance information as, for example, x-y-z coordinate data of each part of the measurement target object. Here, a laser range finder 1 is used in which distance information about the object to be measured is input line by line while moving a laser line with a predetermined scanning width at predetermined scanning intervals.
Through this laser range finder 1, distance information of a plurality of objects a, b, and c is sequentially inputted one scanning line at a time, as shown in FIG. 2a, for example.
In addition, while scanning the laser beam with a predetermined scanning width,
It is also possible to use a laser range finder 1 that sequentially inputs distance information of an object to be measured while moving its scanning line at predetermined scanning intervals. In this case, the distance information of the object to be measured that is input sequentially may be processed in units of one scanning line at a time.

The line segment cutting section 2 is a laser range finder 1.
From the line information sequentially input for each scanning line using
The images are extracted using the continuity of the z-coordinate data or a threshold value for distinguishing the target object from the background. For example, as shown in Fig. 2b, the line segments of target objects a and b on scanning line (K-1) are designated as h1 and h2, and the line segments of target objects a and b on scanning line K are designated as h3 and h4, respectively. Each is extracted from the line information. Information on each line segment h1, h2, to h4 includes, for example, length information and position information in the scanning area (area shown within the frame in the figure) by the laser range finder 1, such as x-y- Each is output as z coordinate data.

Therefore, the set acquisition unit 3 compares the line segment information obtained for each scanning line with the line segment information obtained in the previous scanning line, and We are looking for a set of line segments detected from an object. Specifically, information on each line segment cut out for each scanning line is compared between each scanning line, and line segments with almost the same position information and almost the same length are detected from the same object. It is determined that the line segment is In the example shown in FIG. 2, line segments h1 and h3 are grouped together as one set (set) belonging to target object a, and line segments h2 and h4 are grouped together as one set (set) belonging to target object b. It can be summarized.

In addition, when comparing line segment information between each scanning line, if there are several line segments with similar information values, for example, if |h2|≒|h3|≒|h4| , pair line segments that are close to each other in distance (1
set). Alternatively, a set determination process is performed by using information on straight lines connecting each line segment, which will be described later, of a set of line segments that have already been found. Each time the line segment information of each scanning line is sequentially given on a line-by-line basis, the set importing unit 3 collects the line segments extracted from that scanning line into a set that has already been obtained (lines related to the same target object), as described above. The line segment to be processed is added to the determined set. Note that if a set to which the determined line segment belongs is not found, it is assumed that the line segment relates to a newly appearing target object, and a new set of line segments is created based on this line segment.

For each set of line segments obtained in this way, the straight line generation unit 4 obtains the barycenter (center) coordinates of each set of line segments from the information on the length and position of each line segment belonging to the set, and connects these barycenter coordinates. Calculating the equation of a straight line. For example, as shown in Fig. 2c, an equation representing the straight line f1 connecting the barycenter coordinates of the line segments h1 and h3 is found, and an equation representing the straight line f2 connecting the barycenter coordinates of the line segments h2 and h4 is similarly found. There is. The equations of these straight lines f1 and f2 are output as information indicating the postures (inclinations) of the target objects a and b, respectively, and are also written in the control information memory of the robot.

By the way, as described above, the information on the line segments belonging to the set of line segments representing the target object increases sequentially as the laser range finder 1 scans the line of the target object. Therefore, at the initial stage of posture measurement, the newly added line segment does not necessarily exist exactly on the straight line that the straight line generation unit 4 has calculated from the two line segment information forming one set. In other words, the information on each line segment itself includes a certain error, and the equation of the straight line connecting the two line segments is calculated from the information on two line segments that are almost not far apart from each other in terms of distance between adjacent lines. The amount of error in the linear equation is relatively large. Therefore, the straight line correction unit 5 corrects the previously calculated straight line equation each time a new line segment is added to each set of line segments.

For example, as shown in Fig. 3, after calculating the equation of the straight line F1 connecting the centroid points of the line segments H1 and H2, when the line segment H3 is added to the set of the line segments H1 and H2, the line segments H1, H3 Calculate the equation of the straight line F2 connecting the centroid points of . And these straight lines F1, F2
The average value of each coefficient is calculated, and the equation of the straight line F3 is calculated using each of these average values as a coefficient. The equation of this straight line F3 is the straight line F1
A new equation is created by modifying , that is, an equation of a straight line connecting the line segments H1, H2, and H3 of the set. Thereafter, in the next scanning line, when line segment H4 is added to the set of line segments H1, H2, and H3, the linear equation is similarly corrected.

In this way, the line segment information regarding the recognition target object is sequentially compiled into a set of line segments corresponding to each target object, and as shown in Figure 2d, each line segment is represented as a straight line connecting the line segments belonging to the set. The pose of the target object is then determined. That is, in this example, a set of line segments including line segments h1, h3, h6, and h9 is extracted as representing the target object a, and the equation of the straight line F1 connecting these line segments h1, h3, h6, and h9 is It is output as information indicating the attitude of the target object a. Similarly, line segments h2, h4, h5, h
8 is obtained as information indicating the attitude of the target object b, and the equation of the straight line f3 connecting the line segments h7 and h10 is output as information indicating the attitude of the target object c.

FIG. 4 shows an example of a control sequence for the above-mentioned attitude measurement process.

As shown in this control sequence example, a control parameter K is set in response to line-by-line input of distance image information by the laser range finder 1,
While extracting line segments on the scanning line K, the set to which the line segment belongs is determined, the equation of a straight line connecting the line segments belonging to each set is calculated, and the equation of this straight line is sequentially corrected. In this way, each posture of the target object existing within the measurement target region can be determined as a linear equation.

As explained above, according to this device, the distance information of the object to be measured is input, the distance information is processed in units of scanning lines, and the information of the object included in the line information is extracted as line segment information. Then, the object to be measured is grasped as a set of line segments from the connection relationship of the line segments between the plurality of lines, and the attitude of the object to be measured is measured as an equation of a straight line connecting the line segments belonging to the set. Therefore, as the line scanning of the target object is performed, the posture information of the target object is obtained as the linear equation, and the linear equation is sequentially corrected using new line segment information, so that the posture of the target object can be determined at high speed. You can get information. In other words, it is possible to quickly obtain rough posture information of the target object and to sequentially correct the rough posture information to obtain highly accurate posture information. In addition, the above-mentioned posture measurement can be performed by a simple process of calculating a set of line segments, calculating the equation of a straight line connecting each line segment belonging to the set, and correcting the equation.

Furthermore, since the orientation of the target object is expressed as a straight line equation, the only information necessary to store this is the coefficient information of the straight line equation, making it possible to significantly reduce the required storage capacity. Furthermore, since posture information is expressed as a linear equation, the information can be directly used for robot control. Therefore, there are no problems such as using a large capacity memory or requiring a long time for processing, which is the case in the past, and there are great practical effects.

Note that the present invention is not limited to the embodiments described above. In the embodiment, a cylindrical (cylindrical) part was the measurement target, but all parts having a so-called elongated partial shape can be similarly measured. Furthermore, when calculating the straight line equation, information on the Z coordinate maximum point of the line segment information may be used instead of the center of gravity shown in the embodiment. In this way, it becomes possible to suppress the error caused by the length of the line segment and to measure the posture with high precision. Further, as a method for calculating the equation of a line segment, it is also possible to use a method other than the least squares method. Furthermore, the pitch of the scanning lines, etc. may be determined according to the specifications of the apparatus, and the present invention can be implemented with various modifications without departing from the gist thereof.

[Brief explanation of drawings]

The figures show one embodiment of the present invention, and FIG. 1 is a schematic configuration diagram of the embodiment device, FIGS. 2 a to d are diagrams showing the concept of attitude measurement processing in the embodiment device, and FIG. 3 is a linear equation. FIG. 4 is a diagram showing the processing control sequence of the embodiment apparatus. DESCRIPTION OF SYMBOLS 1...Laser range finder, 2...Line segment cutout unit, 3...Collection capture unit, 4...Line generation unit, 5...Line correction unit, a, b, c...Target object, h1, h2 , ~h10... line segment, f1, f
2, f3...straight line.

Claims (1)

[Claims] 1. Means for irradiating a measurement target object with a laser beam while moving at a predetermined scanning interval, and sequentially inputting distance information regarding the measurement target object as line information; means for sequentially extracting information on the length and position of each line segment indicating the object to be measured in each of the line information; A means for detecting whether line segments are from the same object to be measured, and a means for determining an equation of a straight line connecting the centroids of each line segment from the same object to be measured detected by this means. A posture measuring device comprising: outputting the equation of the straight line obtained by this means as information indicating the posture of the object to be measured. 2. The posture measuring device according to claim 1, wherein the straight line equation is sequentially added to and corrected information on already determined line segments.