JPS622106A - Detection of two-dimensional posture angle using mark pattern - Google Patents

Abstract

PURPOSE:To detect automatically, simply, and at low-cost a two-dimensional posture, by measuring out-of-alignments of marks distributed at four equal distances along the periphery with respect to x- and y-axes and calculating the relative posture angles around the y- and z-axes basing upon them. CONSTITUTION:Conversion of the xyz fixed co-ordinate system into the rotary x'y'z' co-ordinate system by turning by angles psi, theta with respect to the y- and z-axes can be made by simple geometrical calculation. Thus, out-of-alignments of the fixed co-ordinate system in the x- and y-directions of the marks 1, 3 and 2, 4 distributed with equal distances on a circle of radius r0 of, for instance, a wheel are measured by a TV camera, etc. Basing upon this measured values, aforesaid conversion into the rotary co-ordinate system is performed for calculating relative posture angles psi and theta around the y- and z-axes. By these processes, accurate automatic detection of the 2-dimensional posture can be made simply and at low cost without applying a complicated shape recognizing method and automatic assembly operation, etc. can be available.

Description

[Detailed Description of the Invention] [Industrial Application Field] The present invention relates to a method for detecting a two-dimensional attitude angle relative to a pattern in which marks are placed at quarter positions on one circumference. It is.

[Prior Art] For example, an automobile wheel has bolt holes at four equal positions on one circumference of a disc part in the rim, and the wheel is automatically attached to the part of the car body. In this case, it is necessary to install the wheel by aligning the positions of the bolts in the hole of the wheel and the wheel mounting part of the vehicle body.

In addition, not only the wheels of the automobile mentioned above, but also various industrial parts often have some kind of mark such as a hole at four equal parts of a circumference, and it is difficult to automatically assemble the parts. , it is necessary to automatically detect the positions of those holes, that is, the orientation of the parts.

Furthermore, when a moving body such as a movable mouth pot that moves within a space appropriately measures its own posture, a circular shape or the like is placed at an appropriate position within the space at four equal positions on one circumference. By displaying a characteristic pattern inside the cylinder with the mark and detecting that pattern on the moving body, the attitude of the moving body itself can be measured.

Conventionally, although it is possible to detect the attitude of such a pattern by using a relatively advanced pattern recognition technique, no method has been proposed for detecting the attitude of the pattern in a cylinder using an inexpensive device.

[Problems to be Solved by the Invention] An object of the present invention is to solve the problem by using a very simple and inexpensive device and by real-time processing to determine the relative two-dimensional orientation between the patterns, that is, orthogonal to each other. An object of the present invention is to provide a method for automatically detecting relative rotation angles around two axes.

[Means for Solving the Problems] In order to achieve the above-mentioned object 1, the method of the present invention has a method in which a pattern in which marks are located at four equal positions on one circumference is arranged on y-axes and y-axes that are orthogonal to each other. When taking an appropriately rotated posture around the center, the two pairs of marks located opposite to the center of the above are perpendicular to the viewing direction taken in the z-axis direction and the y-axis direction and the y-axis direction, which are perpendicular to the line of sight and mutually orthogonal. A technical means is employed to measure the positional deviation of the y-axis and detect the relative attitude angle around the y-axis based on the measured positional deviation.

According to the method of the present invention, the necessary positional deviation of the mark is measured from the image of the pattern detected using an ITV camera, etc., and simple calculations are performed.
The required attitude angle can be measured in real time.

To explain the method of the present invention in more detail, in an Xy2 fixed coordinate system in a certain space as shown in FIG. 1A,
Rotate by an angle ψ around the y-axis, and rotate by an angle θ around the two axes, and the newly obtained rotated coordinate system can be transformed into x' y ' z ' coordinates as shown in Figure 1B. system, the coordinate transformation from the xyz fixed coordinate system to the x'y'z' rotating coordinate system is performed by geometric calculation as follows:
This can be done using the relationships shown in the table.

In the present invention, the pattern in which the two-dimensional posture is to be detected is, as shown in FIG.
This pattern has an arbitrary posture by changing the y-axis and the rotation around the y-axis with respect to the xyz fixed coordinate system.
It is assumed that the center of the circle in which the marks 1 to 4 are arranged is aligned with the origin of the rotating coordinate system on the y' plane.

Note that the above-mentioned pattern only needs to have at least four marks attached so that it can be easily identified, and therefore, there is no problem even if additional marks and figures are mixed.
When using the above-mentioned pattern to detect the attitude of a moving object as described above, it is possible to make the shape of the mark in a plurality of patterns set in the traveling area of the moving object different, and use the mark shape to detect the position of the moving object. can.

In order to observe the pattern in the above rotating coordinate system from the xyz fixed coordinate system as shown in Figure 2B, x'y'
The coordinates of marks 1' and 2' in the z' rotational coordinate system are
Each.

Let PI' (chi, 0, 0) P2' (0, rO, O) and convert the coordinates of those marks to coordinates P, , P in the xyz fixed coordinate system, then the coordinates P, , P
Each coordinate value of 2 can be expressed as follows by referring to Table 1.

P, (cos Oa cosψ* ro, s'+
nOa r,.

-cosOm sinψ・ro) P2 (-ginθa cosψ・rO1CO3O r
. .

5inO・sinψ・r, ) Here, for each of the two pairs of marks located opposite to the center of the circle in which marks 1 to 4 are arranged, that is, mark 1.3 and marks 2 and 4, the line of sight direction is (Z-axis direction), the positional deviation in the X-axis direction perpendicular to the above two axes and the X-axis is P31(x) and P42(CT). Then, among these positional deviations, those necessary for the following calculations are measured, and based on those values, the first-order y
7-γ7. Any one of , and B, γ2. By calculating either of these, it is possible to measure the relative attitude angles ψ and θ around the y-axis and the two axes using a very simplified formula.

Note that the calculation formulas for the attitude angles θ and ψ shown here are merely examples, and the present invention is not limited to these.

(1) Measurement of θ rnO = sin θ COSθ =cog.

(2) Measurement of ψ COSψ = Cogψ As can be seen from the above equation, the attitude angle θ of the pattern around the two axes can be detected independently of the attitude angle ψ around the y-axis. Furthermore, these attitude angles θ and ψ are unrelated to the distance between the pattern and the ITV camera or the like that detects it.

Note that the distances from the X-axis to marks 1 and 3 are the same, and the distances from the X-axis to marks 2 and 4 are also the same, so when calculating γ Instead of measuring the distance between marks 2.4 and 2.4, it is also possible to measure the distance from the X axis to the mark. Also, mark l from the y axis
The same applies to the distance from the y-axis to the mark 3 and the distance from the y-axis to the mark 2 and mark 4.

Next, considering the measurement error in attitude angle detection of such a pattern, the measurement error Δ for attitude angle θ is
0. and the measurement error Δψ for the attitude angle ψ is the above ratio γ
When the measurement accuracy for 6 is denoted by Δγ, it is expressed as follows.

(1) Total n error Δθ, °, Δ0 = attitude angle θ
−5in20 * Δr, t [r a d ]
3°) Jy,,=. . 8・o°70,°, Δ0 = cos20 * ΔYtl [r
adl(d) Δy,, = co sθφ Δ
θ(f) Δ'! , = −5InO* Δ θ(
2) Measurement error Δψ(b) Δηλ regarding attitude angle ψ
=-sinψφΔψ Figures 3 to 6 show the error -〇 when calculating (a) to (d,) when measuring the attitude angle θ, and Figures 7 and 8 show the attitude The angle ψ is (a), (b)
It shows the error Δψ when calculated using the formula. Note that the error when calculating the attitude angle θ using the equations of degrees (e) and (f) shows the same tendency as fp, FIGS. 7 and 8.

In these drawings, it is assumed that the measurement error for 20 meshes is the ratio of KS meshes during input processing by the image input device for the above pattern, and the KS
This shows the case where Δθ and Δψ are obtained by changing the values of .

Generally, in an 84x48 mesh image input device,
When an image of the above pattern is captured with a 20x20 mesh, the KS can be approximately 0.5 to 0.25, and image processing is usually performed at four times that amount (25B) in television cameras. Therefore, by performing such image processing, the total J11 accuracy becomes simply the same, and excellent measurement accuracy can be easily obtained.

According to Figures 3 to 8, it can be seen that there are differences in errors depending on the calculation method.2 The calculation method to be adopted should be determined in relation to other measurement conditions. be.

Note that the apparatus for carrying out the attitude angle detection method described above converts the above-mentioned pattern into an image using an ITV camera or other image input device, and uses an arithmetic device connected to the image input device to image the opposing position in the above-mentioned pattern. It is only necessary to measure the positional deviation of the two sets of marks and configure the above-mentioned calculation, thereby determining the attitude angles θ and ψ.
can be detected.

[Example] A figure as shown in FIG. 2A was displayed on a flat plate, and the flat plate was rotated around the vertical axis within a range of ψ=-so'' to 80° to measure the error. Figure 9 shows the measurement results.The measurement of the attitude angle 0 was carried out using the equation (c) above, and the measurement of the attitude angle ψ was carried out using the equation (b) above. The value is 0.25.

This experimental result agrees well with the calculated value, so
This means that the reliability of the above calculated values has been confirmed.

[Effects of the Invention] As detailed above, according to the attitude angle detection method of the present invention, the two-dimensional attitude of the above pattern can be automatically detected with high accuracy using a very inexpensive device in a cylinder. .

[Brief explanation of drawings]

FIGS. 1A and 1B are explanatory diagrams of a fixed coordinate system and a rotating coordinate system for explaining the posture angle detection method of the present invention, and FIGS. 2A and 2B are explanatory diagrams of coordinate transformation of a pattern between the above two coordinate systems. 3 to 8 are explanatory diagrams of errors in detecting attitude angles according to the present invention, and FIG. 9 is a graph showing experimental results regarding the present invention. 1-4 ・Φ mark. 111th A B 2nd E A B Fig. 3 θ(″) 4th E θ(+)

Claims (1)

[Claims] 1. In a state where a pattern with marks at four equal positions on a circumference is appropriately rotated around the y-axis and z-axis, which are perpendicular to each other, the pattern is placed in a position opposite to the center of the circle. For a certain pair of marks, measure the positional deviation in the x-axis and y-axis directions, which are perpendicular to the viewing direction taken in the z-axis direction and mutually orthogonal, and calculate the relative positional deviation around the y-axis and z-axis based on them. A method for detecting a two-dimensional posture angle using a mark pattern, which is characterized by detecting a posture angle.