JP5035448B2 - Magnetic sensor control apparatus and method - Google Patents

Description

  The present invention relates to a magnetic sensor control apparatus and method.

Conventionally, a three-dimensional magnetic sensor that is mounted on a moving body and detects the direction of geomagnetism is known. In general, a three-dimensional magnetic sensor includes three magnetic sensor modules for detecting a scalar quantity by decomposing a magnetic field vector into three orthogonal components. The magnetic data that is the output of the three-dimensional magnetic sensor has three components because it consists of a combination of the outputs of such three magnetic sensor modules. The direction and magnitude of the vector having magnetic data as the component are the direction and magnitude of the magnetic field detected by the three-dimensional magnetic sensor. When the direction or magnitude of geomagnetism is specified based on the output of the three-dimensional magnetic sensor, a process for correcting the output of the three-dimensional magnetic sensor is necessary to cancel the magnetization component of the moving body. The operation value of this correction process is called offset. The offset represents a magnetic field vector generated by magnetization of the moving body detected by the three-dimensional magnetic sensor, and the magnetizing component of the moving body is obtained by subtracting the offset from the magnetic data that is the output of the three-dimensional magnetic sensor. Be countered. An offset can be calculated by obtaining the center of a spherical surface that passes through a point having magnetic data as a component.
By the way, a set of points whose components are magnetic data is not a perfect sphere in reality. This is because the output of the three-dimensional magnetic sensor itself has a measurement error according to a Gaussian distribution, and in reality there is no completely uniform magnetic field, so magnetic data necessary for calculating the offset is accumulated. This is because the magnetic field measured by the three-dimensional magnetic sensor fluctuates during the period, and there is a calculation error until the output of the three-dimensional magnetic sensor is extracted as a digital value.
The conventional offset calculation method of a magnetic sensor is to accumulate a large number of magnetic data and calculate the most probable offset by their statistical processing (see, for example, Patent Document 1). However, when statistical processing of a large number of magnetic data is performed, it is possible to calculate an accurate offset if the distribution of the magnetic data group as a population is uniformly wide and unique magnetic data is not excluded from the population. Can not. Therefore, it is necessary to select a population to some extent in order to calculate an accurate offset, and after all, it is not possible to calculate an accurate offset only by statistical processing. In addition, the amount of processing for obtaining the most probable spherical center by statistical processing is considerable, and it consumes a lot of time and resources.

  It is an object of the present invention to provide a magnetic sensor control device, method, and program that do not require the use of statistical processing for offset calculation.

(1) A magnetic sensor control device for achieving the above object includes an input means for inputting a plurality of magnetic data having three components, which are sequentially output from a three-dimensional magnetic sensor, and a plurality of the inputted magnetic data. A selection means for selecting the four magnetic data satisfying a predetermined four-point selection condition, and a center point which is a point equidistant from the four points having the selected four magnetic data as components. Calculating means; and setting means for setting a component of the center point as an offset of the magnetic data.
Four magnetic data satisfying a predetermined four-point selection condition are selected from magnetic data sequentially output from the three-dimensional magnetic sensor, and the points are equidistant from the four points having the selected four magnetic data as components. By calculating the center point that is, it is not necessary to use statistical processing for calculating the offset, so that the offset can be calculated efficiently. It is also possible to evaluate the reliability of the calculated offset by statistical processing.

(2) The magnetic sensor control device may further include a buffer for storing the plurality of input magnetic data. The selection means selects the (n−1) th magnetic data satisfying a predetermined selection condition before selecting the n (n = 2, 3 or 4) th magnetic data, and selects n−1. The magnetic data that has not been selected as the first magnetic data may be acquired from the buffer and may be selected as the nth magnetic data selection candidate.
In order to select four magnetic data satisfying the four-point selection condition from a large number of magnetic data at once, the number of selection trials greatly varies, and the maximum processing amount required for selection may become enormous. Therefore, the selection process is set in stages, and the n-1st magnetic data satisfying another selection condition is selected before the nth magnetic data is selected under a certain condition, whereby the maximum processing required for selection is performed. The amount can be reduced. Further, when selecting the n-th magnetic data under a certain condition, the magnetic data that has not been selected as the magnetic data up to the (n−1) -th magnetic data is set as a selection candidate again, so that magnetic data necessary for calculating the offset is obtained. Can be reduced. As a result, the amount of operation of the moving body equipped with the three-dimensional magnetic sensor required for collecting magnetic data necessary for calculating the offset of the magnetic data is reduced, and the offset calculation time is reduced. Shortened.

(3) The selection means stores the input magnetic data in the three data structures for storing the three magnetic data of the selection candidates, and stores them in the three data structures. The data structure may be updated until the three magnetic data satisfy a predetermined three-point selection condition, and the three magnetic data satisfying the three-point selection condition may be selected.
By selecting up to the third magnetic data before selecting the fourth magnetic data satisfying the four-point selection condition, the maximum processing amount required for selection can be reduced. In the process of selecting the third magnetic data, three data structures capable of storing three magnetic data, in which the magnetic data is updated one by one, can be used. A data structure for storing magnetic data can be defined as an array having, for example, three variables for storing three components of magnetic data.

(4) When the selection means stores the last input magnetic data in any of the three data structures, the selection means updates the last input magnetic data with the three data structures. The one data structure is selected so that the length of the chord of the arc passing through the three points including the three magnetic data stored in the body is extended and at least not shortened, and the selected data structure is The magnetic data inputted last may be updated.
A circle passing through three points having three magnetic data components is the circumference of a sphere cross-section centered on a point whose component is an offset finally obtained. In order to bring the cross-sectional circle closer to the cross-sectional circle of a sphere centered on a point having an accurate offset as a component, the area of the cross-sectional circle is large, and the three points that define the cross-sectional circle are wide on the circumference It is desirable to be evenly distributed. Accordingly, it is desirable that the length of the string passing through the three points having the three magnetic data components is longer. Therefore, when storing the last input magnetic data in any of the three data structures, the length of the chord of the arc passing through the three points whose components are the three magnetic data stored in the three data structures. It is desirable that the data structure to be updated is selected so that the length becomes longer due to the update of the data structure.

(5) The selection unit may reset the three data structures when the number of updates of the three data structures exceeds a predetermined number.
In the process of selecting three magnetic data satisfying the three-point selection condition, if the magnetic field in which the magnetic sensor exists is changed drastically after the second magnetic data is selected, the three-point selection condition is satisfied. Magnetic data may not be input indefinitely. In order to prevent such a problem, it is desirable to reset the three data structures once when the number of updates of the three data structures exceeds a predetermined number. As a result, since the magnetic data is stored in three empty data structures for the changed magnetic field, the problem that the three magnetic data satisfying the three-point selection condition are not selected indefinitely does not occur.

(6) The three-point selection condition may include a condition that correlates with a distortion with respect to a regular triangle of a triangle having three points as vertices having three magnetic data stored in the data structure as components.
A circle passing through three points having three magnetic data components is the circumference of a sphere cross-section centered on a point whose component is an offset finally obtained. In order to make the cross-sectional circle close to the cross-sectional circle of a sphere centered on a point having an accurate offset as a component, it is desirable that the three points defining the cross-sectional circle are dispersed at equal intervals on the circumference. Therefore, it is desirable to set a three-point selection condition for the distortion of a triangular triangle having three points with three magnetic data as components. When a triangle whose apex is three points having three magnetic data components is a regular triangle, the three points are distributed at equal intervals on the circumference.
The distortion of a triangle with respect to a regular triangle can be defined by paying attention to some attributes of the triangle.

(7-9) For example, the three-point selection condition may include that the variation in the internal angle of the triangle is within a predetermined range. Further, for example, the three-point selection condition may include that the variation in the length of the sides of the triangle is within a predetermined range. Further, for example, the three-point selection condition may include that a variation in distance from the center of gravity of the triangle to each vertex is within a predetermined range. .
Variations in values can be arbitrarily defined mathematically, but in consideration of the amount of computation and processing time required for condition determination, and the possibility of reusing calculation results obtained in the condition determination process. It is desirable.

ただし、
ｐ_i（ｉ＝０，１，２）は前記三角形の各頂点であり、ｄ₃は前記三角形の重心である。 (10) Therefore, it is desirable that the three-point selection condition includes that the ratio of the intermediate eigenvalue to the maximum eigenvalue of the following 3-by-3 symmetric matrix A is not less than a predetermined value.

However,
p _i (i = 0, 1, 2) is each vertex of the triangle, and d ₃ is the center of gravity of the triangle.

(11) The four-point selection condition may include a condition that correlates with a distortion of a tetrahedron with four points as vertices having the four magnetic data as selection candidates as a vertex.
Four points having four magnetic data as components define a spherical surface centered on a point having the offset finally obtained as a component. In order to bring the spherical surface closer to a spherical surface centered on a point having an accurate offset as a component, it is desirable that the four points defining the spherical surface are uniformly distributed over a wide range on the spherical surface. When the four points that define the spherical surface are evenly distributed on the spherical surface, a tetrahedron with these four points as vertices is a regular tetrahedron. Therefore, it is desirable to set a four-point selection condition for the distortion of a tetrahedron with four points as vertices having four magnetic data components as selection candidates.
The distortion of a tetrahedron with respect to a regular tetrahedron can be defined by paying attention to some attributes of the regular tetrahedron.

(12-14) For example, the four-point selection condition may include a variation in solid angle of each vertex of the tetrahedron being within a predetermined range. Further, for example, the four-point selection condition may include that the variation in the area of each surface of the tetrahedron is within a predetermined range determined in advance. Further, for example, the four-point selection condition may include that a variation in distance from the center of gravity of the tetrahedron to each vertex is within a predetermined range determined in advance.
As already mentioned, the dispersion of values can be arbitrarily defined mathematically, but it takes into account the amount of computation and processing time required for condition judgment and the possibility of reusing the calculation results obtained in the condition judgment process. It is desirable to define

ただし、
ｐ_i（ｉ＝０，１，２）は前記四面体の各頂点であり、ｄ₄は前記四面体の重心である。 (15) Therefore, it is desirable that the four-point selection condition is that the ratio of any two sets of the eigenvalues of the following 3-by-3 symmetric matrix B is larger than a predetermined value. . Only one combination of threshold values (predetermined values) for the ratio of one set of eigenvalues may be determined, or a plurality of combinations of threshold values may be determined and the ratio of each of the two sets of eigenvalues. It is good also as 4 point selection conditions that each is larger than two threshold values corresponding to any combination.

However,
p _i (i = 0, 1, 2) is each vertex of the tetrahedron, and d ₄ is the center of gravity of the tetrahedron.

  The functions of the plurality of means provided in the present invention are realized by hardware resources whose functions are specified by the configuration itself, hardware resources whose functions are specified by a program, or a combination thereof. The functions of the plurality of means are not limited to those realized by hardware resources that are physically independent of each other. The present invention can be specified not only as a magnetic sensor control device, but also as a magnetic measurement device combining a magnetic sensor control device and a three-dimensional magnetic sensor, as a magnetic sensor control method, and as a magnetic sensor control program. is there.

The flowchart which concerns on embodiment of this invention. The schematic diagram which concerns on embodiment of this invention. The schematic diagram which concerns on embodiment of this invention. The block diagram which concerns on embodiment of this invention. The block diagram which concerns on embodiment of this invention. The flowchart which concerns on embodiment of this invention. The schematic diagram which concerns on embodiment of this invention. The schematic diagram which concerns on embodiment of this invention. The schematic diagram which concerns on embodiment of this invention. The schematic diagram which concerns on embodiment of this invention. The schematic diagram which concerns on embodiment of this invention. The schematic diagram which concerns on embodiment of this invention.

Hereinafter, embodiments of the present invention will be described in the following order.
[1. Overall explanation]
1-1. Basic principle 1-2. Hardware configuration 1-3. Software configuration [2. Process flow]
2-1. Buffer update 2-2.1 point selection candidate tentative decision 2-3. Second point selection candidate tentative decision 2-4.3 selection up to 3rd point ・ 3 point update ・ 3 point selection condition ・ Selection verification ・3-point reset 2-5.4 Selection of the 4th point ・ Overview ・ 4-point selection conditions ・ Calculation of azimuth sphere ・ Verification of azimuth sphere ・ Offset setting

[1. Overall overview]
1-1. Basic Principle First, the basic principle used by the present embodiment will be described.
In the present embodiment, an offset for canceling the magnetization component of the moving body on which the three-dimensional magnetic sensor is mounted is set. In order to obtain the offset, a spherical surface that approximates a solid surface that is a collection of innumerable points in a vector space whose components are magnetic data that is output from a three-dimensional magnetic sensor is specified. This spherical surface is referred to as an azimuth spherical surface. The component of the center point of the azimuth sphere is an offset, and its radius is the strength of the magnetic field. Since there is a measurement error in the output itself of the three-dimensional magnetic sensor, even if the moving body rotates with a constant magnetic field, the set of points having magnetic data as a component does not become a spherical surface. In the present embodiment, four points having magnetic data as components are selected, and from the selected four points, the center point and radius of a spherical surface passing through them are obtained, and the obtained spherical center is set as an offset. In order to reduce the error due to the measurement error of the three-dimensional magnetic sensor itself included in the offset by this method, four points including four magnetic data containing the measurement error of the three-dimensional magnetic sensor itself as components It is important to select so that it is evenly distributed over a wide range on a spherical surface passing through four points. Hereinafter, in order to facilitate understanding, a description will be given by replacing it with a two-dimensional magnetic sensor.

  FIG. 2 shows an azimuth circle (solid line) that passes through three points whose components are magnetic data, which is the output of a two-dimensional magnetic sensor, and an infinite number of points, whose components are magnetic data of an ideal two-dimensional magnetic sensor that has no measurement error. It shows the relationship with the true azimuth circle (dashed line). When the three points whose components are magnetic data of a two-dimensional magnetic sensor are distributed in a narrow range on the circumference passing through them, the center of the azimuth circle (×) obtained from these three points and the ideal two-dimensional to be obtained The distance from the center (+) of the true azimuth circle of the magnetic sensor increases as shown in FIG. On the other hand, when three points whose components are magnetic data of a two-dimensional magnetic sensor are evenly distributed over a wide range on the circumference passing through them, the center (+) of the azimuth circle obtained from those three points, The distance from the center (x) of the true azimuth circle of the ideal two-dimensional magnetic sensor to be obtained is small as shown in FIG. When the three points whose components are magnetic data of the two-dimensional magnetic sensor are distributed at equal intervals around the circumference passing through them, that is, when the triangle having the three points as vertices becomes an equilateral triangle, the offset error is the most. Can be small.

  In the three-dimensional magnetic sensor, when the tetrahedron having four points selected for obtaining the azimuth is a regular tetrahedron, the offset error can be minimized. Therefore, in the present embodiment, the four points having the magnetic data as components are selected so that the distortion of the tetrahedron having them as a vertex with respect to the regular tetrahedron is smaller than a specific threshold value.

  Here, four points are selected in advance, a spherical surface passing through the selected four points is obtained, and the method of this embodiment for obtaining an offset from the spherical surface, and a spherical surface approximate to a solid surface passing through a large number of five or more points. It is obtained by a statistical method and compared with a method for obtaining an offset from the spherical surface. In the method using the statistical method, if the points having magnetic data as a component are unevenly distributed or the magnetic population includes peculiar magnetic data, the center of the azimuth sphere obtained from these points is the true azimuth sphere. Deviation from the center. Therefore, if it is intended to reduce the offset error using a statistical method, it is necessary to verify the variation in the population for obtaining the bearing sphere using the statistical method, or to select the population. In the method of the present embodiment, if appropriate four points can be selected, an offset having an accuracy comparable to the offset accuracy obtained by the method using the statistical method can be used without using a statistical method having a large amount of calculation. Can be requested.

1-2. Hardware Configuration FIG. 3 is a schematic diagram showing an appearance of a mobile phone 3 which is an example of a mobile body to which the present invention is applied. A three-dimensional magnetic sensor 4 is mounted on the mobile phone 3. A three-dimensional magnetic sensor detects the direction and strength of a magnetic field by detecting vector components of magnetic fields in three directions of x, y, and z orthogonal to each other. Various information such as characters and images is displayed on the display 2 of the mobile phone 3. For example, the display 2 displays a map and an arrow or character indicating the direction.

  FIG. 4 is a block diagram showing a magnetic measurement device including the three-dimensional magnetic sensor 4 and the magnetic sensor control device 1 to which the present invention is applied. The three-dimensional magnetic sensor 4 is mounted on various moving bodies and includes an x-axis sensor 30, a y-axis sensor 32, and a z-axis sensor 34 that respectively detect an x-direction component, a y-direction component, and a z-direction component of a magnetic field vector due to geomagnetism. It has. The x-axis sensor 30, the y-axis sensor 32, and the z-axis sensor 34 are all configured by a magnetoresistive element, a Hall element, etc., and may be any one as long as it has directivity. The x-axis sensor 30, the y-axis sensor 32, and the z-axis sensor 34 are fixed so that their sensitivity directions are orthogonal to each other. Outputs of the x-axis sensor 30, the y-axis sensor 32, and the z-axis sensor 34 are time-divisionally input to a magnetic sensor I / F (Inter Face) 22. In the magnetic sensor I / F 22, input from the x-axis sensor 30, the y-axis sensor 32, and the z-axis sensor 34 is amplified and then AD converted. Digital magnetic data output from the magnetic sensor I / F 22 is input to the magnetic sensor control device 1 via the bus 5.

  The magnetic sensor control device 1 is a so-called computer including a CPU 40, a ROM 42, and a RAM 44. The CPU 40 is a processor that controls the entire mobile body such as the mobile phone 3. The ROM 42 is a non-volatile storage medium that stores a magnetic sensor control program executed by the CPU 40 and various programs for realizing the function of the moving body, for example, a navigation program. The RAM 44 is a volatile storage medium that temporarily holds data to be processed by the CPU 40. The magnetic sensor control device 1 and the three-dimensional magnetic sensor 1 can be configured as a one-chip magnetic measurement device.

1-3. Software Configuration FIG. 5 is a block diagram showing the configuration of the magnetic sensor control program 90. The magnetic sensor control program 90 is a program for providing azimuth data to the navigation program 98 and is stored in the ROM 42. The azimuth data is two-dimensional vector data indicating the direction of geomagnetism. The azimuth data may be provided to the navigation program 98 as three-dimensional vector data. When the CPU 40 executes the magnetic sensor control program 90, the CPU 40, the RAM 44, and the ROM 42 function as a magnetic sensor control device. The magnetic sensor control program 90 is composed of a module group such as a buffer management module 92, an offset setting module 94, and an azimuth calculation module 96.

  The buffer management module 92 is a program component that stores magnetic data sequentially input from the magnetic sensor 4 in a buffer for use in offset calculation, and causes the CPU 40 to function as input means. The magnetic data sequentially input from the magnetic sensor 4 can be sequentially selected, and the magnetic data that has not been selected can be discarded. However, if the selection conditions for the nth point (n = 2, 3) and the selection conditions for the n + kth point (k = 1, 2) are different, even if the magnetic data is not selected as the nth point, n + k Can be selected as a point. Therefore, the buffer managed by the buffer management module 92 is used not only for buffering the difference between the input timing and the processing timing but also for the purpose of making magnetic data that has not been selected once available again as a selection candidate. The buffer may be configured by hardware or software.

  The offset setting module 94 is a program component that selects four magnetic data from the magnetic data held by the buffer management module 92, calculates an offset from the selected four magnetic data, and sets the calculated offset. The CPU 40 is caused to function as selection means, calculation means, and setting means.

  The azimuth calculation module 96 is a program component that generates azimuth data using magnetic data sequentially input from the magnetic sensor 4 and a set offset. Specifically, the azimuth calculation module 96 outputs two or all three of the three components obtained by subtracting each component of the offset data from each component of the magnetic data that is three-dimensional vector data as azimuth data. To do.

  The navigation program 98 is a well-known program that searches for a guidance route to a destination and displays the guidance route on a map. For ease of map recognition, the map is displayed on the screen so that the orientation on the map matches the actual orientation. Therefore, for example, when the mobile phone 3 rotates, the map displayed on the display 2 rotates with respect to the display 2 so as not to rotate with respect to the ground. Direction data is used for such map display processing. Of course, the azimuth data may be used only for displaying the north-south-north-west with characters or arrows.

[Process flow]
2-1. Buffer Update FIG. 6 is a flowchart showing the flow of buffer update processing. The process shown in FIG. 6 proceeds when the CPU 40 executes the buffer management module 92 when an offset update request is generated.
In step S100, a buffer for storing a plurality of magnetic data used for offset calculation is initialized. Specifically, for example, data stored in the buffer area of the RAM 44 is deleted.

In step S102, input of new magnetic data from the magnetic sensor 4 is awaited.
When new magnetic data is input from the magnetic sensor 4, it is determined whether the buffer needs to be updated (step S104). In the situation where the mobile phone 3 is hardly moving, if there is an input from the magnetic sensor 4 sequentially at short time intervals, the distance between two points having two magnetic data that are continuously input as components becomes close. It is not desirable that two points that are close to each other are included in the four points that are necessary for obtaining the offset. In addition, storing a plurality of magnetic data whose magnetic data components are close to each other in a buffer having a limited capacity is a waste of memory resources and causes unnecessary buffer update processing. . Therefore, the necessity for updating the buffer is determined as follows. For example, if the distance between the point having the magnetic data input immediately before and the point having the magnetic data input last as the component is smaller than a certain threshold value, it is determined that there is no need to update the buffer. The last input magnetic data is discarded without being stored in the buffer. Otherwise, it is determined that there is a need to update the buffer. As a result, for example, when magnetic data is input in the order shown in FIG. 7, the magnetic data represented by black circles is not stored in the buffer, and the magnetic data represented by white circles is stored in the buffer. In FIG. 7, the input order of magnetic data is represented by white circles and black circles.

If the following conditions are further added to the above determination conditions to determine whether or not the buffer needs to be updated, the probability of successful calibration can be increased. Here, it is assumed that N _qmax data strings stored in the buffer are q ₀ , q ₁ ,... Q _Nmax−1 , and the set Q is Q = {qi | 0 ≦ I ≦ N _qmax −1}.
Additional update condition: For a certain threshold value q _min , only when the point q having magnetic data as a component satisfies || qq _i || ₂ > q _min for all i where i <N _q . Store the last magnetic data and update the buffer.

Note that the failure of calibration means that an offset update request is generated and magnetic data accumulation is started, and as a result that the offset cannot be obtained from the accumulated magnetic data, it is accumulated until then. It means to discard all magnetic data.
If it is determined that there is a need to update the buffer, the buffer is updated (step S106). The update algorithm is, for example, FIFO (First In First Out).

FIG. 1 is a flowchart showing a flow of offset setting processing. The processing shown in FIG. 1 proceeds when the CPU 40 executes the offset setting module 94 when an offset update request is generated. When the processing described below is progressing by executing the offset setting module 94, the processing by the buffer management module 92 is proceeding in parallel in the multitasking environment.

In step S200, initialization for discarding all selection candidates is executed. Specifically, the four data structures p ₀ , p ₁ , p ₂ and p ₃ for storing selection candidates declared by the offset setting module 94 are reset and the buffer is also initialized. The data structures p ₀ , p ₁ , p ₂ , and p ₃ store magnetic data as selection candidates until the fourth point is selected and an offset is set, as will be described later. Each data structure p ₀ , p ₁ , p ₂ , p ₃ is an array of three variables for storing three components of magnetic data.

In step S202, the above-described buffer update is awaited.
When the buffer is updated, the first selection candidate is provisionally determined (step S204). That is, the magnetic data input first after the buffer initialization (step S200) is the first selection candidate. The first selection candidate is stored in the data structure p ₀ . Since the data structure p ₀ in which the first selection candidate is stored may be updated as described later, the first selection candidate is provisionally determined in this process. Not too much.

2-3. Temporary Selection Candidate Selection for Second Point In step S206, buffer update is awaited again.
When the buffer is updated again, the second selection candidate is provisionally determined. That is, the second input magnetic data after buffer initialization (step S200) becomes the second selection candidate. The second selection candidate is stored in the data structure p ₁ . Since the data structure p _{1 in} which the second selection candidate is stored may be updated as described later, the second selection candidate is temporarily determined in this process. Not too much.

2-4.3 Selection up to the third point-Update of the three points In step S210, the reset counter is reset. The reset counter is set to count up at a predetermined count. This reset counter is used in order to prevent the occurrence of a problem that the calculation of the offset is not completed indefinitely without the third selection candidate being determined after the second selection candidate is provisionally determined. Specifically, if the magnetic field in which the mobile phone 3 exists immediately after the provisional determination of the second selection candidate is changed drastically, the selection candidates up to the second point are not updated. There is a possibility that the selection candidate of the eyes is not determined. As will be described later, the third selection candidate is determined unless the size of the circle passing through the three selection candidates falls within a certain range and the triangle having the three points as vertices does not satisfy the predetermined condition. This is because these conditions are not satisfied if the strength of the magnetic field changes drastically.

In steps S212 and S214, the reset counter is incremented when new magnetic data is input. That is, the number of magnetic data input after the second selection candidate is provisionally determined in step S208 for the first time is counted.
In step S216, buffer update is awaited again.
When the buffer is updated again, in step S218, it is held as a candidate to be selected according to the three-point selection condition, so that the magnetic data last input and stored last in the buffer is up to the third existing point. It is determined which of the selection candidates should be updated and retained (if selection using the three-point selection condition has not been executed, only selection candidates up to the second point exist) However, the processing content to be executed is the same). That is, it is determined in which data structure p ₀ , p ₁ , p ₂ the magnetic data last input and stored last in the buffer is stored. This is because the data structures p ₀ , p ₁ , and p ₂ have different meanings as follows.

The three points whose components are magnetic data stored in the data structures p ₀ , p ₁ , and p ₂ uniquely define an arc passing through them as long as they are not on the same straight line. The data structures p ₀ and p ₁ are meant as data structures that store magnetic data that are components of points at both ends of the arc. The data structure p ₂ is meant as a data structure that stores magnetic data that is a component at the midpoint of the arc. Therefore, until the third point is selected, in order to hold the selection candidates up to the third point, the magnetic data last input and stored last in the buffer is any data structure p ₀ , p ₁ , p Whether it is stored in ₂ must be determined.

By the way, the reason why the meaning is given to the data structures p ₀ , p ₁ and p ₂ is as follows. The three selection candidates are located on the spherical surface of an azimuth sphere that passes through the four selection candidates. As described above, the offset error becomes smaller as the four points used for obtaining the azimuth sphere are evenly distributed over a wide range on the spherical surface passing through these four points. Accordingly, the longer the chord length of the arc passing through the three selection candidates is, the smaller the offset error becomes. If the chord length is too short, the offset error becomes larger. For this reason, it is desirable to update the selection candidates up to the third point each time the magnetic data is stored in the buffer so that the chord length of the arc passing through the three points that are selection candidates is increased to some extent. .

FIG. 8 is a schematic diagram for explaining the update of three selection candidates. Now, it is assumed that the magnetic data shown in FIG. 8B is stored in the data structures p ₀ , p ₁ and p ₂ . (The numbers 1, 4, 5, and 6 shown in FIG. 8 are numbers for managing the storage order in the buffer, and the magnetic data managed in the buffer by that number is stored in each data structure. Here, the magnetic data managed by each number is referred to as magnetic data 1, magnetic data 4, magnetic data 5, and magnetic data 6.) In this state, selection candidates The arc chord passing through the three points is a line segment (broken line) connecting a point having the magnetic data 1 as a component and a point having the magnetic data 5 as a component. Next, it is assumed that the magnetic data 6 is stored in the buffer. At this time, an arc chord through which three points having three magnetic data components pass is defined as p ₀ , p ₁ , and p ₂ , and a point having the magnetic data stored in p ₁ as a component and p It is calculated as a distance from a point having magnetic data stored in ₀ as a component. Therefore, if the positional relationship between the point having the magnetic data 6 as a component and the three points having the magnetic data 1, 4, 5 as the component is in the state shown in FIG. 8A, it is stored in p _0. If the selected candidate is not updated, the calculation result that the chord of the arc has become long cannot be obtained. Also if, when the magnetic data 6 has a value which does not lengthen the arc chord, unless update the selection candidate stored in the p _2, to become shorter arc strings as the calculation result, the p ₂ The stored selection candidates need to be updated.

A condition that the chord of the circular arc through which three points having three magnetic data components pass is updated at least efficiently without being shortened at each update is derived as follows. Let q be the last magnetic data stored in the buffer. Here, p ₀ , p ₁ , and p ₂ are magnetic data stored in each data structure. In FIG. 9, points having q, p ₀ , p ₁ , and p ₂ as components are represented by black circles, an azimuth sphere passing through these four points is represented by a solid line, passes through p ₀ and p _1, and a line segment p _0. Two planes S ₀ and S ₁ perpendicular to p ₁ are represented by broken lines. The range in which the chord of the arc passing through the three points that are candidates for selection becomes longer by q is a point having q as the component outside the space sandwiched between the planes S ₀ and S ₁ and penetrated by the line segment p ₀ p ₁ . When there is. At this time, the following expressions (3) and (4) hold.

If equation (3) holds, the string can be extended by updating p ₂ to p ₁ and newly updating p ₁ with q. If equation (4) holds, the string can be extended by updating p ₂ to p ₀ and newly updating p ₀ with q. If neither of these, updating p ₂ with q, but the strings are not extended.

In step S220, the selection candidate to be updated, determined as described above, is updated with the magnetic data stored last in the buffer. As a result, for example, the data structures p ₀ , p ₁ , and p ₂ storing the magnetic data shown in FIG. 8B are updated as shown in FIG. 8C. At this time, the circle passing through the three points that are selection candidates is updated to a new circle included in the plane intersecting the plane including the circle. Also, this process tends to increase the area of a circle passing through the three points that were candidates for selection.

Three-point selection condition In step S222, it is determined whether the magnetic data that is the three selection candidates satisfies one of the three-point selection conditions. The three selection conditions are derived from two viewpoints. The first point of view is that the three points whose components are magnetic data that are candidates for selection are sufficiently dispersed on the circumference passing through these three points, that is, the triangle having the three points as vertices is close to an equilateral triangle. This is the point of view. Therefore, when selection from the first viewpoint is executed, the distortion with respect to a regular triangle of three points having apexes of magnetic data as selection candidates as components is calculated. The second point of view is whether the area of a circle passing through the three points whose components are magnetic data that are selection candidates is sufficiently large. When the area of the circle passing through these three points is small, an unexpected magnetic field is measured when calculating the exact offset or whether the three points are distributed in a narrow range on the spherical surface of the azimuth sphere passing through these three points. Will be. If the area of the circle passing through these three points is too large, an unexpected magnetic field is measured in calculating an accurate offset. Therefore, when selection is performed from the second viewpoint, the radius of a circle passing through three points whose components are magnetic data that are selection candidates is calculated. In the following description, the condition for determining whether a selection candidate should be selected from the first viewpoint is also simply referred to as a three-point selection condition, and it is determined whether the selection candidate should be selected from the second viewpoint. This is also referred to as verification of 3-point selection.

First, the three-point selection condition will be described from the first viewpoint applied in step S222. In this embodiment, as shown in FIG. 10, the distortion with respect to the regular triangle of the triangle having three points as the vertices, the magnetic data of which is a candidate for selection, as the starting point, the centroid d ₃ of the triangle as the starting point, and each vertex as the ending point. It is evaluated by the eigenvalues of a 3-by-3 symmetric matrix expressed using the sum of vectors. The center of gravity d ₃ of the triangle is obtained by the following equation (5).

The three-point selection condition derived from the first viewpoint is that the ratio of the intermediate eigenvalue λ _{A2 to} the maximum eigenvalue λ _A1 among the eigenvalues λ _A1 , λ _A2 , and λ _A3 of the symmetric matrix A defined by the following equation (6) The predetermined threshold value E ₀ or more.

ただし、以下の通りである。

Equation (6) can also be written as Equation (7) when pi = (pix, piy, piz), d ₃ = (d3x, d3y, d3z).

However, it is as follows.

From the definition of the symmetric matrix A, all eigenvalues are non-negative real numbers. If the calculation error is not taken into consideration, the minimum eigenvalue λ _A3 is zero. As described above, the three-point selection condition derived from the first viewpoint is that the evaluation amount E _A defined by the following equation (14) satisfies E _A ≧ E ₀ with respect to the threshold value E ₀ . is there.

By the way, the distortion with respect to the regular triangle of the triangle whose vertex is the three points whose magnetic data is the candidate for selection is the variation of the internal angle of the triangle, the variation of the length of the side of the triangle, and the centroid of the triangle. It can be defined by the variation of the distance up to. Further, these variations can be defined by the evaluation amount E _A using the eigenvalues of the symmetric matrix A described above. Therefore, the three-point selection condition derived from the first point of view is a variation in the internal angle of a triangle whose apex is the three points whose components are magnetic data as selection candidates, or a variation in the length of the side of the triangle, or It can also be said that the variation in distance from the center of gravity of the triangle to each vertex is within a predetermined range.

If the three-point selection condition derived from the first viewpoint is not satisfied, it is determined whether the reset counter is counting up (step S224), and if not counting up, the process returns to step S216, Repeat the process described above. That is, the update process of the data structures p ₀ , p ₁ , and p ₂ for making the magnetic data finally stored in the buffer as a new selection candidate is repeated. When counting up, the selection up to the third point may not be completed indefinitely due to a sudden change in the magnetic field in which the mobile phone 3 exists immediately after the temporary selection of the second selection candidate Therefore, returning to the process of step S200, the buffers and the data structures p ₀ , p ₁ and p ₂ are all reset.

Selection Verification When the three-point selection condition derived from the first viewpoint is satisfied, the magnetic data selected from the first viewpoint is verified from the second viewpoint (step S226). In this process, as described above, the radius of a circle passing through three points whose components are magnetic data that are selection candidates is calculated. Various methods for calculating such a radius are known. In this embodiment, a method using the eigenvectors of the symmetric matrix A is adopted in order to reduce the overall calculation amount and speed up the processing. . If this method is employed and the Jacobian method is employed as a method for calculating the above eigenvalues of the symmetric matrix A, the total amount of calculation is reduced. This will be specifically described below.
When p ₀ , p ₁ , and p ₂ are rewritten in the coordinate system based on the eigenvector of the matrix A, the component of the eigenvector l _A3 with respect to the minimum eigenvalue λ _A3 takes a constant value for each point. focusing only on the components of the eigenvector l _A2 for eigenvector l _A1 and the intermediate eigenvalue lambda _A2 with respect to the maximum eigenvalue lambda _A1 normalized, is regarded each point and three points on the two-dimensional space, the equation of the circle by simultaneous linear equations Can be sought. This will be described more specifically.

First, the matrix L is defined by the following equation (15). The matrix L is obtained in the process of calculating the above eigenvalues of the symmetric matrix A using the Jacobian method.

For the point p _i (0 ≦ i ≦ 2), the linear transformation of the following equation (16) is performed, the third component of p ′ _i is ignored, and only the first component and the second component are used.

Thus, although the following equation (17) is derived, for the right side third term can be ignored, 3 points in two-dimensional space _{p "0 = (p '01} , p' 02), p" 1 = (p '11 , p is resulting in _{'12), p "2 =} (p' 21, p '22) to determine the circle passing through.

When the center point of the circle _{c "3 = (c '31} , c' 32) and is, simultaneous linear equations for obtaining the center point is the following equation (18).

c ′ _3r is a value that does not need to be used in particular, but is added to establish a ternary linear simultaneous equation. The radius r ₃ is finally obtained as the Euclidean distance between c ′ ₃ and p ′ ₀ calculated by the following equation (19).

If the radius of the circle passing through the three points, whose components are the three magnetic data satisfying the three-point selection condition, determined as described above is within the range from the threshold value r _max to r _min , the test is accepted. . Otherwise, it fails and returns to the process of step S200, and the buffers and data structures p ₀ , p ₁ , p ₂ are reset.

2-5.4 Selection of the fourth point ・ Outline When the third point is selected by the above processing, the point is sufficiently scattered on the circumference of a sufficiently large cross-sectional circle of the azimuth circle finally obtained 3 A point is selected. Using the magnetic data whose component is one point that is not on the plane including those three points, an equation of a sphere passing through these four points can be obtained. However, if the fourth point is randomly selected, the offset error will not be reduced. Therefore, in the present embodiment, in order to reduce the error of the offset, the optimum fourth point selection is tried for a plurality of fourth point selection candidates. That is, selection is attempted using all the magnetic data stored in the buffer as fourth selection candidates. If there is an empty space in the buffer, the fourth selection is not executed until all the empty spaces are filled (step S228). When the fourth point is selected, as shown in FIG. 11, magnetic data that is not determined as a selection candidate up to the third point (including those temporarily determined but rejected thereafter) is included. Sequential selection is made as the fourth selection candidate. In FIG. 11, identifiers for managing the input order of magnetic data in the buffer are indicated by numbers from 1 to q _max . A smaller number represents a data structure in which previously inputted magnetic data is stored.

-Four-point selection condition In step S230, all magnetic data stored in the buffer are sequentially combined one by one with the magnetic data having the three selected points as components, and four magnetic data are obtained in four points. It is sequentially determined whether the selection condition is satisfied, and one of the magnetic data stored in the buffer that satisfies the four-point selection condition is selected as the fourth highest one. The four-point selection condition is that the four points whose components are magnetic data that are candidates for selection are sufficiently dispersed on the spherical surface passing through these four points, that is, the tetrahedron having the four points shown in FIG. This is a condition for selecting the fourth point from the viewpoint of whether it is close to a tetrahedron. Therefore, the distortion of a tetrahedron with four points as vertices having magnetic data as a selection candidate as a component is calculated. The distortion of a tetrahedron to a regular tetrahedron is defined by the variation in solid angle of each vertex of the tetrahedron, the variation in the area of the surface of the tetrahedron, the variation in the distance from the center of gravity of the tetrahedron to each vertex, etc. be able to. These variations are evaluated by the eigenvalues of the symmetric matrix B of 3 rows and 3 columns expressed using the sum of vectors having the center of gravity d _{4 of the} tetrahedron as the starting point and each vertex as the ending point. The center of gravity d ₄ of the tetrahedron is obtained by the following equation (20).

The symmetric matrix B is defined by the following equation.

The larger the ratio of the minimum eigenvalue λ _B1 to the maximum eigenvalue λ _B3 of the symmetric matrix B, the smaller the distortion of the tetrahedron with respect to the regular tetrahedron, that is, the variation in solid angle of each vertex of the tetrahedron, the surface of the tetrahedron It can be said that both the variation in area and the variation in the distance from the center of gravity of the tetrahedron to each vertex are small. Therefore, the 4-point selection condition is that E _B1 obtained by the following equation (22) is maximized.

E _B1 is calculated for all the magnetic data stored in the buffer, and when the maximum value is determined, the fourth magnetic data that maximizes E _B1 is the data structure p ₃ as the final selection candidate. Stored in For p ₀ , p ₁ , p ₂ , and p ₃ in which magnetic data that are selection candidates up to the final fourth point determined in this way are stored, another evaluation amount E _B2 is expressed by the following equation (23 ).

Λ _B2 is an intermediate eigenvalue of the symmetric matrix B. If the following equation (24) is not satisfied with respect to the predetermined threshold values E ₂ and E ₃ determined in advance, there is a possibility that a large error may be mixed when obtaining the spherical equation, so the process returns to step S200. All the buffers and data structures p ₀ , p ₁ , p ₂ , p ₃ are reset. When the following equation (24) is satisfied, the fourth selection candidate is finally selected.

However, rather than selecting the fourth point using one combination of two threshold values, two or more combinations of two threshold values are set, and the following equation (25) is set for all i On the other hand, when the buffer and the data structures p ₀ , p ₁ , p ₂ , and p ₃ are all reset when it does not hold, it is more reasonable from the viewpoint of reducing the number of calibration failures.

Calculation of azimuth sphere When magnetic data up to the fourth point are selected by the above processing, that is, when p ₀ , p ₁ , p ₂ , and p ₃ are determined, the center point and radius of the azimuth sphere are obtained ( Step S232). The center point c ₄ = (c _4x , c _4y , c _4z ) of the azimuth sphere is obtained by the following equation. c _4r is a value that is not particularly used later, but is added to establish a quaternary linear simultaneous equation.

The radius r of the azimuth sphere is obtained by the following equation (27) as the Euclidean distance between c ₄ and p ₀ .

-Verification of bearing ball In step S234, the validity of the calculated bearing ball is verified. If the radius of the azimuth sphere is not within a predetermined range, the calculated center point of the azimuth sphere is not adopted as an offset, and the process returns to step S200 to return to the buffer and data structures p ₀ , p ₁ , p ₂ and p ₃ are all reset.

In the present embodiment, the validity of the azimuth sphere is verified by statistical processing using not only the selected four magnetic data but also a part or all of the magnetic data stored in the buffer. This verification is effective to some extent in order to reduce the variation in offset error, but is not necessarily a necessary process. In this verification, prompted evaluation amount S by the following equation (28), for a given threshold S _t which is determined in advance, in the case of S> S _t is the center point of the calculated orientation sphere The buffer and data structures p ₀ , p ₁ , p ₂ , and p ₃ are all reset without returning to the process of step S200. Otherwise, the offset is set as follows:

-Offset setting When the calculated azimuth is valid, an offset is set in step S236. That is, each component of the calculated center point of the azimuth sphere is set as each component of the offset.

1: magnetic sensor control device, 4: three-dimensional magnetic sensor, 90: magnetic sensor control program, 92: buffer management module (input means), 94: offset setting module (selection means, calculation means, setting means)

Claims (5)

Input means for inputting a plurality of magnetic data having three components sequentially output from a three-dimensional magnetic sensor;
A selection means for selecting four magnetic data satisfying a predetermined four-point selection condition from the plurality of input magnetic data;
Calculating means for calculating a center point which is a point equidistant from four points having the four selected magnetic data as components;
Setting means for setting the component of the center point as an offset of the magnetic data;
With
The four-point selection condition includes a condition that correlates with a distortion of a tetrahedron having four points as vertices having four magnetic data components as selection candidates.
Magnetic sensor control device. The four-point selection condition includes that a variation in solid angle of each vertex of the tetrahedron is within a predetermined range.
The magnetic sensor control apparatus according to claim 1. The four-point selection condition includes that the variation of the area of each surface of the tetrahedron is within a predetermined range determined in advance.
The magnetic sensor control apparatus according to claim 1. The four-point selection condition includes that a variation in distance from the center of gravity of the tetrahedron to each vertex is within a predetermined range.
The magnetic sensor control apparatus according to claim 1.

The four-point selection condition includes that the ratio of each two sets of eigenvalues of the following 3 × 3 symmetric matrix B is larger than a predetermined value.
The magnetic sensor control apparatus as described in any one of Claims 1-4.

However,
p _i (i = 0, 1, 2) is each vertex of the tetrahedron, and d ₄ is the center of gravity of the tetrahedron.

Images