JP2633630B2 - Navigation device for vehicles - Google Patents

Description

Description: BACKGROUND OF THE INVENTION 1. Field of the Invention The present invention relates to a navigation device for a vehicle, and more particularly, to a method in which a current position of a vehicle estimated by dead reckoning is matched to a road on a map in which the current position is stored in advance. The present invention relates to a vehicle navigation device for guiding the traveling of a vehicle.

[Related Art] In recent years, in vehicle navigation, for example, there is a dead reckoning method of estimating a current position of a vehicle based on a traveling direction of the company obtained by using geomagnetism and an accumulated traveling distance. If this dead reckoning is applied to vehicle navigation, the detected position of the vehicle and a map of the surroundings are displayed on the screen of the display device to guide the traveling of the vehicle. However, it is said that this dead reckoning navigation has a weak point where errors are accumulated.

As an improvement of this weak point of dead reckoning, for example,
There is a method called map matching. This map matching is, for example, as described in Japanese Patent Application Laid-Open No. 61-209316, when the vehicle has passed the position characteristic of the shape of the road such as an intersection, the current Is corrected to the above characteristic position.

However, the above-mentioned map matching is based on dead reckoning navigation. Further, the above-mentioned characteristic points do not always exist conveniently. Even if measurement errors are accumulated to the extent that map matching cannot be performed between one characteristic point and the next point, map matching will be performed, which is different from the road on which the vehicle is actually traveling. On the road, the current position may be matched. And once
When the current position of the own vehicle is matched on a different road, it is impossible to perform the matching later thereafter.

In view of this point, the applicant of the present application has developed a method for surely performing map matching without relying on uncertain intersections or the like that may not be available when map matching is truly needed (Japanese Patent Application No. 63-163). −106556).

It combines dead reckoning and map matching, sets a plurality of data nodes along the roads on the map, and connects each node with coordinate position information and roads. Is to have a link relationship with the node. The road previously stored in this way is referred to as a “stored road”. Accordingly, since the map matching can be performed periodically, the map matching can be performed before the measurement error of the vehicle position obtained by dead reckoning becomes too large. Therefore, there is no need to correct the position of the vehicle on the wrong road.

[Problems to be Solved by the Invention] By the way, the above-described map matching generally searches for the most appropriate one for the current position estimated by dead reckoning from among many candidate points. The burden on the CPU for the control circuit and the like of the navigation device becomes large. Furthermore, since the map matching is performed in real time while running, the excessive load is even more significant.

On the other hand, for example, on a road such as an expressway where a certain running shape continues, no problem occurs even if map matching is not frequently performed. The reason is that the frequency of map matching is reduced, and as a result, even if the current position accuracy due to dead reckoning is reduced and map matching is normally difficult, as described above, constant driving such as an expressway This is because, while the vehicle is traveling on a road having a continuous shape, even if the map matching is performed on this highway, the matching will not be mistaken.

Therefore, the present invention aims to reduce the load on the navigation device by changing the map matching pattern to the one most suitable for the road on which the vehicle is currently traveling.

[Means and Actions for Solving the Problems] In order to solve the above problems, a navigation device for a vehicle according to the present invention, based on the traveling direction and traveling distance of a vehicle, as shown in FIG. Estimating means for estimating the current position by dead reckoning, and information on roads on a map, and a plurality of types of matching patterns for map matching including different weights depending on the types of roads, corresponding to the types of roads. Storage means for storing; determining means for determining the type of the road currently traveling from the current position obtained by the estimating means; and a matching pattern corresponding to the road type based on the determination result by the determining means. It reads out from the storage means, performs map matching according to this matching pattern, and stores the current position obtained by the estimation means on a road on the map. Map matching means for determining whether or not map matching should be performed by evaluating the traveling direction of the vehicle related to the road estimated to be currently running. The evaluation unit changes the weight of the evaluation with respect to the traveling direction according to the type of the road determined by the determination unit and changes the weight in accordance with the weight value read from the storage unit. .

That is, the map matching pattern is switched according to the type of the road on which the vehicle is currently traveling, so that a map matching method suitable for the road is applied. In particular, the evaluation means uses a weight value corresponding to the traveling road when evaluating whether or not map matching should be performed by evaluating the traveling direction of the vehicle associated with the road estimated to be currently traveling. This makes it possible to perform an evaluation suitable for the traveling road.

According to a preferred aspect of the present invention, the matching pattern includes a map matching evaluation threshold value that differs depending on the type of road, and the evaluation unit determines whether or not to perform map matching on the type of road. It is characterized in that it is changed according to the corresponding threshold value, and the accuracy of evaluation is further improved.

According to a preferred aspect of the present invention, when a road estimated to be traveling is estimated to be a straight road, unnecessary processing is reduced by reducing the frequency of executing map matching.

Hereinafter, preferred embodiments of the present invention will be described with reference to the accompanying drawings.

<Configuration of Example Device> FIG. 2 is a control system diagram of the vehicle navigation device according to the present example. This navigation device 10
Is a terrestrial magnetism sensor 11 for detecting the traveling direction (D _M ) of the own vehicle, and a vehicle speed sensor 12 for calculating a traveling distance of the own vehicle, for example, outputting a pulse signal _REV for each rotation of a tire.
A map information storage device 13 in which map information such as a map in which the contents necessary for traveling guidance of the vehicle such as roads and buildings are represented, and map information stored in the map information storage device 13 And a control circuit 15 configured by a CRT or the like to display the road map on the display screen.

The control circuit 15 has an arithmetic circuit 16, a storage circuit 17 and an input / output interface 18 connected thereto, and the geomagnetic sensor 1 is connected via the input / output interface 18.
1. A vehicle speed sensor 12, a map information storage device 13, and a display 14 are connected to the arithmetic circuit 16. Accordingly, a signal indicating the traveling direction of the vehicle from the geomagnetic sensor 11 and a signal indicating the traveling distance of the vehicle from the vehicle speed sensor 12 are input to the arithmetic circuit 16 via the input output interface 18. ,
As a result, the estimated position of the own vehicle is calculated in the arithmetic circuit 16, and the calculation result is temporarily stored in the storage circuit (RAM) 17. In the control circuit 16,
The surrounding map is drawn from the map information storage device 13.
Then, the map matching is periodically performed, and when the map matching is not performed, the estimated position of the vehicle detected as described above is displayed, and when the map matching is performed, the vehicle's estimated position is displayed on the map. The estimated position is matched, and the position on the map is displayed on the screen of the display unit 14. The map information storage device 13 includes, for example, a CD-ROM and a CD drive (player) for reproducing the CD-ROM.

In the present embodiment, the map information storage device
As map information stored in 13, a large number of nodes are set on the road and stored in advance, and data indicating the connection relationship of these nodes is input. These nodes are, for example, intersections and corners. As well as the department,
It is assumed that they are appropriately provided at predetermined intervals on a straight road.

<Node> FIGS. 3A and 3B show the relationship between a road and nodes set on the road. In Figure 3A, there is shown a grid-like road as an example, the node N ₀ to N ₈ at each intersection of these roads are set. As described above, the map information storage device 13 stores the position coordinates of these nodes and, for each node, the connection relationship between that node and other nodes. The connection relationship in terms of the node N ₀ of Figure 3A, the N ₀ directly represented as a link (i.e., first layer) connected to the other nodes. Then, around the node N _0, a is Figure 3B that represent conveniently draw a connection relationship between the nodes.

<Data Used for Control> FIG. 4 shows the storage states in the RAM of flags, intermediate data, and the like used in a program for controlling this embodiment. These data will be briefly described.

In d, the distance traveled by dead reckoning is stored. In addition, D _M is the azimuthal angle of the current read geomagnetism.
D _P is the average value of the geomagnetic direction is stored. D _P0 , D _P1 , ‥‥
Each of D, D _Pi , ‥‥‥, and D _Pn is a list sequentially stored with the movement of the vehicle having the geomagnetic average value D _P. In this embodiment, D _P is the average value of Dm data measured 32 times.

P _x is the estimated position of the vehicle computed by dead reckoning, a display position of the vehicle for P _D is to be displayed on a display device 14. P _C is the position when it recognizes a corner a corner of the vehicle has passed. P _L is the position recognized as out of the memory road.
P _I is a position set the virtual node. N _I denotes the number of the virtual node. N _L is the name of the last node that arrived.

F _ON is a flag that stores the recognition result of whether the _vehicle is currently traveling on the storage road (F _ON = 1) or not (F _ON = 0). F
_N is a flag indicating that a virtual node is currently being set. F _C is a flag for storing the detection of the corner passage of the road. Further, F _D is a flag for storing the detection of the corner entrance. F _A is a flag indicating that the node has arrived.

N _j , _{DC Lj} , and I _xj make up a candidate node list for driving on a storage road, and each element is indexed by a subscript j. N _j
Is the node name, and _DCLj is the last node that arrived (= _NL )
J from the direction of the candidate node j, the I _xj is the current point P _x
This is the remaining distance to the candidate node.

N _k , D _CIk , and I _xk make up a candidate node list while traveling on a road outside of memory, and each element is indexed by a subscript k. N _k
Is the node name, D _CIk is the direction from the virtual node to the k candidate node, and I _xk is the remaining distance from the current point P _x to the k candidate node. C _k is used to evaluate each candidate node
Indicate the number of D _P.

<Control> Main Routine FIG. 5 is a flowchart showing the entire navigation control according to the present embodiment. Fig. 6A is on a storage road (F _ON
= 1), a program flow chart relating to the node search control, FIGS. 6B and 6C show a program flow chart relating to the node search control when traveling on a road outside the memory, and FIG. 6D shows a corner detection control. FIG. 6E is a program flow chart relating to map matching control.

First, the entire navigation control will be described with reference to FIG. The program shown in FIG. 5 is an interrupt routine that is started every time a _REV pulse is input from the vehicle speed sensor 12.

Step S2 is an update step of the vehicle moving distance d. _REV is generated for each revolution of the vehicle. In this embodiment, since the vehicle travels 0.4 m with one rotation of the wheel, d = d + 0.4. In step S4, the geomagnetic azimuth angle _DM is read from the geomagnetic sensor 11. In step S6, it is determined whether the vehicle has performed the operations in steps S2 and S4 32 times.
Until that time, the target node search is not performed by returning to the main routine from step S6.

When the data of azimuth D _M 32 number collected, the process proceeds to step S8, determining 32 the moving average value D _P azimuth D _M. This moving average is, as mentioned above, It is. With one _REV interruption, the vehicle will travel 40 centimeters, and 32 times will travel 12.8m. In other words, the control from step S8 is activated every time the vehicle travels 12.8 m. In addition, this 3
The number of times may be changed according to the size of the wheel.

In step S10, the D _P obtained in step S8, numeric columns D
Store in _Pn . The numerical sequence D _Pn is an array (= array) for storing the D _P calculated for each 12.8 m from the last reached node point until reaching the next node.
This arrangement will be described with reference to FIG. _Assuming that D _P0 is the one calculated at the last arrival at the node, the number of D _Pn increases as the _distance from the node increases, such as D _P1 and D _P2 . At present, the counter n stores the number of stored data. In step S12, the counter n is incremented.

In step S14, a current position P _x (x, y) based on dead reckoning is calculated based on the section movement distances d and D _P. In step S16, the flag F _ON or traveling on a storage road
Find out from.

Node search (on storage road) A case where the vehicle is currently traveling on a storage road (F _ON = 1) will be described first. FIG. 8 illustrates a case where the last arrival node is N ₀ , and nodes N ₁ , N ₇ , N ₅ are linked from that node, and the vehicle is actually heading to node N ₇ . . Then, it is assumed that the traveling direction data D _{P0 to} D _P5 have been calculated before reaching the current position P _x (x, y). Note that nodes N ₁ and N
_7, N ₅ at the time of arriving at the node N _0, are those become known from the connection relationships between nodes that are also connexion stored before a movable node will be referred to such movable nodes and candidate nodes .

Now, in the memory road travel in, from among the nodes N _1, N _7, N ₅ are connected will be described technique to identify the traveling direction of node N _7.

The connection relationships between nodes N ₀ and N ₁ , N ₀ and N ₇ , and N ₀ and N ₅ have been previously _solved as connection directions D _C01 , D _C07 , and D _C05 , respectively.
From among the three candidate nodes, to explore the destination node, for each candidate node, compared with the connecting direction D _C between the vehicle traveling direction D _P and the node, and the destination node are relatively close candidate node You do it. In particular, in this embodiment,
It performed based on the comparison to the evaluation function T _j.

_{θ i = | D Pi -D CLj} | where, j denotes the j-th candidate node, D _CLj is connecting direction value between the final arrival node N _L and the candidate node N _j. In addition, Σ is calculated for i with respect to D _Pi stored up to the current location P. for division by n, the connection direction value D _C in the traveling direction with respect to the stored value D _P0 to D _Pn
This is to obtain the average value of the differences. A node having a smaller value of the evaluation function _Tj is a candidate node closer to the destination node.

For example, in the example of FIG. 8, the candidate node N ₁ is For candidate node N ₇ , For the candidate node N _5, Whether the node is a destination node is determined based on whether the value of the evaluation function T for each of the candidate nodes is larger than a predetermined threshold K.

In this determination, whether or not a node j is the destination node, go Shibotsu the destination node by an erase method of the evaluation value T _j is the case of larger than the threshold value is not considered to destination node . The weight W of the shall have the characteristic values shown in FIG. 9 in accordance with the value of the difference θ between the connecting direction D _C to the traveling direction D _P. The reason for giving the characteristic as shown in FIG. 9 is that the larger the value of the direction difference θ, the earlier the candidate node having the direction difference needs to be deleted from the candidates.

Return to the flow chart, and in step S16, F _ON is set to "1".
At this time, the details of the node search subroutine in step S18 will be described with reference to FIG. 6A.

First, in step S100, the arrival flag F _A is set to "0". This flag is set to "1" when it reaches the next node (step S132). In step S102, the subscript counter j is set to 1. Since this counter j is an index indicating a candidate node, j = 1 is the first node in the candidate node list.

The loop from step S104 to step S108 to step S126 is applied to each of the candidate nodes pointed to by j. That is, the evaluation function T _j is calculated for the node j in steps S104 and S106. In step S108
Compare _Tj with threshold K.

If T _j > K, it is considered that the node of j is not the destination node, so in step S109 this j node is deleted from the list of candidate nodes.

When the T _j <K, enough to be removed from the candidate node currently the j, be considered that not deviate from the traveling direction D _P, not delete, the process proceeds to step S120, the current point P The distance I _xj from _x (x, y) to the node of j is calculated. In step S122, it is checked _whether the distance I _xj has become 0 or less. When I _xj becomes 0 or less, it means that the node of j is a destination node.

If I _Xj ≦ 0, the subscript j is incremented to check the next candidate node.

Then, the process returns to step S104 via step S126, and it is detected that the above operation has been executed for all the candidate nodes in the next node: in step S126, or in step S110, the number of remaining candidate nodes is “0”. Or: Repeat in step S122 until it is detected that the target node has been attached.

The case will be described. If it is determined in step S126 that all candidate nodes have been checked, the process returns to the main routine of FIG. Then, the corner detection subroutine of step S22 is executed to detect the corner (the details will be described later), and the map matching is executed in step S24. In step S26, based on the result of the map matching, Whether the position P _x (x, y) is displayed (flag F _A = 0), (II) the position of the arrival node is displayed (F _A = 1), or (III) the node position near the recognized corner is Show (F _C = 1).

When the program of FIG. 5 is started 32 times by the interruption by the next 32 _REV pulses, the new traveling direction D
_P (step S8), current position P _x (x, y) (step S14)
Is calculated, and the node search subroutine of FIG. 6A is executed.

By repeating such an operation, nodes satisfying T _j > K are excluded from candidates.

FIG. 7 shows that the number of candidate nodes is initially 20 but is gradually deleted.

If the vehicle continues to run on the storage road, it should arrive at a node where I _xj ≦ 0 (step S122). In this case, the process proceeds to step S130, and the node _satisfying I _xj ≦ 0 is added to the list of arrival nodes. That is, this step
In S130, a list of nodes that have passed in the past is formed. In step S132, the flag F _A indicating that the arrived at the node 1. The flag F _A is used to Matsupumatsuchingu as described below. Then, the latest arrival node _NL is updated, and the counter n is set to "0" in step S136. That is, D _P0 ,
D _{1 to} D _Pn are cleared to “0”. In step S138,
For the node N _L arriving now examine the road information on the connection relations such as the connection relationship seeking (Figure 3B) and connecting direction D _C, it makes a candidate node list (Fig. 4).

Node Search (Road Driving Outside Memory) Next, a case where the number of remaining candidates becomes 0 in the process of deleting candidate nodes in step S109 will be described. At this time, the flag F _{ON is set} to “0” in step S112 (see FIG. 7). That is, when the candidate becomes “0”, it is determined that the vehicle is traveling outside the storage road. And
In step S113, the current position P _x (x, y) is stored as the position P _L lastly deviating from the storage road, as P _L (x, y) = P _x (x, y). Also in this case, when branching from a storage road to a road outside of storage, a node is usually set at a branch point to such a road outside of storage. However, for example, when a new road is laid after the storage of the road information in the map information storage device 13, a node is not always provided at the branch point. Therefore, it is to store the P _x as P _L.

Thus, after storing the point out of the last stored road, further when 12.8m traveling calculates the traveling direction D _P from P _L in step S8, calculates the current position P _{x (x,} y) at step S14 Then, the node search subroutine of FIG. 6B is executed.

About the outline of the node search outside this storage road,
This will be described with reference to FIG. In Figure 10, outside the storage road at a point P _L, shows a case advanced the _{P x0 → P x1 → P x2} → P x3 → P x4 every 12.8m. First, after leaving the memory road,
When _traveling to the first point P _x0 , draw a circle having a radius of r, which is a value obtained by multiplying the distance L _x that has been traveled by the predetermined constant β, and examine whether a real node exists in the circle. . If a real node exists in this circle, this P _x0 is set as a virtual node. In the example of FIG. 10, the virtual node N _I is set at P _x0 .
_{Are set,} and nodes N ₁ and N ₂ exist in a circle of r ₁ = β × (distance between P _x0 and P _L ). From this virtual node N _I, it calculates a connection direction D _CIk to actual nodes. Here, k is a subscript for indexing a real node. Note that this calculation of D _CIk is calculated based on the coordinate value of the actual node existing in the current value P _x and a circle based on dead reckoning. Nodes within these circles are relatively likely to be nodes arriving at any of them, and will be referred to as candidate nodes as in the case of node search on a storage road. Then, from these candidate nodes, nodes that are not likely to be destination nodes are excluded. The exclusion algorithm is substantially similar to the above-described node search on the storage road. That is, if a candidate node is indicated by a subscript k, Nodes having a large evaluation function T _k of θ _i = | D _Pi −D _CIk | are excluded. here,
In the above equation, the denominator C _k will be described below.

In the case of FIG. 10, since there are two nodes N ₁ and N ₂ , the connection directions D _CI1 and D _CI2 with these nodes are calculated,
Further, for the nodes N ₁ and N ₂ , T ₁ = W × | D _CI1 −D _P0 | T ₂ = W × | D _CI2 −D _P0 | is calculated, and the T ₁ and T ₂ are set as thresholds K ′. Compare. Then, nodes having an evaluation value larger than the threshold value K 'are excluded. Note that this threshold value K 'has a value different from K in the above-described node search on the storage road, and is selected as K'<K. The refinement has become severe.

T ₁ = 1/2 ｛W · | D _P0 −D _CI1 | + W · | D _P1 −D _CI1 |｝ T ₂ = 1/2 ｛W when the vehicle advances from P _x0 to P _x1・ | D _P0 −D _CI2 | + W · | D _P1 −D _CI2 |｝, and when it progresses to P _x2 , T ₁ = 1/3 ｛W · | D _P0 −D _CI1 | + W · | D _P1 −D _CI1 | + W · | D _P2 −D _CI1 |｝ T ₂ = 1/3 ｛W · | D _P0 −D _CI2 | + W · | D _P1 −D _CI2 | + W · | D _P2 −D _CI2 |｝ It is evaluated by.

Meanwhile, C _k of the denominator of the evaluation formula, as in the case of node searching in the memory road described above is the number of D _P from P _L. Also, in the case of searching for a node on the storage road described above, the next target node must always exist in the candidate node. However, traveling outside the storage roads, guarantee that the target node exists in the circle of radius r ₁ which the initially calculation is not. Therefore, in this embodiment, P _x0 → P _x1 → P
_{As the} vehicle progresses as _x2 → P _x3 → P _x4 , at each point, r = β × L _x is calculated from the traveling distance L _x from _PL to each point, and within this radius r, To check if a real node has been found. If it is newly found, it is added to the list as a candidate node. In such a case, the problem is how to evaluate the connection direction between the virtual node and the newly discovered new candidate node. This will be described in more detail. In FIG. 10, until P _x2, and new node has failed discovered. And
In P _x3, newly in a circle with a radius r3, a node When N ₃ is found, the evaluation function T ₃ of _{N 3 T 3 = 1/4} {W · | D P0 -D CI3 | + W · | D _{P1 -D CI3 | + W · |} D P2 -D CI3 | + W · | D P3 -D CI3 |} and should be, or, T ₃ = W · | it should be | D _P3 -D _CI3. In this embodiment, the latter evaluation function is used. That is, when a new node is found, only the traveling direction after that point (in the example of FIG. 10, only D _P3 ) is to be evaluated. If the vehicle further proceeds from P _x3 to P _x4 , this newly added node N
₃ will be evaluated by D _P3 and D _P4 .

The reason for this is as follows. Roads that are not stored mainly occur when small roads are omitted due to the storage capacity of road information. Therefore, it should be noted that such a small road need not be stored as node information only if it is a substantially straight road. In such a case, the evaluation of the newly discovered nodes, are lines of connexion from P _L, also discovered point (the
In the example of FIG. 10, the same result can be obtained even when the vehicle _travels from _Px3 ) because the traveling direction of the vehicle is almost immediately.
However, in the case of a newly laid road, it is impossible to predict how the road will bend. If, when the memory outside road is large Kyokutsu, the newly discovered nodes can be evaluated by adding also the previous D _P is because thereby increasing the retirement connexion error. For example, as shown in Fig.
Actual storage outside roads have summer as flex greatly P _x3, if the vehicle has to be Kotsu the node N ₃ is the addition of travel direction data D _P0 to D _P3 on the evaluation of the node search This obviously increases the evaluation error.

Returning to FIG. 6B, the details of the control while traveling on the road outside the memory will be described.

First, in step S150, the node arrival flag F _A is reset. In step S151, the point P _L
Calculating a cumulative traveling distance L _x to the current point P _x from. Then, in step S152, a circle having a radius of r r = L _x × β is drawn, and it is checked whether a real node exists in the circle.
Here, β is set to 20% (= 0.2) as an example, and r is fixed within the following range.

50m ≦ r ≦ 500m If there is no node in this range, step S156
In, examine the flag F _N. This flag _FN is a flag indicating that a virtual node has been set. If not, the flow advances to step S158 to delete the _DPI stored in step S10 (FIG. 5). This is to ignore the traveling direction data before the virtual node is set. Then, the process returns to the main routine.

After returning to the main routine, the vehicle travels 12.8 m further, and then the control shown in FIG. 6B is executed as follows.

If a node is found in the circle in step S154,
Proceed to step S160. In step S160, the flag F _N is checked whether it is excisional. Only when the flag is not excisional, it sets a virtual node N _I at step S164. That is, excisional the in step S162 the flag F _N, in order to set the virtual node N _I to the current point P _x at step S164, the P _x and the virtual node point P _I. And step S1
In step 66, the real node found in step S154 is set as a candidate node.
And N _k, to create a connection relation list between those candidate nodes N _k and the virtual node N _I. That is, the k-th candidate node
A connection direction D _CIk between N _k and the virtual node N _I is created for each candidate node found in the circle. Then, in step S168, the counter C _k for each candidate node N _k 1
Initialize to Then, the process returns to the main routine.

The meaning of this C _k = 1, the like should be evaluated as a traveling direction data refers to a single D _P which is excisional at step S10. Because, the D _P of up to it is because has been deleted in step S158.

Thus, a flag indicating that the virtual node has been set
F _N is set to 1. Next, operation control during the flag F _N = 1 will be described.

Further, when the vehicle travels 12.8 m and proceeds to step S154, it is checked whether a node exists within the circle of the radius r calculated in step S152. Generally, there is a high possibility that the node detected last time is detected as it is. In such a case, proceed to step S160 → step S174.
It is checked whether or not there is a new node that is further added to the previously detected candidate node. If there is nothing new, the process proceeds to step S182, for each existing candidate node, increments the counter C _k for storing the number of the traveling direction data D _P to be evaluated. In the example of Fig. 10, P _x2
Until then, no new candidate nodes have been found, so D
_In order to indicate that _{P0 to} D _P2 are targets for evaluating the candidate nodes N ₁ and N ₂ , C ₁ = C ₂ = 3 should be satisfied.

If a new candidate node is found in step S174, the process proceeds to step S176 for additional candidate nodes.
Creates a list N _k , D _CIk of candidate nodes. And
For this new candidate node, C _k = 1, and for the existing candidate node C _k , step S180
And C _k = C _k +1.

After the execution of step S180 or step S182, the process proceeds to step S190 in FIG. 6C.

If the candidate node no longer exists in the circle in step S154 after F _N = 1, as long as the candidate node discovered in the past is not deleted (F _N = 1), step S156 is performed. Then, the process proceeds from step S174 to step S182 to set C _k = C _k +1 for the remaining candidate nodes, and then returns to the main routine.

The program in FIG. 6C will be described. The control of this program is an evaluation control of each candidate node as a target node.

In step S190, a subscript k indicating a candidate node is initialized.

The loop from step S192 to step S204 → step S192 evaluates all candidate nodes as destination nodes. That is, in step S192, k
Is the connection direction D between the candidate node N _k and the virtual node N _I
And _CIk, the directional difference theta _i of the respective travel direction data D _Pi, θ _{i =} | computed from | D _Pi -D _CIk. Then, in step S194, for the k candidate nodes, the number of the latest traveling direction data D _Pi corresponding to the number held in the counter C _k of this candidate node is set.
_{Is used} to calculate the evaluation function Tk.

Then, at step S196, and compares the T _k and the threshold value K ', to determine whether to candidate node N _k of k is excluded as a candidate.

If it is within the threshold, the current position P is determined in step S198.
Calculate the distance I _xk from _x (x, y) to the candidate node,
In step S200, it is determined whether the candidate node is sufficiently close. If not, the process proceeds from step S202 to step S204 and then to step S192, and the above control is repeated.

If T _k > K ′ in step S196, in step S230, the k candidate node is excluded from the candidate list. Then, if there are any remaining candidate nodes in step S232, the process proceeds to step S202, and the control is performed.

The case where I _xk ≦ 0 in step S200 will be described. In this case, in step S210, the k candidate nodes are listed as arrival nodes, and
In S212, F _ON = 1 is set in order to reset F _N and store the fact that it has returned to the storage road. Then, in step S214, this arrival node is stored as the latest arrival node _NL . In step S216, the arrival flag F _{A is set} to F _A = 1. Then, in step S218, a list of candidate nodes movable from the arriving node is created for the node search control of FIG. 6A. This candidate node list is used in the node search control (during running on the stored road = FIG. 6A). In step S220, the list of candidate nodes _Nk is cleared.

In step S232, if the remaining candidate node is determined to not exist, the process proceeds to step S234, to reset the flag F _N. And, since the virtual node N _I ever believed that meaning is that there is no, in step S236, to remove all of the candidate list, at step S238, new, point out the current position from the storage road P _{As L} , the same control is repeated again from step S150 in FIG. 6B. That is, searching for a candidate node → setting a virtual node → evaluating a candidate node. Referring to FIG. 11, the reason why the virtual node setting is repeated again even if all the candidate nodes evaluated from the virtual node once set are deleted from the candidates will be described. (A) of FIG.
In FIG. 7, a node is indicated by ◯, a storage road is indicated by a broken line, a non-stored road is indicated by a dashed line, and a solid line indicates a traveling locus of the vehicle. Now, the vehicle is shown in FIG.
It is assumed that a virtual node is set at the time when the vehicle enters a road outside the memory (F _N = 1). A virtual road is drawn from this virtual node to nodes N ₃ and N _4.
It is assumed that ₃ is excluded from the candidates. It is assumed that the vehicle further advances and N ₄ is also excluded as shown in FIG. 3C (F _N = 0).
As the vehicle further advances, the virtual node is set again (F _N = 1) and the node N ₄ is set as the destination candidate node, as shown in FIG. Is so different that this N ₄ is excluded from the candidate and again (F _N =
0).

In the state of FIG. 11 (e), when the vehicle returns to the storage road, the returned point is not a node.
The virtual node is set again. Then, it sets N ₄ again as a candidate node. Vehicle proceeds aerodrome N _4, this time, since the moving direction of the connection direction and the vehicle between virtual nodes and N ₄ are the same, until arriving at the N _4, always N ₄ seems relatively certainly candidates It will continue to be recognized as a node.

As described above, if the virtual node setting fails once while traveling on the non-memory road, the virtual road may return to a point on the storage road where no node is set, so it is necessary to set the virtual node again. There is.

<Corner Detection> The principle of corner detection control will be described with reference to FIG. This corner detection is based on the following principle. In this embodiment, the nodes are set such that the nodes appear on a straight road before the position detection error by dead reckoning becomes too large, and the nodes are always set at intersections, curves, corners, and the like. I have. Therefore, if an intersection, a curve, a corner, or the like is detected, map matching is performed on a node in the vicinity thereof. Detection of intersections, curves, corners, and the like is performed by judging that the vehicle has passed the corner when each road in the traveling direction of the vehicle changes greatly.

In Figure 12, the excisional the flag F _D upon detecting an entering from straight road curved road by l times (= corresponding to the traveling distance of 0.4l m) collecting the traveling direction data D _P of, after the Furagusetsuto by l times (= corresponding to the traveling distance of 0.4Lm) collecting travel direction data D _P of, when determining that Modotsu the straight road running, it is determined that passes through the corner, as well as reset the F _D, to excisional the corner detection flag F _C. When this flag F _C is excisional the step S24 looks for (Figure 5) rows of connexion neighboring nodes the Matsupumatsuchingu of, is that to correct the current position to the node.

The corner detection control is performed by the flow chart in Fig. 6D.
Now, a description will be given of a state in which the vehicle is traveling on a straight road, and will soon approach a corner.

First, in step S300, the latest l direction data
The D _Pi is read, and in step S301, the D _Pi is integrated (moving average) to estimate an average traveling direction over a distance of 0.4 lm.

In step S302, the difference from the previously calculated moving average value is calculated. In step S303, it is determined whether the difference is larger than a threshold value to determine whether the vehicle is currently traveling on a straight road. If it is determined that the road is straight,
Check excisional state of F _D at S304. Since it is assumed that the vehicle is still traveling on a straight road, F _D = 0 and therefore,
In step S306, the process returns to the main routine stores the moving average value calculated in step S301 as a straight road traveling direction phi _1.

After that, when the vehicle further travels and enters a curved road, the process proceeds from step S303 to step S308, and now F _D = 0
Since it is, the process proceeds from step S308 to step S310, in order to memorize that enters the corner, and excisional the flag F _D, and returns to the main routine.

F _D = 1 After that, the vehicle further travels and the flow of the step S300 → the step S303 → the step S308 → the main routine is repeated while the traveling on the curved road is detected.

Upon returning to the straight road, the process proceeds to step S300 → step S301 to calculate a moving average value, and further proceeds to step S303 → step S304 → step S312. Here, by resetting the flag F _D, in yet step S314, step S301
In the traveling direction phi ₂ after corner escape calculated average value.

Further, in step S316, the absolute value Δφ of the traveling direction difference before entering the corner and after exiting the corner is determined.

Δφ = | φ ₁ −φ ₂ | In step S317, φ ₁ = φ ₂ is set for the next corner detection. Then, in step S318, by comparing this Δφ with the site H, it is determined in step S318 whether this corner is a real corner. If Δφ ≦ H, the process returns to the main routine. If, if Atsuta in [Delta] [phi> H, in step S320, the control of hitherto been determined that detects a true corner to excisional flag F _C.

F _C = 1 Then, in step S322, the current position P _x (x, y) (= corner exit point) is stored as the corner position P _C (x, y).

P _C (x, y) = P _x (x, y) As described above, a true corner is detected while the noise-like curve is not erroneously detected as a corner, and the result is represented by a flag F. _C and the corner point P _C (x, y) are stored.

<Matsupumatsuchingu> flags affect this Matsupumatsuchingu control is the node arrival flag F _A and the corner detection flag F _C.

When F _A = 0 and F _C = 0 At this time, the vehicle is traveling on either the storage road or the non-storage road and is halfway between nodes. At this time, in step S404, the position P _x (x, y) based on the dead reckoning calculated in step S14 is displayed on the display 14
_D (x, y).

When F _A = 1 In this case, whether the vehicle has arrived at the node while traveling on the storage road (step S130), returns to the storage road while traveling on the road outside the storage and arrives at the actual node (step S210).
At this time, in step S402, the current position P _x (x,
y) is corrected to the coordinate position (stored) of the arrival node. Further, at step S403, the display position P _D the position of the arrival node.

When F _C = 1 This is when a corner is detected. First, at step S408, the corner position P detected at step S322
Check if a node is within a certain range centered on _C (x, y). If there is no such node, in step S414, the position P _x (xy) based on the dead reckoning calculated in step S14 is set as the display position P _D (x, y) of the display 14.

As described above, navigation control during traveling on a stored road, traveling on a road outside of storage, arriving at a node, or traveling on a corner can be realized.

<Modification of Candidate Node Narrowing> In the above embodiment, in the node search, the weighting function W used to calculate the evaluation function T for excluding the candidate node has the characteristics shown in FIG. Direction difference θ
(The traveling direction D _P and the connection direction D _C ) have a larger value. This is significantly different nodes with the vehicle traveling direction D _P as the arrival node is standing on the idea that it should be excluded from the earlier candidate nodes. This property is
In the above embodiment, the same characteristics are given to all roads. However, when the vehicle is traveling on a memory road (step S108) and while traveling on a non-memory road (step S196), the evaluation function T
The threshold value compared with is only different between K and K ′ (K ′ <K) in order to select nodes more rigorously while traveling on an unremembered road. Thus, some modified examples of narrowing down candidate nodes will be described. These modifications change the manner of narrowing down according to the road type.

Weight function W In this modification, the characteristics of the weight function W are determined by the type of road (for example, the type of expressway, main road, or general road).
As shown in FIG. 13, it is changed according to the direction difference θ. FIG. 15 illustrates a specific example in which the weight is changed by the direction difference theta (= difference between connecting direction D _C of the traveling direction and the node of the vehicle). This is done for the following reason.

That is, for example, it is assumed that the vehicle runs on a highway. In this case, there is little chance of leaving the road except for entering the interchange and the service area. Therefore, even when the direction difference θ is large while driving on a highway, such a node may not be excluded from the candidates and may be pulled into the node “forcibly” (“matching”). From this point of view, as shown in FIG. 13, the higher the “linearity” is, the smaller the weighting function W is, so that the influence of the direction difference θ is less likely to appear in the evaluation function T.

The weighting function W is stored in the storage device 13 together with the node information and the like in a configuration as shown in FIG. Then, step S106 (FIG. 6A) or step S194 (FIG. 6C)
, The weight function W belonging to the final arrival node _NL is used.

Change of thresholds K, K 'For the same reason as in the case of the weighting function W, the value of threshold K is changed according to the type of road, and the value of threshold K is increased for roads with higher "linearity", and Make the probability of being excluded from a node relatively low.

Change of Map Matching Frequency For the same reason as in the case of the weight function W, the frequency of map matching may be changed depending on the type of road.
In other words, the higher the “straightness”, the lower the frequency of map matching. If the frequency of map matching is reduced, the error due to dead reckoning increases.
However, on a road with a high “straightness”, even if the node is to be excluded from the candidate nodes in the normal control due to an increase in the error, the candidate node is left, and so-called “forced” matching is performed. This is because it does not matter.

It should be noted that the above-described modified example of the candidate node narrowing in which the mode of narrowing is changed according to the road type can be naturally applied to a conventional map matching that does not have a concept of a node.

<Effects of Embodiment> The following effects can be obtained by the embodiment described above.

: Estimated position by dead reckoning by providing nodes not only at characteristic points on the map but also periodically (not necessarily at equal intervals) at predetermined distance intervals on the road Is performed before the error of becomes too large, that is, so that even if the matching is performed, the wrong node is not matched. In addition, by executing the map matching, the error is canceled, and the accuracy of the dead reckoning navigation and the accuracy of the map matching are maintained.

: In the search for a movable node (= candidate node) traveling on an unremembered road, the node search range is expanded according to the distance traveled while performing dead reckoning (step S152), so dead reckoning itself is performed. Unavoidably includes an error, the leakage of the target node from that range is reduced.

: In narrowing down nodes while traveling on the storage road and in narrowing down target nodes while traveling outside the storage road, an evaluation function is performed after weighting according to the direction difference θ.
Since the calculation of T _n is performed, the precision of the refinement is improved.

: Since the weighting in the node narrowing down while traveling on the storage road and the weighting in narrowing down the target node while traveling outside the storage road are set so that the narrowing down in the latter is strict, the value is set outside the storage road. It is possible to compensate for a deterioration in accuracy due to the necessity of relying on dead reckoning navigation during traveling.

: Distance from the vehicle position to the target node when narrowing down the target node while traveling outside the storage road (I _xk )
Not only the direction progression (direction D _P and node connection direction D _C )
Since the target node is specified in consideration of the above, the accuracy of specifying the target node is improved.

: While traveling on a road outside the storage road, by setting the virtual node many times even if there are no candidate nodes, accurate navigation control is possible even when returning to the middle of the storage road.

: In the conventional map matching, while driving on a newly laid road, the memory road is "forcibly" matched to the memory road, and subsequent navigation becomes impossible, but as in this embodiment, By recognizing that the vehicle is traveling on the road that is not stored, incorrect map matching is not performed, so that the navigation accuracy is relatively improved.

[Effects of the Invention] As described above, according to the navigation device of the present invention, the determination and evaluation of whether or not to perform the map matching is performed.
Since the weight is changed according to the weight value set according to the type of the road, the evaluation for the road on which the vehicle is traveling is performed, thereby reducing the load for the evaluation. In particular, both reduction of the burden and maintenance of the evaluation accuracy are achieved at the same time.

Further, by changing the threshold value in accordance with the type of road, it is possible to easily obtain the mode of change in the map matching pattern.

In particular, when the type of road is estimated to be a straight road,
By reducing the frequency of executing map matching, useless processing can be reduced.

[Brief description of the drawings]

FIG. 1 is a diagram for explaining the configuration of the present invention, FIG. 2 is a diagram showing the configuration of a navigation device according to an embodiment, and FIGS. 3A and 3B are diagrams showing the relationship between notes and roads and the connection relationship between nodes. FIG. 4 is a diagram for explaining a storage state in a RAM such as a flag used for control. FIG. 5 is a flowchart of a main routine of a program according to the embodiment control. FIG. 6A is a storage. 6B and 6C are flowcharts of a subroutine relating to node search control while traveling on a road outside the memory, FIG. 6D are flowcharts relating to corner detection control, FIG. 6E is a flowchart relating to map matching control, FIG. 7 is a diagram for explaining an outline of control operations from a storage road to a road outside the storage, and FIG. 8 is a node narrowing control during traveling on a storage road. FIG. 9 is a diagram illustrating an outline of the operation, FIG. 9 is a diagram illustrating one characteristic of the weighting function W, FIG. 10 is a diagram illustrating an outline of a node narrowing-down control operation while the vehicle is traveling on a non-memory road, and FIG. To (e) are diagrams illustrating the outline of the control operation when returning from a road outside the storage to a storage road that is not a node, FIG. 12 is a diagram illustrating the principle of corner detection, and FIG. 13 is a modification of the weight function. FIG. 14 is a diagram illustrating characteristics according to an example, FIG. 14 is a diagram illustrating a data storage configuration according to a modification, and FIG. 15 is a diagram illustrating the principle of a weight function. In the figure, 10 ... navigation device, 11 ... geomagnetic sensor, 12 ...
... vehicle speed sensor, 13 ... map information storage device, 14 ... display unit, 15 ... arithmetic circuit, 17 ... storage device (RAM).

Continuation of the front page (56) References JP-A-1-277713 (JP, A) JP-A-1-212309 (JP, A) JP-A-1-114712 (JP, A) JP-A 64-80813 (JP) JP-A-61-250661 (JP, A) JP-A-1-276014 (JP, A) JP-A-1-277714 (JP, A)

Claims (3)

(57) [Claims] An inference means for estimating a current position by dead reckoning navigation based on a traveling direction and a traveling distance of a vehicle, and information on roads on a map, and different weights for map matching depending on the type of roads. Storage means for storing a plurality of types of matching patterns including assignments in correspondence with the type of the road; determining means for determining the type of the road currently traveling from the current position obtained by the estimating means; Reading a matching pattern corresponding to the road type from the storage unit based on the determination result by the unit;
Map matching means for performing map matching according to the matching pattern and matching a current position obtained by the estimating means with a road on a map, wherein the map matching means relates to a road estimated to be currently running. Evaluation means for determining whether or not map matching should be performed by evaluating the traveling direction of the vehicle. The evaluation means weights the evaluation of the traveling direction with respect to the road determined by the determination means. A navigation device for a vehicle, wherein the navigation device changes according to the type and the weight value read from the storage means. 2. The method according to claim 1, wherein the matching pattern is a threshold for map matching evaluation, and includes a different threshold depending on a type of road, and the evaluation means determines whether or not to perform map matching. 2. The vehicular navigation device according to claim 1, wherein the value is changed according to a threshold value corresponding to the type of road determined by the means. 3. The map matching device according to claim 1, wherein the map matching means reduces the frequency of executing the map matching when the road estimated to be currently running is estimated to be a straight road. Vehicle navigation device.