JPH03241470A - Posture detecting method - Google Patents

Abstract

PURPOSE:To miniaturize an image pickup means for the detection of the posture of a recognition object and a memory circuit and to shorten detecting time by using the image pickup means having two line sensors, and installing the image pickup means at a position where the recognition object is crossed with the two line sensors and is image-formed. CONSTITUTION:The image pickup means 1 having the two line sensors outputs data by converting an image signal in accordance with luminance to image data from the two line sensors. Furthermore, a driving mean 2 which rotates the image pickup means 1, an illumination 4, a robot 5 to move the recognition object 3, and a control part 6 which performs the control of the robot 5, etc., based on the image data from the image pickup means 1 are provided. Therefore, the image pickup means 1 is installed at the position where the recognition object 3 is crossed with the two line sensors and is image-formed, and the posture of the recognition object 3 can be detected based on each image signal generated from the two line sensors. Thereby, it is possible to miniaturize a posture detector and to shorten a time required for the detection of the posture.

Description

[Detailed description of the invention] [Industrial application field] The present invention relates to a posture detection method.

[Prior Art] For example, when an industrial robot grips arbitrarily placed rod-shaped parts and assembles a product, it is necessary to detect the posture of the parts.
For example, an image of the above-mentioned rod-shaped part is captured by an imaging means such as a video camera having one screen composed of 500 x 500 pixels, and the image signal output from each pixel is converted into image data and stored in a storage circuit. There is a method of detecting the posture of the above-mentioned rod-shaped parts based on the following.

[Problems to be Solved] The conventional method described above requires many sensors and a large-capacity storage circuit in order to obtain image data for one screen. For this reason, the device becomes large and data processing takes time.

In other words, of the image data on one screen, the bar-shaped recognition target and its surroundings occupy only a small portion, so there is no need for one screen's worth of image data, and much of the image data is unnecessary. be.

An object of the present invention is to provide an attitude detection method that realizes downsizing of an attitude detection device and shortens the time required for attitude detection.

[Means for Solving the Problems] The present invention provides a posture detection method that images a rod-shaped recognition target, outputs the image signal, and detects the posture of the recognition target based on the image signal. The imaging means is installed at a position where the recognition target intersects with the two line sensors to form an image,
Based on each image signal generated from the book line sensor,
The above problem is solved by detecting the posture of the recognition target.

[Example] Hereinafter, one embodiment of the present invention will be described based on the drawings.

In FIG. 1, reference numeral 1 denotes an imaging means having two line sensors, and when an image is captured, image signals corresponding to brightness from the two line sensors are converted into image data and output.

2 is a driving means for rotating the imaging means 1; 3 is a rod-shaped recognition object; 4 is an illumination; 5 is a robot for moving the recognition object 3; 6 is control of the robot 5 based on image data from the imaging means 1. This is a control unit that performs such operations.

In FIG. 2, the same numbers as in FIG. 1 indicate the same components. 1a is an imaging circuit that outputs an image signal; 1b is an image data output circuit that converts the image signal from the imaging circuit 1a into image data and outputs the image data; these constitute the imaging means 1.

Reference numeral 6a denotes a storage circuit consisting of a line memory that stores image data from each line sensor of the image data output circuit 1b, and 6b detects the peak of the image signal from each line sensor of the imaging means 1 based on the image data of the storage circuit 6a. A position detection circuit that calculates each value, 6c is a position detection circuit 6
A posture detection circuit 6d calculates the posture of the recognition target based on each peak value calculated in step b, and 6d controls the driving means 2 and the robot 5 based on the posture of the recognition target 3 calculated by the posture detection circuit 6c. These control circuits constitute the control unit 6. 8 is the position detection circuit 6
This is a notification means for notifying an error when the peak value is not calculated in step b.

FIG. 3 shows the internal configuration of the imaging means 1 and the recognition target 3, in which 1c and 1d are line sensors, and 1e is a lens system. Line sensors 1c and 1d are parallel,
It is installed at a position where the light beam passing through the center of the lens of the lens system 1e reaches the center of the line sensor 1c.

Next, the posture detection operation will be explained along the flowchart of FIG.

The width and the like of the recognition target 3 are input in advance using an input means (not shown) such as a keyboard. Next, when a recognition switch (not shown) is operated, imaging is performed by the imaging means 1,
Image signals are output from line sensors 1c and 1d,
The image data output circuit 1b converts the image data into image data, which is stored in the storage circuit 6a (step 2).

Next, the position detection circuit 6b detects each peak value of the image signals from the run sensors 1c and 1d based on the image data stored in the storage circuit 6a (step 2).

For example, if the images of the line sensors 1c and 1d and the recognition target 3 intersect as shown in FIG. 5(a), the waveform from the line sensor 1c as shown in FIG. 5(b), An image signal having a waveform as shown in FIG. 5(c) is output from d. n is the number of the light receiving element in the line sensor,
■ is the output voltage value from the light receiving element.

When each peak value is detected by the position detection circuit 6b,
Based on this, the posture detection circuit 6C calculates the posture of the recognition object 3, that is, the coordinates on the imaging screen of the position where the peak value of the line sensor C is detected and the inclination of the recognition object 3.

The coordinate system on the imaging screen is shown in FIG.

The X axis is parallel to line sensor c, and line sensor I
Set the origin O at the center of C. The run sensor d is placed at a position parallel to the line sensor C by a distance P in the y-axis direction. The peak value of the image signal of the line sensor 1c is σ, and the peak value of the image signal of the line sensor 1d is σ
, the pixel distance coefficient is β, the X-axis offset is X, the y-axis offset is yo,
When the lens magnification is α, the coordinates of the position where the peak value σ is detected are: X−βσ CO5eo +X.

a y−βσ　sjn　θo+y.

becomes. θ. is the angle when the imaging means 1 is rotated. In addition, the angle θ between the line sensors C and 1d and the image of the recognition target 3 is, if σ > σb, θ−180+θ −tan”(aP/a −a )
When Oa b σb>σ3, θ electric θ + jan'(αP/ab−a8) is obtained (step ■).

Next, based on the posture calculated by the posture detection circuit 6c, the control circuit 6d moves the center of the hand portion of the robot 5 to the coordinate (x, y) position, and then moves the hand portion to zero.
Rotate in the 0 degree direction to open the hand, lower the hand section to the position of the recognition object 3, grab the recognition object 3 with the hunt, move it to a predetermined place, and release the hand (step 2).

By the above operation, the two line sensors C and 1d
The posture of the recognition target 3 is detected based on the image signal from the robot 5, and the robot 5 automatically moves it.

By the way, if the peak value is not detected in the first imaging,
The imaging means 1 is rotated by θθ degrees by the driving means 2, and the recognition object 3 is imaged again (step 2).

For example, when the peak value is not detected, see Figure 7 (
As shown in a), line sensors 1c and 1d and recognition target 3
There is a state in which the images of the two images intersect.

At this time, the waveform of the image signal from the line sensor 1c is
As shown in FIG. 7(b), the waveform of the image signal from the line sensor 1d is as shown in FIG. 7(c).

This makes it impossible to identify the peak value. For this reason, the imaging means 1 is set at θ. By counter-rotating, Fig. 7 (
As shown in d), line sensors 1c and 1d and recognition target 3
The intersection angle θ with the image formation can be increased, and peak values clearly appear in the waveforms of the output signals of the line sensors 1c and 1d.

By the way, if the peak value is not detected during the second imaging, the control circuit 6d determines that the recognition target 3 does not exist within the field of view of the imaging means 1, and the notification means 8 issues an error notification (step ■).

In the above embodiment, the line sensors C and 1d are arranged in parallel, but the arrangement is not limited to this, and as shown in FIG. 8(a)
Alternatively, the arrangement may be as shown in FIG. 9(a).

Figure 8(a) shows line sensor C and line sensor 1d.
This is a case where the angle φ with respect to φ is set to φ≠0990°. As shown in FIG. 8(b), the coordinate system has its origin at the intersection of the extended lines of line sensor c and line sensor d, and the X axis is parallel to line sensor IC. The angle θ between the line sensor C and the image of the recognition target 3 is: When the peak value is σ 〈σb, o ”” tan-l (ασb + [3o) SINφ (
ασb+13o ) CO3−−(ασyu+＾0) If the peak value is σ 〉σb, o −11110tan−′ ((X (7b ”Bo
) SINφ (ασ + A ) − (ασb + Bo) CO
It becomes 8φO. Ao is the offset from the origin of line sensor c, and Bo is the offset from the origin of line sensor d.

Figure 9(a) shows line sensor c and line sensor 1d.
This is a case where the angle φ with respect to φ is set to φ−90′. In this case, the line sensors c and 1d are arranged as shown in FIG. 9(b), and the image of the recognition target 3 is formed on the line sensors C and 1d by the beam splitter 9. As shown in FIG. 9(C), the coordinate system has the origin at the intersection of line sensor IC and line sensor d,
The X axis is parallel to line sensor c. Line sensor I
The angle θ between C and the image of recognition target 3 has a peak value of σ
If ×σb is 0, then θ−tan’ (σ /σ ) a If the peak value is σ ×σb>O, then e−180−tan−7 (σ/a ) b
It becomes a.

[Effects] According to the present invention, the posture of the rod-shaped recognition target can be detected by the imaging means having only two line sensors, so that the imaging means and the image data storage circuit can be miniaturized and the posture detection time can be shortened. can be achieved and is very convenient.

[Brief explanation of drawings]

FIG. 1 is a block diagram showing an embodiment of the present invention, FIG. 2 is a block diagram showing an embodiment of the present invention, and FIG. 3 is an example of the internal structure of the imaging means 1 and the recognition target 3. An explanatory diagram,
Figure 4 is a flowchart for explaining the operation in Figure 2, Figures 5, 6 and 7 are explanatory diagrams for explaining the operation in Figure 2, and Figures 8 and 9 are in the main text. FIG. 7 is an explanatory diagram showing another embodiment according to the invention. 1...Imaging means lc, ld...Line sensor 3...Recognition target and above Figure 3 Figure 5 Figure 2 Figure 4 Figure 8

Claims (1)

[Claims] A posture detection method for capturing an image of a rod-shaped recognition target, outputting an image signal thereof, and detecting the posture of the recognition target based on the image signal, using an imaging means having two line sensors, The imaging means is installed at a position where the recognition target intersects with the two line sensors to form an image, and the posture of the recognition target is detected based on each image signal generated from the two line sensors. A posture detection method characterized by the following.