JPH09229698A - Position sensing device - Google Patents

Abstract

PROBLEM TO BE SOLVED: To shorten the time for a matching process for a map to be made on the basis of the position and azimuth of a car acquired by the autonomous navigation method, satellite telemetry navigation method, etc., by selecting optimum ones as the restrictive conditions of a candidate road. SOLUTION: Two different procedures may be considerable; (a) links existing within the limit error range E of the presumed position are extracted from a certain extent of road map data, and among these links, a one having an azimuth whose difference from the presumed azimuth of the car lies within the threshold is selected, (b) links having an azimuth whose difference from the presumed azimuth of the car lies within the threshold are extracted from a certain extent of road map data, and among them, link (s) positioned within the limit error range E is selected. Which procedure should be adopted is decided on the basis of the estimated time required to execute the procedure, and the object link (s) for map matching is selected in conformity to the adopted procedure, and the map matching is executed for the selected links.

Description

Detailed Description of the Invention

[0001]

BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a position detecting device for detecting the position of a vehicle traveling on a road on a map.

[0002]

2. Description of the Related Art Conventionally, in order to detect the position of a vehicle running on a road, a distance sensor, a direction sensor, and a processing device for performing necessary processing on output signals from both sensors have been provided. Autonomous navigation has been proposed in which the current position data of the vehicle is obtained while integrating the distance change amount and azimuth change amount caused by running of the vehicle, but the distance sensor and the azimuth sensor necessarily have There is a problem that the errors in the current position data are accumulated as the running continues, and the errors included in the obtained current position data are also accumulated.

In consideration of such problems and on the premise that a vehicle travels on a road, a current position data obtained based on the autonomous navigation and a road stored in a map memory in advance are used. Comparing with the map data, the deviation amount (distance difference and azimuth difference) of the current position data from the road is calculated as a cumulative error, and the current position data is corrected for the cumulative error. A map matching process has been proposed in which the current position data is made to match the road map data (see Japanese Patent Application Laid-Open No. 61-56910, etc.).

At the time of executing the map matching process, in order to reduce the amount of calculation required for the process, the directions of the roads stored in the map memory are classified, and the roads having the directions belonging to the classification are grouped for each classification. A proposal has been made to search for a relevant road from a group of roads in which the detected azimuth of the vehicle and the azimuth of the road are similar when specifying the target candidate road (see Japanese Patent Application Laid-Open No. 1-260312).

According to this method, the number of candidate roads can be reduced, so that the efficiency of map-map matching and the required time can be reduced.

[0006]

However, in practice,
When performing the map matching process, that is, when performing the map matching between the detected direction of the vehicle and the direction of the road, the target road is not limited and has a certain range (for example, several km square around the estimated position of the vehicle). Selected from the roads in the range).

When selecting a target road from the roads in the "fixed range", all the roads in the "fixed range" can be set as candidate roads, but in order to reduce the number of candidate targets. In order to improve the efficiency of map map matching, it is preferable to set a narrower range in stages, determine a candidate road from the narrower range, and gradually expand the range when there is no candidate road.

As described above, when the target road is limited, it is expected that the processing time will be different depending on whether the direction is limited first or the distance is limited first. For example, if there are multiple roads near the position of the vehicle obtained by autonomous navigation and their directions are completely different, the number of target roads can hardly be reduced even if the distance is limited first. . In addition, when there are roads in a wide range of directions similar to the direction of the vehicle obtained by autonomous navigation, the number of target roads cannot be reduced even if the direction is limited first.

According to the present invention, when the map matching processing is performed based on the position and direction of the vehicle obtained by the autonomous navigation, the satellite positioning navigation, etc., the matching processing is performed by selecting the optimum one as the limiting condition of the candidate roads. It is an object of the present invention to realize a position detection device that can reduce time.

[0010]

The position detecting device of the present invention determines which of the procedure (a) and the procedure (b) is adopted next in order to select a target link for map matching. The determination is performed based on the estimated time required to execute the procedure, the target links for map matching are selected according to the determined procedure, and the map matching is executed for these links. (a) From a certain range of road map data, find the links that exist within the range of the marginal error of the estimated position caused by the vehicle position error, road map error, etc., and from the links thus obtained, A link is selected that has an orientation whose difference from the estimated orientation of the vehicle is within a threshold value. (b) From the road map data within a certain range, find a link whose direction from the estimated direction of the vehicle is within a threshold,
From the links thus obtained, the links existing within the range of the marginal error of the estimated position caused by the error of the vehicle position and the error of the road map are selected.

According to the above configuration, before selecting the target link for map matching, either (a) or (b) the selection procedure is determined based on the estimated time required to execute the procedure. Therefore, it is possible to actually select target links for map matching and execute map matching for these links in the shortest time.

The "estimated time required to execute the procedure" is the area of the road map within the certain range,
Area ratio of the margin of error of the estimated position, and all directions (3
The estimation may be made based on the ratio of the angle of the threshold to 60 ° (claim 2). According to the second aspect of the invention, the estimated time is estimated based on the assumption that the link density is constant and the link directions are distributed in all directions.

[0013]

Embodiments of the present invention will be described below in detail with reference to the accompanying drawings. FIG. 1 is a block diagram showing one embodiment of the position detecting device of the present invention. The position detection device includes a traveling distance detection unit 1 and a traveling direction detection unit 2
And a road map memory 3 in which road map data of a predetermined range is stored in advance, a traveling distance detection signal and a traveling direction detection signal output from the detection units 1 and 2, and a road stored in the road map memory 3. An estimated position calculating unit 4 that calculates an estimated position based on the map data and the evaluation data output from the matching evaluation unit 6; and a road map data stored in the road map memory 3 and the estimated position calculating unit 4. A target link extraction unit 5 that extracts a target link by using the output position data as an input, and a matching evaluation that evaluates a predetermined magnitude relationship with respect to the target link extracted by the target link extraction unit 5 and outputs a current position. 6.

More specifically, the traveling distance detecting unit 1 has, for example, a magnetic sensor for detecting the rotation of the wheel, and the counter measures the wave number of the sine wave signal output from the magnetic sensor. A so-called wheel speed sensor configured to obtain the number of revolutions and calculate the traveling distance per unit time by multiplying the count data output from the counter by a predetermined constant indicating the outer circumference of the wheel by a multiplier. It suffices to have, but it is possible to use a conventionally known configuration such as a vehicle speed sensor configured to calculate the traveling speed of the vehicle based on the Doppler shift or the like and integrate it to calculate the traveling distance.

As the traveling direction detecting section 2, for example, a gyro which outputs rotation angle data per unit time can be used. Examples of the gyro include a vibration gyro, an optical fiber gyro, and a differential wheel speed sensor. Further, a geomagnetic sensor for detecting a horizontal component of the geomagnetism can be used, and a combination of the geomagnetic sensor and the gyro can be employed.

The road map memory 3 stores road map data covering a predetermined range in advance, and a semiconductor memory, a cassette tape, a CD-ROM or the like can be used. The road map data is obtained by dividing a road map (including a highway national road, a motorway, a national road, a prefectural road, a municipal road, and other living roads) into a mesh shape, and coordinates and adjacent nodes of each mesh unit. Consists of a pointer to the node. A nationally closed network is formed for roads over national roads.

The mesh has a vertical and horizontal distance of about 1 km × 1.
It has a square shape with km. A node is generally
It is a coordinate position for identifying a road intersection or a bending point. A node representing the intersection is called an intersection node, and a node representing a road bending point (excluding the intersection) is called an interpolation point node. Here, the structure of the road map data is illustrated in FIG. Node coordinates are stored for each node in the mesh, and a pointer to a terminal node having the node as a start node is stored. Although not shown, the actual number of nodes in the mesh is stored corresponding to the mesh.

A link connecting the start node and the end node is a link, which is a polygonal line with a direction along the shape of the road. The estimated position calculating unit 4 uses the distance data Δd output from the traveling distance detecting unit 1 and the azimuth data θ output from the traveling direction detecting unit 2 to calculate the east-west direction component Δx (= Δd × cos θ) of the distance. ), And north-south direction component Δ
The current estimated position data (Px, Py) is calculated by calculating y (= Δd × sin θ) and adding the components Δx, Δy to the previous estimated position data (Px ′, Py ′).
Is calculated.

Further, a limit error corresponding to the amount of travel of the vehicle with respect to a range of a limit error (including a distance error, an azimuth error, and an error of a road map) which may be included in the previous estimated position data. By adding the increments of the above, the range of the marginal error that the current estimated position data may have is calculated. Since this marginal error range generally has an elliptical shape, it is conceivable to set the ellipse as it is when setting it on a road map. However, in this embodiment, a rectangular shape is set as shown in FIG. This is because the rectangular shape simplifies the determination as to whether or not it belongs to the marginal error range E. (If it is a rectangular shape, it is only necessary to determine the magnitude relationship of the coordinates. It has to be converted to a curve equation for determination).

It goes without saying that the limit error range E usually extends over a plurality of meshes.
In FIG. 3, a marginal error range E including the estimated position P of the vehicle and four adjacent meshes M ₁ , M ₂ ,
Depicts the M _3, M _4. The target link extraction unit 5 includes:
The link in the marginal error range E is extracted.

If all links in the mesh are targeted, the number is too large, and it is necessary to limit the links.
Therefore, in this embodiment, the target links are limited as follows. In limiting, there are a method of limiting to a link within the limit error range E in the mesh and a method of specifying a link having an orientation close to the orientation of the vehicle among the links in the mesh. The logical product (AND) of the two methods is adopted. However, which method is to be performed first is determined in advance for each mesh.

This determination method will be described. First, the time for determining whether one node in the mesh belongs or does not belong to the marginal error range E is estimated, and this time is defined as t _a .
The time t _a is determined by dividing actually measured by executing a program time the number of nodes. Next, the number of nodes within the marginal error range E in the mesh is estimated. For this reason, it is assumed that the nodes are present in the mesh with a constant density. Focusing on any one of the meshes M _i , the ratio P _a of the overlapping portion (E∧M _i ) of the marginal error range E to the total area of the mesh is calculated. And
The number of nodes N _i in the mesh is multiplied by the proportion P _a of the overlapping portion to estimate the number of nodes existing in the overlapping portion.

Next, of the links in the mesh M _i , the time required to specify a link having an orientation close to that of the vehicle is estimated. Specifically, this time is calculated by using the following equation, tan ⁻¹ (y _j −y _i ) / (x _j −x _i ), from the difference between the coordinates of the nodes, and calculating the link Regarding azimuth, it is the time required to determine whether | vehicle azimuth−link azimuth | <threshold (30 °) holds. This time is t _b per node. This time t _b is
It is determined by actually executing the program, measuring the time, and dividing by the number of nodes.

Next, the ratio of "links having an orientation close to that of the vehicle" with respect to all links in the mesh is estimated. Therefore, it is assumed that the link directions in the mesh are evenly distributed in all directions. Therefore, the ratio P _b is P _b = 2 × threshold value (30 °) / 360 °. Then, the number N _i of nodes in the mesh is _multiplied by the ratio P _b to estimate the number of links having an orientation close to the orientation of the vehicle.

Then, for each mesh, it is determined in advance which of the above two methods is the shorter processing time. The case is first: the mesh M _i, the execution time of the method to determine the links within the limits error range E in the mesh becomes N _i t _a. For the link within the limits error range E in the mesh, the execution time of the method to identify a link having an orientation close to the orientation of the vehicle becomes N _i P _a t _b. Therefore, the sum of the two is N _i (t _a + P _a t _b ). First: For the mesh M _i , the execution time of the method of identifying a link having an orientation close to that of the vehicle is N _i t _b . The link for this specific orientation, the execution time of the method to determine the links within the limits error range E in the mesh becomes N _i P _b t _a. Therefore, the sum of both, the _{N i (P b t a +} t b).

A numerical example will be given. t _a = 12 μsec, t _b
= 11 μsec, area ratio P _a of the marginal error range E in the mesh P _a = 0.1, threshold value for orientation determination 30 °, P _b = 1 /
6, if the number of nodes in the mesh = 10,000, then N _i (t _a + P _a t _b ) = 131,000 [μ _sec ] N _i (P _b t _a + t _b ) = 130,000 [μ _sec ], It can be estimated that the processing is faster if the method of (1) is executed first and then the method of (2) is executed.

If the area ratio P _a of the marginal error range E in the mesh is 0.01, then N _i (t _a + P _a t _b ) = 121,100 [μ _sec ] N _i (P _b t _a + T _b ) = 130,000 [μ _sec ], and it can be estimated that the processing is faster when the method of (1) is executed first and the method of the subsequent method is executed.

Generally, the smaller the area ratio P _a of the marginal error range E in the mesh, the marginal error range E in the mesh.
It is more efficient to determine the links in the first, and the method for identifying the links having the bearings closer to the vehicle bearing is the smaller the threshold value for determining the links having the bearings closer to the vehicle bearing is. It is more efficient to execute first.

Next, the procedure for actually executing the method will be described. The target link extraction unit 5 determines the limit error range E
One or a plurality of meshes having a common part with is specified, and among the nodes included in the mesh, the nodes included in the marginal error range E are extracted. As a result, a link in which both the start node and the end node are included in the limit error range E (see L _{a in} FIG. 3) and only the end node has the limit error range E
Link (see L _{b in} FIG. 3) included in the marginal error range E (see L _{c in} FIG. 3).
Is specified.

Even if the start node and the end node are not included in the limit error range E, as shown in L _d of FIG. This link L _d may be added as it exists. In the determination of the link L _d, the diagonal point coordinates of the marginal error range E are (X ₁ , Y ₁ ), (X ₂ , Y ₂ ), and the start or end node coordinates are (x _i , y _i ), etc. Let (x _j , y _j ) be the node coordinates of the determination formula (X ₁ <x _i <X ₂ ) and [(y _i <Y ₁ ) or (Y
₂ <y _i )] and (Y ₁ <y _j <Y ₂ ) and [(x
_j <X ₁ ) or (X ₂ <x _j )].

As shown by L _e in FIG. 3, a link L _e in which no link is included in the marginal error range E is excluded. In the case of actually executing the above method, the end node in the case where the target node is the start node is specified, and the link azimuth is expressed by the following equation: tan ⁻¹ (y _j −y _i ) / (x _j −x _i ), and for that link bearing, it is determined whether or not | vehicle bearing−link bearing | <threshold value holds, and links that do not satisfy this determination formula Exclude. This threshold value is determined based on the detection error of the estimated heading of the traveling direction detection unit 2.

In this way, when a link that exists in the marginal error range E for one mesh and has a bearing close to the bearing of the vehicle is determined, the same processing is performed for the other meshes. Finally, it exists in the margin of error range E,
And a link having an orientation close to the orientation of the vehicle is found across the mesh. The matching evaluation unit 6 performs a map matching process on these links.

Since this map matching method has been publicly implemented, it will be briefly described below with reference to FIG. Figure 4 is a diagram showing a part of a road map, a number of links, the link L _0, the link L ₁ to the node n ₁ and the starting end, the link L ₂ to the node n ₂ to start, the node n ₃
Only the link L ₃ starting from is shown, and
The state where the vehicle is traveling through the links L ₀ to L ₁ is shown by a broken line S. It should be noted that the alternate long and short dash line T indicates the locus of the estimated position, a, b, c, d indicate branches or curves, and E indicates the marginal error range.

First, the matching evaluation unit 6 calculates the distance (correction) between the estimated position P of the vehicle and each node n ₁ , n ₂ , n ₃ at the time when the estimated direction of the vehicle is largely rotated (at the time of turning the intersection). Calculate the initial value). Then, as the vehicle travels, the distances from the estimated position of the vehicle to the links L ₁ , L ₂ , L ₃ starting from the nodes are added to the distances. This value is called “correction value γ”.

At the same time, the matching evaluation unit 6 obtains the heading difference between the estimated heading of the vehicle and the link heading. This azimuth difference is called “correlation degree α”. The matching evaluation unit 6 evaluates the magnitude relationship between the correlation degrees α corresponding to the respective links, specifies the link corresponding to the smallest correlation degree α ₀ ,
One or a plurality of links having a correlation degree α whose difference | α−α ₀ | from the correlation degree α ₀ is smaller than the threshold value are determined.

Then, the change amount Δγ of the correction value γ is obtained for each of the determined links, and this change amount Δγ is obtained.
Determines the smallest link and identifies the current position of the vehicle on this link. Since the amount of change Δγ changes every moment, an average of the latest period may be taken, and this average value may be used as a determination target. At this time, the correction values γ for other links are also kept stored in principle. Therefore, as the vehicle travels, one link and some candidate links are saved.

The link is deleted when the correlation degree α becomes larger than the threshold value or when the correction value change amount Δγ becomes larger than the threshold value. That is, the correlation α
Is performed at predetermined timings, and the registration of the link corresponding to the correlation α that is larger than the predetermined threshold value or the link whose change amount Δγ of the correction value is larger than the threshold value is deleted. The correction value γ corresponding to the link whose registration has not been deleted will be sequentially updated to a new correction value γ in the target link extracting unit 5 thereafter, and the above evaluation will be performed again. Then, by continuing the above-described series of evaluation operations, only one link is finally left under the influence of a road branch or the like, and the position on the remaining link is continuously displayed as the current position.

The embodiment of the present invention is not limited to the above-described embodiment. For example, instead of the traveling distance detecting section and the traveling direction detecting section, a GPS (G
(lobal Positioning System) It is also possible to use a receiver. In addition, various design changes can be made within the scope of the present invention.

[0039]

As described above, according to the position detecting device of the present invention, before selecting a target link for map matching, a selection procedure that can be performed in the shortest time is set as an estimated time required to execute the procedure. Make a decision based on Therefore, it is possible to actually select target links for map matching and execute map matching for these links in the shortest time, compared to the conventional method that fixed the target link selection procedure. Thus, the time required for the map matching process can be shortened.

[Brief description of drawings]

FIG. 1 is a block diagram illustrating an embodiment of a position detection device according to the present invention.

FIG. 2 is a structural diagram of road map data.

FIG. 3 shows a marginal error range E including an estimated position P of a vehicle and four adjacent meshes M ₁ , M ₂ , M ₃ ,
Is a road map depicting the M _4.

FIG. 4 is a diagram showing a part of a road map for explaining a map matching method.

[Explanation of symbols]

 1 traveling distance detection unit 2 traveling direction detection unit 3 road map memory 4 estimated position calculation unit 5 target link extraction unit 6 matching evaluation unit

Claims (2)

[Claims] 1. A position detecting device for determining an estimated position of a vehicle and executing map matching between the estimated position and road map data to identify a current position of the vehicle on a road. To select
Decide whether to adopt the procedure of (a) or the procedure of (b) based on the estimated time required to execute the procedure, and select the target link for map matching according to this determined procedure. Then, a position detection device characterized by executing map matching for these links. (a) From a certain range of road map data, find the links that exist within the range of the marginal error of the estimated position caused by the vehicle position error, road map error, etc., and from the links thus obtained, A link is selected that has an orientation whose difference from the estimated orientation of the vehicle is within a threshold value. (b) From the road map data within a certain range, find a link whose direction from the estimated direction of the vehicle is within a threshold,
From the links thus obtained, the links existing within the range of the marginal error of the estimated position caused by the error of the vehicle position and the error of the road map are selected. 2. The "estimated time required to execute the procedure" is the area of the road map within the certain range,
Area ratio of the margin of error of the estimated position, and all directions (3
The position detecting device according to claim 1, wherein the position detecting device is estimated based on a ratio of a threshold angle with respect to 60 °.