JPH0799560B2 - How to update the sensor correction coefficient - Google Patents

Description

TECHNICAL FIELD The present invention relates to a method for updating a sensor correction coefficient, and more particularly to a method for updating a correction coefficient for correcting output data of a sensor for detecting a mileage and an azimuth of a vehicle in an in-vehicle navigation device. Is.

BACKGROUND ART In recent years, a vehicle-mounted navigation device that guides a vehicle to a predetermined destination by storing map information in a memory and reading the map information from the memory on a display device together with the current location of the vehicle has been researched and developed. Has been done.

In such a navigation device, by detecting the mileage and the azimuth of the vehicle based on the output data of the mileage sensor and the azimuth sensor mounted on the vehicle, by estimating the present location of the vehicle that changes momentarily based on this , The current location is displayed on the map displayed on the display.

In this case, it is preferable that the current position is always displayed on the road of the map, but due to the accuracy of the sensor and the like, the current position will deviate from the road on the display, especially as the traveling distance increases.

Since it is inevitable that the mileage sensor and the azimuth sensor have an error in the output data obtained by their accuracy, it is necessary to correct the output data using a correction coefficient. However, the error of the mileage obtained by the mileage sensor may be not only due to the sensor accuracy but also due to the accuracy of the map, the change of the tire air pressure, or the slip. If is always used, the mileage cannot be accurately obtained.

Also, even when using a geomagnetic sensor that detects the direction of the vehicle based on the geomagnetism (earth magnetic field) as the orientation sensor, the north indicated by this geomagnetic sensor is not the map north,
The output data of the sensor is corrected using the azimuth correction coefficient. However, since this correction coefficient changes depending on the region, if the azimuth correction coefficient is kept constant, the azimuth of the vehicle cannot be accurately detected.

SUMMARY OF THE INVENTION The present invention has been made in view of the above points, and an object of the present invention is to provide a method of updating a sensor correction coefficient that can always detect an accurate traveling distance and a vehicle azimuth.

A method for updating a sensor correction coefficient according to the present invention obtains a distance from a current position to a nearest position on a road based on map data, further obtains a deviation of a change rate of the distance after traveling a predetermined distance, and the deviation is a predetermined value. When it is smaller than the reference value, the correction coefficient for correcting the output data of the sensor for detecting the traveling distance and the azimuth of the vehicle is updated.

Example Hereinafter, an example of the present invention will be described with reference to the drawings.

FIG. 1 is a block diagram showing the configuration of a vehicle-mounted navigation device according to the present invention. In the figure, 1 is a geomagnetic sensor for outputting the azimuth data of the vehicle based on the geomagnetism, 2 is an angular velocity sensor for detecting the angular velocity of the vehicle,
3 is a travel distance sensor for detecting the travel distance of the vehicle,
Reference numeral 4 is a GPS (Global Positioning System) device for detecting the current position of the vehicle from latitude and longitude information and the like, and the output of each of these sensors (devices) is a system controller 5
Is supplied to.

The system controller 5 receives the outputs of the respective sensors (devices) 1 to 4 as an input, performs an A / D (analog / digital) conversion, and the like, and performs various image data processing and sequentially sends from the interface 6. A CPU (central processing circuit) 7 that calculates the amount of movement of the vehicle based on the output data of each sensor (device) 1 to 4, and this C
A ROM (read only memory) 8 in which various processing programs of the PU 7 and other necessary information are written in advance, and a RAM (random access memory) in which information necessary for executing the program is written and read. ) 9
And a so-called CD-ROM, IC card, etc., and a recording medium 10 on which digitized (numerical) map information is recorded.
And a graphic memory consisting of V-RAM (Video RAM), etc.
11 and the graphic data such as a map sent from the CPU 7 are drawn in the graphic memory 11 and the CRT is used as an image.
Etc. and a graphic controller 13 for controlling to display on the display 12. The input device 14 is composed of a keyboard and the like, and issues various commands to the system controller 5 by key input by the user.

The map information is recorded on the recording medium 10. The data structure of the map information will be described below. First, as shown in FIG. 2 (A), the whole map of Japan is, for example, 16384 (= 2 ¹⁴ ).
[M] The mesh is divided into four meshes, and one mesh at this time is called a territory. Territory is territory No.
It is identified by (Tx, Ty), and each territory is given a territory No. based on the territory at the lower left of the figure, for example.
The territory No. is obtained from the current location (Crntx, Crnty). The territory is the largest management unit in this data structure. The structure of the entire map data file is shown in Fig. 2 (B), and the territory ID file is shown in Fig. 2 (C).
As shown in, the start address in the file of the territory No. (Tx, Ty), the lower left latitude of the territory (real number),
Data such as the longitude (real number) at the lower left of the territory and the declination angle (real number) of the geomagnetism are written for each territory.

The territory file is the most important file in this data structure, and various map data and data necessary for map drawing are written in it. In FIG. 3A, the navigation ID and section table are road and intersection search files in navigation, the picture ID is a display management file, and road section data to intersection data are actual map data. As shown in FIG. 3 (B), the map data has a hierarchical structure. The lowest layer is polygon data of rivers, seas, lakes, etc., the top is line data of roads, railways, etc. Character data such as marks, character data such as place names and the like, and intersection data at the top layer. The intersection data of the uppermost layer is data used for pulling in an intersection described later and is not displayed on the display.

Next, as shown in FIG. 4 (A), one territory is divided into 256, for example, and 1024 (2 ¹⁰ ) obtained by this is obtained.
[M] The four-sided mesh is called a unit. This unit is also managed by the unit No. (Nx, Ny), and that No. (Nx
x, Ny) is obtained from the current location (Crntx, Crnty). A unit is an intermediate management unit, and map information is recorded in this unit, and 256 units form a territory file. Since it is based on this unit when drawing a map, it can be said to be the basic unit for drawing. In the navigation ID file, as shown in FIG. 4 (B), the unit number (Nx,
Data such as the line start address, the intersection start address, the road section start address, and the intersection start address in the Ny) file are written for each unit.

Further, as shown in FIG. 5 (A), one unit is divided into, for example, 16 units, and 256 (2 ⁸ ) [m] obtained by this
The four-sided mesh is called a section. This section is also managed by the section No. (Sx, Sy), and the No. (Sx
x, Sy) is obtained from the current location (Crntx, Crnty). The section is the smallest management unit, and the information on line segments (roads are represented by connecting line segments) and intersections within this range is the fifth.
As a section table as shown in FIGS. 6 (B) and 6 (C), FIG. 6 (A), (B) and FIG. 7 (A),
As shown in (B), it is registered in the territory file as section data.

Further, as shown in FIG. 3A, there is a file called a picture ID for display management in the territory file. In this embodiment, the scale of the map data is set to three types, for example, 15,000 / 10,000 and 1 / 100,000,
As actual map data, we have only the largest scale, 1 / 25,000. The map of each scale is divided into areas as shown in each of FIGS. 8 to 10 (A), and this area is managed by area No. (Anx, Any).
Area No. (Anx, Any) is obtained from the current location (Crntx, Crnty). If the scale is 1 / 2,500, the area number and unit
The No. is the same, and in the case of 1 / 50,000, one area is for 4 unit files, and in the case of 1 / 100,000, it is 1
One area is for 16 units. In addition, as shown in each figure (B) of FIG. 8 to FIG. 10, the picture ID of each scale has the beginning of the polygon, line, character, and character data required to display the map of that scale. Address and data size are recorded.

Next, the polygon data and the line data will be described. Polygon data and line data are represented by connected vectors (line segments) represented by a start point and an end point, as shown in FIGS. 11 (A) and 12 (A). Here, map data of the largest scale, 1 / 5,000, is 1 / 50,000.
If you represent a map of 1 / 100,000 or 100,000, the distance between the start point and the end point will be shortened, so as long as you see it on the display, it may be possible to display all points. Taking this into consideration, the information about the points that can be omitted visually when displayed on the display is shown in FIG.
As shown in FIG. 12 (B), it is inserted in advance in each thinning bit of polygon and line data. Then, the number of line segments (vectors) to be displayed can be reduced by checking the thinning bit at the time of displaying each scale and removing the point that the thinning bit contains information as necessary, so-called thinning.

Further, as shown in FIG. 13 (A), serial numbers (xn, yn) are assigned to all the intersections existing in one unit. By the way, there are various kinds of intersections, such as an orthogonal type, a Y-shaped road, and a five-forked road. Especially, at an intersection that has a plurality of roads with similar directions, the accuracy of the sensor when passing through this intersection,
There is a possibility that a road is erroneously selected due to a calculation error, map accuracy, or the like, and the road whose current position is displayed on the display does not match the road actually being traveled.
Therefore, for such an intersection, as shown in FIG. 13 (B), difficulty data indicating the difficulty of the intersection is put in the difficulty bit in the intersection data. Then, when passing through the intersection, by performing processing based on this difficulty level data, it is possible to prevent erroneous road selection. The processing will be described later.

Next, regarding the display of map data, a graphic memory
The case where V-RAM is used as 11 will be described.
As shown in FIG. 14 (A), the display configuration is 512
The screen is divided into 16 areas on (dots) x 512 (dots) V-RAM, and an independent map is displayed in each area. One area is 128 (dots) x 128 (dots)
It is one unit of × 32 (dots), and by further dividing into 16 units, one area becomes one section of 32 (dots) (see FIGS. 14 (B) and 14 (C)). The actual in-vehicle display has 256 screens, which correspond to the four screens in the center of Fig. 14 (A).
An area of (dots) x 256 (dots) (area surrounded by a thick line) is displayed, and this area represents the movement of the vehicle's current position by moving on the V-RAM.

Next, the basic procedure executed by the CPU 7 will be described with reference to the flowchart of FIG.

The CPU 7 first initializes to execute the program (step S1), and thereafter determines whether or not the current position of the vehicle is set (step S1).
2). If the current location has not been set, the current location setting routine is executed (step S3), for example, the current location is set by key input on the input device 14. next,
The mileage is set to zero (step S4), and then it is determined whether or not there is a key input from the input device 14 (step S4).
Five).

If there is no key input, the map around the current location is displayed on the display 12, and the current position and direction of the vehicle are displayed on this map, for example, by the vehicle mark, and when the vehicle moves, the map scrolls with the movement. When the vehicle position is likely to exceed the range of the map data currently on the graphic memory 11, the necessary map data is read from the recording medium 10 and displayed on the display 12 (step S6).

If there is a key input, each routine of resetting the current location (step S7), correcting the sensor (step S8), setting the destination (step S9) and enlarging / reducing the map (step S10) is executed according to the input data. To do.

Further, the CPU 7 performs a process of constantly calculating the heading of the vehicle based on the output data of the geomagnetic sensor 1 and the angular velocity sensor 2 at a constant time interval as shown in FIG. 16 by the interruption by the timer (steps S11 and S12). ).

When the data is input from the mileage sensor 3, the CPU 7 further performs an interrupt process by the mileage sensor. In this interruption processing, as shown in FIG. 17, the present position is calculated from the traveling distance and the azimuth (step S13), right turn or left turn is determined (step S14), and the road is pulled in (step S1).
5), intersection pull-in (step S16), and pull-in by travel distance (step S17) are executed. The respective processes in steps S13 to S17 will be described in detail later.

Also, the latitude and longitude data obtained from the GPS device 4 is the first
As shown in Figure 8, it is processed by the GPS data reception interrupt, and the coordinates are converted as the current position data (step S1.
8).

The travel distance of the vehicle is obtained from the output of the travel distance sensor 3. The mileage sensor 3 may be, for example, the rotation speed of a so-called speedometer cable of a car (6 in JIS standard).
Although a configuration is used in which the traveling distance is obtained by integrating the distance of one rotation from (37 rotations / Km), it is inevitable that an error will occur in the traveling distance obtained by the accuracy of the sensor 3. Further, not only the accuracy of the sensor 3, but also the accuracy of the map, the change of the tire air pressure, the slip, etc. are factors of the error of the traveling distance. Therefore, unless the traveling distance is corrected frequently, the distance cannot be accurately obtained. Therefore, the distance correction coefficient rs is obtained from the actually measured distance obtained from the output of the travel distance sensor 3 and the distance obtained from the map data, and the distance correction is performed using this correction coefficient rs, so that the traveling distance is always accurate. Can be detected.

The direction of the vehicle is obtained from the output of the geomagnetic sensor 1. This azimuth detection method is described in the specification of Japanese Patent Application No. 60-282341 filed by the present applicant. The north indicated by the geomagnetic sensor 1 is magnetic north, not map north. Therefore, if the magnetic north is displaced from the map north, as shown in FIG. 19, the estimated current position P ₁ obtained from the output of the geomagnetic sensor 1 when traveling a certain distance from the reference position is the actual current position P _1. There will be a deviation from ₂ . Therefore, it is necessary to convert the azimuth obtained from the geomagnetic sensor 1 into a map azimuth. As shown in FIG. 20, this conversion work is performed by the rotation angle obtained by the coordinate conversion of the two-dimensional geometry, that is, the azimuth correction coefficient θs. This azimuth correction coefficient θ
s changes depending on the region, and also changes due to a mounting error that occurs when the geomagnetic sensor 1 is mounted on the vehicle body. As shown in FIG. 21, the bearing correction coefficient θs can be obtained from the error between the current position and the arrival point obtained by inertial navigation while traveling between two points whose positions are known with the coefficient being zero. . By performing the heading correction using the heading correction coefficient θs, the heading of the vehicle can always be detected accurately.

The method of calculating the distance correction coefficient rs and the azimuth correction coefficient θs is described in the specification of Japanese Patent Application No. 60-282344 filed by the present applicant.

Next, the procedure of interrupt processing by the mileage sensor 3 executed by the CPU 7 will be described with reference to the flowchart of FIG. The estimated point of the present location is calculated at any time based on the output data of the traveling distance sensor 3, and this routine is called at a predetermined timing as a present location display routine.

The CPU 7 first determines whether or not it has traveled the unit distance lo (step S20). Here, the unit distance refers to a certain distance that the vehicle actually travels, and is set to 20 [m], for example. Then, this routine is executed for each constant traveling distance, and first, the relationship with the map data, that is, the distance lm to the nearest line segment L, the angle θn of the line segment L with the map north, etc. are obtained. Further, when there are two or more line segments at substantially equal distances, this is indicated by a flag (step S21). In addition, the presence / absence of a nearby intersection may be obtained here. Then, it is determined whether or not the distance lm exceeds a preset threshold lth (step S22). If it does not exceed, it is assumed that the current position is near the line segment, and the error component lm
Is corrected (step S23). This error component lm is caused by a detection error of the traveling distance sensor 3, a digitizing error of map data, and the like. This correction is performed because it is necessary to cancel those errors in order to display the current location next time. After that, the process proceeds to a pattern pull-in routine described later.

On the other hand, if the distance lm exceeds the threshold lth, it is next determined whether or not the vehicle has made a curve (turn right or turn left) (step S24). The method of detecting the curve will be described later. When no curve is made, the vehicle heading θ obtained from the output data of the geomagnetic sensor 1 and the angle θn of the line segment L
Is compared with the set reference value θth (step S25). ｜
If θ−θn |> θth, the process proceeds to the pattern pull-in routine without doing anything. In this case, for example, there may be a case where the vehicle is hitting a T-shaped road in the hitting direction or traveling on a road which is not stored as map data. Then, it is determined whether or not there is an intersection with a smaller angle, such as a Y-shaped road, having a smaller angle and having a high degree of difficulty (step S26).
For example, if there is a Y-shaped road in the vicinity, it may lead to a different road from the one currently running,
Do nothing and proceed to the pattern pull-in routine. Since the data indicating the difficulty level of the intersection is inserted in the difficulty level bit of the intersection data in advance as shown in FIG. 13 (B) when the map is digitized, the CPU 7 can check this bit in step S26. It's good.

If the above two conditions do not apply, it is determined that an error has occurred in the sensor or the like and the vehicle is deviating from the road data. In this case, the current position is corrected, that is, the road data is pulled in (step S27). New current location guess point Pcpd
As shown in Fig. 24, is the current estimated point calculated from the relative relationship from the previous estimated point Pppd obtained from the sensor output.
Change the display from Pcp to the point that intersects the perpendicular line drawn to the nearest line segment L. The distance lm and the coordinates (Xm, Ym) of the current estimated point Pcp are stored as correction values at that point for use in the pattern pull-in routine described later.

If it is determined in step S24 that a curve has been made, an intersection pull-in routine is entered. First, the traveling distance lc from the point displayed as the previous intersection is obtained, and the intersection detection threshold lcth is obtained by multiplying the traveling distance lc by a constant value ac (step S28). The constant value ac is a value related to the accuracy of the traveling distance sensor 3, and is set to a value of about 0.05, for example. The distance lc from the current location Pcp to each intersection is calculated for the intersection data included as map data (step S29), lc <lcth
It is determined whether there is an intersection (step S3)
0). In step S30, it is also determined whether or not it is within a certain distance range (for example, about several hundred [m]). Here, if there is no intersection, the process proceeds to the pattern pull-in routine. Also, when it is determined that there are a plurality of neighboring intersections and the distances lc to the intersections are similar to each other and the neighboring intersections cannot be specified (step S31), the process proceeds to the pattern pull-in routine.

When a nearby intersection is specified, the intersection is pulled in as a new current position estimation point Pcpd (step S3).
2). At this time, the distance lc to the intersection and the coordinates (Xc, Yc) of the current position Pcp are stored as the pull-in amount. Also, the current position estimation point Pcpd is stored (updated) as a new display intersection. As a result, when the current position of the vehicle deviates from the road on the map displayed on the display 12, the current position of the vehicle is forcibly placed on the intersection on the map, so-called intersection drawing is performed. Subsequently, the correction coefficients rs, θs for the distance and the bearing are updated based on the coordinates of the previously recognized intersection, the sum of the correction values from the intersection, and the coordinates of the current position (step S33). In this way, if the intersections are recognized, or if the distance between the intersections is long, the correction coefficients rs and θs for the distance and bearing are updated each time the vehicle travels a certain distance lp, so that a more accurate current position can be estimated. It will be possible.

Next, the pattern pull-in will be described. This routine is executed when running a certain distance lpo. Distance lpo
Is, for example, a value of 1000 [m]. In addition, step
When the intersection is recognized in S31, the traveling distance is reset. While traveling a fixed distance lpo, the error distance lm to the nearest line segment is to be measured n = lpo / lo [times], and n error correction amounts ei are stored as data. Further, for one measurement, the difference between the error correction amount ei-1 at the previous measurement and the error correction amount ei at this time is calculated as the change amount ci.
It shall be calculated as (ei-ei-1).

If you determine that you have run a certain distance lpo (step S34),
Calculation is performed for the variation in the value of the change amount Ci. First,
Average cm (Step S35), and then the deviation α Is calculated (step S36). Then, this value α is compared with a predetermined threshold value αthh (step S37). This value α
Is obtained by considering the detection error of each sensor, the traveling distance, and the like. When α> αthh, it is determined that the pull-in is impossible, and the pattern pull-in is not performed. On the other hand, if α <αthh, it is further determined whether or not retraction is currently performed (step S38). If retraction is not performed, that is, if the current position is at a position deviating from the road data, the current position is the latest. The line segment is pulled in (step S39).
Further, it is compared with the threshold value αthl which is set similarly to the threshold value αthh (step S40), and when α <αthl, the distance and azimuth correction coefficients rs and θs are updated (step S41).

By the method described above, it becomes possible to re-engage on another road after traveling once off the road. That is, when the vehicle travels on a non-digitized road and then on the digitized road again, the road is recognized at the time when the vehicle travels a certain distance, and it becomes possible to accurately estimate the current position. It is also possible to calculate the deviation for a longer distance lt with respect to the constant distance lpo, shorten the distance lpo to improve accuracy, and shorten the response time. The situation is shown in FIGS. 25 (A) and (B).

As described above, the attraction to the nearest intersection and the attraction to the nearest line segment are performed. To perform this attraction, the nearest road (nearest line segment) or intersection (nearest neighbor intersection) to the current position is used. ) Is required to find out.
The task of searching for the nearest intersection or the nearest line segment takes a long time if the amount of line segment or intersection data is large, that is, if the search area is large, it may not be possible to smoothly display the current location that changes moment by moment. Become. However, in the present embodiment, as is clear from the data structures shown in FIGS. 2 to 5, the search area from the current position is made as small as possible, and the data of the line segment and the intersection which enter the area are managed. By providing the data (section data, section table), it is possible to search for a line segment or an intersection from the minimum unit section as a search area, so that the time required for the search can be shortened. The procedure for searching for the nearest line segment and the nearest intersection from the current position, which is executed by the CPU 7, will be described below.
It will be described according to the flowchart of FIG.

CPU7 starts from the current location (Crntx, Crnty) and ends in territory No.
(Tx, Ty), Unit No. (Nx, Ny), Section No. (S
x, Sy) are respectively calculated (steps S50 to S52). This can be obtained by a simple calculation (division) because each area is divided into 2 ⁿ units. Next, using the section as a search area, line segments and intersection data existing therein are loaded by referring to the section table and section data (steps S53 to S55). Based on the loaded data, calculate the distance from the current position to all line segments in the search area (length of the perpendicular to the line segment), the distance to all intersections, and compare them to find the nearest line segment. The nearest intersection can be obtained (step S56). The speed at which the search is performed is proportional to the number of line segments and the number of intersections, but the search method based on the above-mentioned data structure has a small search area (section) and the number of line segments to be calculated. Since the number of intersections and intersections is small, high-speed search is possible.

By the way, in the navigation system, when displaying map data of various scales, if map data of all scales are held, the display can be performed easily and at high speed, but the disadvantage is that the data size on the one hand becomes large. There is. Conversely, if you have only the map data with the largest scale and other scales are represented by simple reduction,
Although it has a smaller data size, it has the drawback of slower display.

On the other hand, in this embodiment, as is clear from the data structures shown in FIGS. 8 to 10, only the map data having the largest scale is provided to reduce the data size, and the map data of other scales is used. When the data is displayed, the management file for display and the thinned-out data are used to speed up the display. The procedure for enlarging / reducing the map executed by the CPU 7 will be described below with reference to the flowchart of FIG.

When the CPU 7 first determines that the scale to be displayed is keyed in from the input device 9 (step S60), the current position (Crn
Area No. (Anx, Any) corresponding to the scale is calculated from tx, Crnty) (steps S61 to S63), and then the picture ID of the scale is referred to (steps S64 to S66), and the map is selected according to the start address and data size. Load data and 1 on V-RAM
Drawing is performed on each of the six areas (step S67). In this way, it is possible to refer to the identified map data to be displayed by using the picture ID for display management (restricting the displayed roads, place names, etc. to important ones as the scale becomes smaller), thus speeding up the display. Can be realized.

Also, for polygon and line data, as described with reference to FIGS. 11 and 12, there is no problem even if the display is omitted. 1 or 1 / 100,000 of the map is drawn, the thinning bit is checked (step S68), and the points other than the thinning target are drawn (step S6).
9). In this way, when the map is reduced, when it is displayed on the display, the number of line segments to be displayed can be reduced by thinning out the points that may be omitted from the appearance. It is possible to achieve higher speed.

In the above embodiment, the thinning bit is provided for the polygon and line data, and the information indicating that the thinning bit can be omitted may be included. However, the polygon and line data are equally spaced. Plot it with
At the time of display, thinning may be performed according to a predetermined rule (for example, if the scale is 1 / 50,000, one jump, if it is 1 / 100,000, four skips, etc.), and the same effect can be obtained.

Next, a method of determining the curve (right turn / left turn) in step S24 in the flowchart of FIG. 22 will be described.

Basically, a right turn or a left turn is discriminated based on the output data of the azimuth sensor, for example, the geomagnetic sensor 1, and when it is detected that a turn is made, the intersection drawing is performed by the processing from step S28. However, the geomagnetic sensor 1 is vulnerable to disturbance, and when a large vehicle (for example, a truck or a bus) passes by the vehicle at the level crossing, at the railway bridge, or at the side of the vehicle, a large error is included in the output data. If this data is used as it is for right / left turn determination, it will be mistaken for a straight turn, and the intersection will be pulled in even if it is not an intersection, and the current location will be displaced from the correct position.

Therefore, in the present embodiment, the radius of curvature and the vehicle speed are used as the criteria for determining that the vehicle is bent,
It is possible to make an accurate right / left turn decision. Below, CPU7
The procedure of the method for determining a right turn / left turn executed by will be described with reference to the flowchart in FIG. 28. First, the CPU 7 determines that a curve (a right turn or a left turn) is made when the vehicle turns a certain angle (for example, 40 degrees) or more while traveling a certain certain distance (for example, 15 [m]) (step S70). However, when curved, the curvature radius R at that time is a constant value Rmin (for example, 3.5
In the case of [m]) or less, it is determined that the data is incorrect, and it is not determined that the data is curved (step S71). This is because the vehicle cannot bend below the minimum turning radius of the automobile. Furthermore, when the vehicle speed S is a certain speed Smax ((for example, 40 [Km / h]), which is the maximum speed for judgment, it is not normally considered to turn at an intersection, so at this speed or more, a curve occurs. Is not determined (step S72).
For example, if the east direction is 0 degrees, the north direction is 90 degrees, the west direction is 180 degrees, and the south direction is 270 degrees.
This can be done by increasing or decreasing the azimuth (step
S73). In other words, turn left in the direction of increasing bearing (step S
74), the direction in which the direction decreases is the right turn (step S75), and thus the right turn or the left turn can be determined.

As shown in FIG. 29, the radius of curvature R is θ [radian], where θ is the angle formed by the azimuth of the vehicle at a point a and the azimuth of the vehicle at a point b traveling a certain distance 1 from the point a. , L = R · θ, it can be obtained from the following equation R = 1 / θ obtained by modifying this equation.

Also, due to the intersection pull-in performed by the processing according to the flow of FIG. 22, it is controlled so that the current location is always located on the road of the map displayed on the display. For example, the distance between the intersections is long. In this case, the current location is slightly corrected during that time, but the distance difference between the actual current location from the previously pulled-in intersection and the current location on the map is caused by the distance error due to sensor accuracy, calculation error, map accuracy, etc. The error increases as the distance between the intersections increases. In such a case, if there are a plurality of intersections close to each other in the vicinity of the intersection to be retracted next, there is a possibility that the retraction may be performed at an incorrect intersection. In such a case, if the vehicle travels a certain distance between the intersections, the so-called travel distance may be used. The procedure will be described below with reference to the flowchart in FIG.

First, an initial value is set (step S80). As this initial value, a certain fixed current location is required, but this can be the current location when the user first sets it or pulls it to a fixed point such as an intersection, or if it is already fixed If the data is registered in the non-volatile memory, it only has to be set once. The mileage is reset to zero at this confirmed current position (step S8
1) The point on the map that was reset to zero based on the map data (previously detected position) when running a certain distance while constantly monitoring whether the vehicle has turned at an intersection (step S82) or has run a certain distance (step S83). ) To obtain a point at a certain distance, change the current position to that point, and pull in (step S84). While driving for a certain distance, draw a perpendicular line on the apparently closest line segment and draw in the intersection (step S85) so that the current position of the vehicle is placed on the road of the map displayed on the display. You can When it is detected that the vehicle is bent, the intersection is pulled in (step S86). This intersection pull-in is as described above.

In addition, even if it is judged that the vehicle has made a turn at an intersection, the pull-in based on the traveling distance can be effectively used. That is, it is possible to determine from the map data the curved intersections depending on the distance between the intersections and the traveling distance.

EFFECTS OF THE INVENTION As described above, according to the present invention, the distance from the current position to the nearest position on the road is obtained based on the map data, and the deviation of the change rate of the distance is obtained after traveling a predetermined distance, When this deviation is smaller than the predetermined reference value, the sensor output is constantly corrected with the latest correction coefficient by updating the correction coefficient that corrects the output data of the sensor that detects the traveling distance and azimuth of the vehicle. Because you can
Therefore, the traveling distance and the azimuth of the vehicle can always be detected accurately.

[Brief description of drawings]

FIG. 1 is a block diagram showing the configuration of an in-vehicle navigation device according to the present invention, and FIGS. 2 (A) to (C) to 13 (A) and (B) are stored in the recording medium in FIG. Diagram showing the data structure of map information, Fig. 14 (A) to (C)
Is a diagram showing a screen structure on the V-RAM, FIGS. 15 to 18 are flowcharts showing basic procedures executed by the CPU in FIG. 1, and FIGS. 19 to 21 are azimuth correction coefficients θ.
FIG. 22 is a flowchart showing the procedure of the intersection pull-in routine and the pattern pull-in routine executed by the CPU, and FIGS. 23 and 24 are the positional relationship between the current position on the map and the nearest line segment. Fig. 25 is a diagram showing another method of pulling in the road, Fig. 26 is a flowchart showing a procedure for searching for the nearest line segment and an intersection, and Fig. 27 is a procedure for enlarging / reducing the map. Fig. 28 is a flow chart showing the procedure of the method for determining a right / left turn, Fig. 29 is a diagram showing how to obtain the radius of curvature, and Fig. 30
The figure is a flow chart showing the procedure of the retracting method according to the traveling distance. Explanation of symbols of main parts 1 ... Geomagnetic sensor, 2 ... Angular velocity sensor 5 ... System controller 7 ... CPU, 10 ... Recording medium 12 ... Display, 14 ... Input device

Claims (1)

[Claims] 1. A method of updating a correction coefficient for correcting output data of a sensor for detecting a mileage and an azimuth of a vehicle, wherein each position on a road on a map and an intersection position are digitized and stored as map data. The error distance from the current position to the nearest position on the road, which is obtained based on the output data of the sensor, is sequentially obtained based on the map data, and the rate of change of the error distance is further calculated for each predetermined distance traveling. A method for updating a sensor correction coefficient, which comprises obtaining a deviation and updating the correction coefficient when the deviation is smaller than a predetermined reference value.