JPS5923214A - Object position measuring system - Google Patents

Abstract

PURPOSE:To raise the position measuring accuracy of a long-distance object, by measuring from different positions by plural observing devices. CONSTITUTION:An observing device 12 is installed at each observing point, and each device is connected to each other by a circuit 13. An object whose position is unknown and a calibrating object are denoted as 14 and 15, respectively. When the device 12 is installed at the i-th observing point, a sensor 17 inputs a beam 16 of the object 14, a detecting circuit 18 detects an image position of an object on the sensor face, executes a calculation, and calculates a unit eye vector Uii to the object. The Uii value is sent to other observing device through the circuit 13 by a transmitter 26. A unit eye vector Ujj sent from other observing device is received by a receiver 27 and is sent to a position calculating circuit 19. A rotation matrix Gij showing a relative position between observing points (i), (j), and a position vector rji are stored in a memory 20 with respect to all (j)s, and the circuit 19 executes an operation by use of the vector Ujj, the matrix Gij and the vector rji, and can calculate the position of the object.

Description

DETAILED DESCRIPTION OF THE INVENTION The present invention relates to a method for measuring the position of an object in a three-dimensional space, and is particularly suitable for precisely determining the position of an object through distributed processing using a plurality of simple passive sensors. Regarding the method.

The position of the object in the three-dimensional space (Xl, Yl z) is expressed as H
Conventionally, optical distance measurement has been used as a passive means of measuring the distance. Using the distance information from the observation point to the target object obtained from the distance ^1 eye 1 and a separate optical sensor, we calculate the position of the image on the sensor surface. The position of the object relative to the observation point can be determined by the information on the direction of the object as seen from the observation point that can be retrieved. However, this method often has the disadvantage that the distance meter's ability to measure objects at a distance decreases, and the accuracy of measuring the position in that space decreases.

Therefore, an object of the present invention is to develop a method for accurately measuring the coordinates of objects located at long distances r and 'lf in three-dimensional space.

In order to achieve this purpose, the present invention utilizes the azimuth information of a target object from a plurality of sensors installed at different locations to determine the location of the above-mentioned sensors. There is a patent on the method used to calculate the position of an object.

Here, the principle of the 11 three-dimensional shapes used in the present invention will be explained. First, target from the direction observation information of the two observation points!
We will explain how to calculate the position of one snout. 1 in Figure 1.
2 is the observation point and 3 is the target object point.The coordinate system 10 is fixed at the observation point 1 (x+ + Yl + z+)
Let be the reference coordinate system. In this coordinate system, the position vector of point 2 is +21, the line of sight vector from point 1 to 3 (unit length)
4 as u, 1, unit-long line-of-sight vector from point 2 to point 3, 5
It1, the position vector r of unknown point 3? shall be expressed as At this time, the position vectors r and l of point 3 are determined by the following equation as the intersection of the line of sight from point 1 to 3 and the line of sight from point 2 to 3.

ri “tfull” +21 It2 u□ (1) However, tI
+F is a parameter (scalar quantity).

Foreshadowing vectors can be obtained, for example, as follows.

Consider an optical sensor assembly centered at observation point 6 in FIG. Assume that a sensor image 9 of a target object i exists on a focal plane 7 located at a distance from the origin 6 by the focal length f of the sensor on the y-axis of a coordinate system 8 having the origin at the point 6. The coordinates of image 9 are (x
f, f, Zf), the line of sight vector along the line of sight connecting point 6 and image 9 is u=(xf/d f/d Zf/d
) T The force can be seen in (2).

However, d = (It2-4- f'+z f'')Z2(3)T: Symbol indicating a transposed matrix.The line-of-sight vector u11 defined in the coordinate system 10 at point l in Fig. 1 is determined by this method. Similarly, the line of sight vector at point 2 is ``t'', which is defined in the coordinate system lNX2 + y! + zt ) fixed at point 2.
t can be obtained from equation (2). u22 can be converted to ull defined in coordinate system 10 using the following equation.

u! +"Mountain tuz (4) Here, G1. is a rotation machine (IJ Lux) for aligning each axis direction of the coordinate system 11 with each axis direction of the coordinate system 10.

To summarize the above, rotation matrix G12, coordinate system 1
If the position vector r21 of point 2 defined by
1) According to equation 1, the coordinates of the object at unknown point 3 are '5F.

Next, a method for calculating the relative positional relationship between sensor images at two observation points will be described. Based on this, the position vector r21 in equation (1) for the two observation points 1.2 in Figure 1,
(4) Rotation matrix G in Eq.

seek. An object i (i=1
．． ..., N), and its images are obtained by sensors at points 1 and 2. Let the position vectors of object i in coordinate systems 10 and 11 be r , l , r
, I, there is rz' = 01'ii' between them.
(r+' ”21) (5)(G +2-
''' (inverse matrix of G It).If the observation sensor is an optical sensor centered at observation point 6 in Figure 2, then the senna at observation point 1 (denoted as St)
The position of object 10) (g) observed at observation point 2 is also given by the position of the image at the sensor (denoted as S2) at observation point 2.
Known by the distance of point s of St, r r' - (X+' y+
'z-+')'+rt' = (X2')'t'
zt')τ. Therefore, if the positions of images of N objects on both sensor surfaces can be measured, (5), <6),
Using the least squares method from the relational expression (7), rotation matrix G,2 and unit vector r9. ′ is found.

Here distance between 2) (8) It is. The solution method is described, for example, in the following document.

D, Qennery, 5tereo-Came
raCalibration, Proc, of
AR,PA ImageUnderstondfng
Workshop, 101-107° Nov.
1978° In this method, the position of observation point 2 relative to 1 is a unit vector ball 7 that indicates the direction. ′ is the only thing I want! 1', the distance rt+ is indefinite. This rt+ is obtained by the following method. In other words, it is assumed that the distances between two points in a set of N point objects are known. Let these two points be t and the distance r
Let it be kL. Rotation matrix G81 obtained by the above method,
Unit vector r2. ', and observation point 1 of the object at point 1 (
From the position of the image in both sensors of
) and (1), the position vector rk' at the point can be obtained. However, in formula 1) □Engine 2. 7. ’ is used. The position vector tl is similarly obtained for point t. Therefore, using the known distance r, the distance r, 1 between the two observation points 1 and 2 is given by calculating kt'...-1,...-4...1 (9). From this, the unknown position vector rz+ is determined according to equation (8). G,2 and r21 can be determined using the above method.

Hereinafter, the present invention will be explained in detail based on examples. Third
The figure is an overall configuration diagram of an object position measurement system in a three-dimensional space according to the present invention. Observation equipment 12 (
- Points j) are installed, and they are connected to each other by a circle i13. It is assumed that a target object sound whose position is unknown is 14, and a calibration object is 15.

FIG. 4 is a configuration diagram of the observation device 12. It is assumed that the nuclear device is installed at the i-th observation point. Sensor 17 captures light rays 16 from object 14 . The detection circuit 18 detects the body position 11v of the object on the sensor surface, calculates the equation (2), and calculates the unit line of sight vector "11" toward the object.
The value is sent by the transmitter 26 over the line 13 to the music observation device. The U-th unit line-of-sight vector (j\1) sent from another observation device through the entire line 13 is received by the receiver 27 and sent to the position calculation circuit 19. memory 2o
A rotation matrix QB indicating the relative positional relationship between the observation point i and another observation point j and a position vector r++75Xf are stored for all j. The position calculation circuit 19 calculates the unit line-of-sight vector I at the observation point j, which is the output of the receiver 27.
J1+ and read from memory 20, matrix 0
17 and vector rjH, (Equation 21 and (
1) Calculate the equation to calculate the position rt of the object 14.

In this way, the unknown position of an object in three-dimensional space can be measured. This result is sent to the operating device 21, and predetermined control can be performed based on this value. The above is the basic operation of the position measurement method according to the present invention. It is assumed that the memory 20 stores in advance a matrix ()+1 and one vector necessary for calculation via an external input device 25 (for example, a keyboard input device) f:. This is based on the position of each observation point and the direction of sensor arrival (the direction of the sensor surface is x, y).

(expressed as an angle with the Z axis) is accurately known in advance.

Next, the operation when the position of each observation point and the mounting direction of the sensor are not known in advance will be explained. At this time, the matrix ()+j and vector rjI are calculated by the calibration calculation circuit 22 using the method described below, and the results are stored in the entire memory 20. After this calibration operation is completed, the unknown object position measurement operation is as described above. As shown in FIG. 3, a plurality of calibration objects 15 are installed at arbitrary positions (the number is N). Each observation device 12 takes images of these calibration objects 15 one after another. All installation positions of observation device 12 in Figure 4
shall be. As in the case of measuring the position of the unknown object described above, the line of sight vector toward the object 15 is obtained by the sensor 17 and the detection circuit 1B. Let the line-of-sight vector to the first (th) calibration object be uIIk. For all 1(u,,'), send to other observation devices 12 via all transmitters 26 and all lines 13. Observation at other j points The line of sight sent from the front 12 of the device through the entire line 13 (k=1 + ・-・・* N
) After being received by the u receiver 27, it is sent to the calibration calculation circuit 22. The circuit 22 outputs ul and the output of the detection circuit 18.
and the output of the receiver 27 uJ/' (k= 1
＋・・・・・・．

Based on N), equations (5), (6), and (7) are solved by the above-mentioned least squares method to obtain the rotation matrix G+1 and the level vector ru'-fr. On the other hand, calibration object 1
Assume that the distance rkt for 21m (k and t) of 5 is known and given from the external input device 24 (fC, for example, the keyboard input device t). (Using -, the circuit 22 calculates the distance r between observation points i and j using equation (9).
B? The desired position vector rB is calculated using equation (8).
k. Perform the above calculation 1) qx positive true calculation circuit 22
If so, a normal microprocessor can be used. In this way, Matrix G+1 and Pelitor R4. are stored in the memory 20. The above calibration calculation is performed for all observation points J (j\1) when the observation point is 1, and the results are stored in the memory 20.

The method described above in which a plurality of observation devices are connected to measure the entire object position Ii can be operated in principle if there are two observation devices. However, by increasing the number of observation devices to three or more, it becomes possible to improve accuracy, improve reliability, and realize cooperative operations.

If there are multiple observation devices in a trench, it is possible to calibrate the mutual position and direction of each combination of observation devices and other devices, and measure the position of unknown objects, and process the results by averaging them all. This makes it possible to improve calibration accuracy and object position measurement accuracy.

Secondly, if there are multiple observation devices, even if one or more of them breaks down and becomes inoperable, two or more remaining devices will continue to function, improving overall reliability.

In addition, cooperative operation of multiple observation devices is possible. For example, if an observation device measures the distance in terms of object weight from its observation point, and you want to apply a predetermined 416 operation only to the device with the smallest distance, the following can be done.

Assume that the first observation device calculates the position vector r at the object by the above-described operation. The determination circuit 23 in FIG.
uk Mr extrusion for all j, lr r++l
The minimum value of ``Irl'' is compared, and the device that differs from the minimum value is determined to be the device closest to the target object.

If it is determined that the device 1 is the closest, the operating device 21 is caused to perform seven predetermined operations.

The above-mentioned calibration, that is, the calculation of the mutual positional relationship of the sensors, and the calculation of the (Sγ1N measurement of the unknown object) are performed by importing the signals from multiple sensors into a central processing unit (not shown) and sizing them up. However, using a distributed processing method in which processing is performed for each i1+11 device as in the present embodiment has the effect of increasing overall reliability.

As described above, according to the present invention, from the fNγ1F7 report measured from different phases 1jt by a plurality of observation instruments Wr,
The unknown position of an object can be estimated by three-dimensional geometric operations. This allows for higher precision in position measurement of distant objects than measurement using conventional optical 1i'4"i distance meters. Also, direction measurement can be performed with higher accuracy using a simpler device than distance measurement. In addition, according to the calibration method of the present invention that uses a calibration object, the position and direction of each observation device can be calculated, so there is no need for strict surveying for calibration.
It becomes possible to move and install the observation device at the location of the child. By using a plurality of observation devices in combination with the awn, it is possible to increase the accuracy of position measurement according to the number of devices.

It is also possible to measure the position of a moving object by repeating the physical position measurement every scent cycle.

[Brief explanation of drawings]

Figure 1 is a diagram showing the principle of measuring object position from azimuth observation, Figure 2 is a diagram showing the relationship between the image position on the sensor surface and the line of sight, and Figure 3 is a diagram showing the relationship between the image position on the sensor surface and the line of sight. An overall view of the apparatus, FIG. 4 is a block diagram of the observation apparatus which is a component of FIG. 3.

Claims (1)

[Claims] 1. Measuring means that includes a light receiving sensor and a calculation circuit and uses the light receiving sensor to observe an image of a target object and calculate the orientation information of the object is distributed at multiple points, and the measuring means is distributed at multiple points. Exchanging direction information between measuring means,
An object position measurement method in which the azimuth and distance 'kN of the target object weight are calculated by the arithmetic circuit indicated in - for each measuring means.ff:%.