JP5402968B2 - Vehicular road shape recognition method and apparatus, and recording medium - Google Patents

Description

  The present invention relates to a vehicle road shape recognition method and apparatus, and a recording medium.

  Conventionally, in order to determine the object to be controlled, based on the position (distance, lateral position) of the forward target detected by the sensor, the turning state of the own vehicle obtained based on the steering angle and the yaw rate, and the own vehicle speed. There is known a method of calculating an own lane probability instantaneous value that is a probability that a recognition target exists on a travel path from a curve radius of the vehicle travel path. Then, a predetermined filter process is performed using the corrected instantaneous value of the own lane probability to calculate the own lane probability, and a preceding vehicle is selected based on the own lane probability.

  However, there is a difference between the shape of the road on which the actual control object is traveling and the turning state of the host vehicle. Therefore, in Patent Document 1, for example, by recognizing the road shape, the curve radius R of the turning state of the own vehicle is corrected, the instantaneous value of the own lane probability is corrected based on the road shape, and the corrected own lane probability is corrected. Using the instantaneous value, a predetermined filtering process or the like is performed to calculate the own lane probability, and the preceding vehicle is selected based on the own lane probability.

Japanese Patent No. 3417375

  However, since the above-mentioned conventional technology estimates the roadside end according to the road shape, such as IC, rampway, uphill lane, express bus stop, etc., when the vehicle does not run along the shape of the road There is a possibility that correct estimation is not performed.

  For example, as shown in FIG. 11, when a right-curved road is connected to a left-curved road on which the host vehicle is traveling, a curve radius (curvature radius) R calculated on the host vehicle side is When R calculated from the road shape recognition is corrected, the corrected R is different from the shape of the road on which the vehicle travels. Therefore, even if the road shape can be accurately recognized on the own vehicle side, the correction is disadvantageous, and accurate road shape recognition may not be performed in the demerit scene as shown in FIG.

  The present invention has been made in view of the above points, and it is an object of the present invention to provide a vehicle road shape recognition method and apparatus, and a recording medium, which can calculate a more accurate road shape frequently in a demerit scene.

  In order to achieve the above object, the invention according to claim 1 recognizes the road shape ahead of the vehicle based on the distance to the object and the angle in the vehicle width direction, and based on the recognized road shape and the degree of recognition thereof. The possibility that the object is on the same lane as the own vehicle is determined, and a correction value for correcting the own lane probability is calculated based on the determination result. Then, it is determined whether or not there is a divergence between the curvature of the own vehicle traveling path and the curvature of the recognized road shape, and if there is no divergence, the own lane probability is corrected with a correction value. The own lane probability is not corrected by a correction value. Thereby, the own lane probability according to the situation where there is a large difference between the estimated R and the road shape R can be obtained. That is, it can be avoided that the correction of the own lane probability is a disadvantage, and a more accurate road shape can be calculated frequently.

  The invention according to claim 2 is an example of an apparatus for realizing the vehicular road shape recognition method according to claim 1, and also in the vehicular road shape recognition apparatus, It is determined whether or not there is a deviation in the curvature of the recognized road shape. If it is determined that there is no deviation, the own lane probability calculated by the own lane probability calculating means is calculated by the correction value calculating means. While correcting with the correction value, when it is determined that there is a deviation, the own lane probability calculated by the own lane probability calculating means is not corrected with the correction value calculated by the correction value calculating means. Yes. As a result, similarly to the first aspect, the road shape can be recognized more accurately.

  Then, as in the invention according to claim 3, the curve curvature calculating means, the object recognizing means, the own lane probability calculating means, the road shape recognizing means, the lane identity determining means, and the correction value calculating means of the vehicle road shape recognizing device are provided. The function realized by the computer system can be provided as a program activated on the computer system side, for example. In the case of such a program, for example, it is recorded on a computer-readable recording medium such as a magneto-optical disk, a CD-ROM, a hard disk, a flash memory, etc., and is used by being loaded into a computer system and started as required. it can. In addition, the ROM or backup RAM may be recorded as a computer-readable recording medium, and the ROM or backup RAM may be incorporated into a computer system and used.

  In addition, the code | symbol in the bracket | parenthesis of each means described in this column and the claim shows the correspondence with the specific means as described in embodiment mentioned later.

It is a block diagram which shows the structure of the vehicle control apparatus with which this invention was applied. It is explanatory drawing which shows the outline | summary of a preceding vehicle selection process. It is explanatory drawing at the time of converting each target position into the position at the time of straight road running. It is explanatory drawing of the own lane probability map. (A) is explanatory drawing of a prediction X-axis intersection, (b) is explanatory drawing of road edge recognition. (A) is explanatory drawing of the determination method when the road edge can be recognized farther than the target, and (b) is an explanatory diagram of the determination method when the road edge can be recognized only up to a short distance from the target. It is. (A) is explanatory drawing of the determination method in the case of using the distance of the position of each target used for road edge recognition, and the own vehicle-target curve, (b) is explanatory drawing of the area | region of the road edge vicinity. is there. (A) is the figure which showed the branching points, such as IC exit of a highway, (b) is the figure which showed the time of a lane change. It is the figure which showed the relationship between the information of a segment, and its conditions as a table. It is explanatory drawing of the map of parameter (alpha) for calculating | requiring the own lane probability. It is a figure for demonstrating a subject.

  Hereinafter, a vehicle control device 1 to which the present invention is applied will be described with reference to the drawings. This vehicle control device 1 is mounted on an automobile and outputs a warning when an obstacle exists in a predetermined area in a predetermined situation, or controls the vehicle speed according to the preceding vehicle (preceding vehicle). It is.

  FIG. 1 is a block diagram of the system. The vehicle control device 1 is configured around a computer 3. The computer 3 mainly includes a microcomputer and includes an input / output interface (I / O) and various drive circuits and detection circuits. Since these hardware configurations are general, detailed description thereof is omitted.

  The computer 3 inputs predetermined detection data from a distance / angle measuring device 5, a vehicle speed sensor 7, a brake switch 9, and a throttle opening sensor 11 as an obstacle detection device for a vehicle. The computer 3 outputs predetermined drive signals to the alarm sound generator 13, the distance indicator 15, the sensor abnormality indicator 17, the brake driver 19, the throttle driver 21, and the automatic transmission controller 23.

  The computer 3 further includes an alarm volume setting unit 24 for setting an alarm volume, an alarm sensitivity setting unit 25 for setting sensitivity in an alarm determination process to be described later, a cruise control switch 26, and a steering sensor 27 for detecting an operation amount of a steering wheel (not shown). And a yaw rate sensor 28. Further, the computer 3 includes a power switch 29, and starts predetermined processing when the power switch 29 is turned on.

  Here, the distance / angle measuring device 5 includes a transmission / reception unit 5a and a distance / angle calculation unit 5b. From the transmission / reception unit 5a, laser light is transmitted forward of the vehicle around a predetermined optical axis (center axis). Based on the time until the reflected light is detected by the distance / angle calculation unit 5b, the distance to the object ahead is detected based on the discontinuous sweep irradiation (scanning) in the predetermined angle range and output. It is a device that detects the distance r. In addition to those using laser light, radio waves such as millimeter waves, ultrasonic waves, or the like may be used, and the scanning method is not limited to scanning the transmission unit. For example, the reception unit It may be one that scans.

  With this configuration, the computer 3 performs an alarm determination process for alarming when an obstacle exists in a predetermined alarm area for a predetermined time. Examples of the obstacle include a front vehicle traveling in front of the host vehicle, a front vehicle that is stopped, or an object on the roadside (a guard rail, a pillar object, or the like). In addition, the computer 3 outputs a drive signal to the brake driver 19, the throttle driver 21, and the automatic transmission controller 23, thereby simultaneously performing so-called inter-vehicle control that controls the vehicle speed according to the situation of the preceding vehicle. ing.

  Next, the internal configuration of the computer 3 will be described as a control block. The data of the distance r and the scan angle θ output from the distance / angle calculation unit 5b of the distance / angle measuring device 5 is sent to the coordinate conversion block 41 between polar coordinates and orthogonal coordinates, and the laser radar center is set to the origin (0, 0), the vehicle width direction is converted to XZ orthogonal coordinates with the X axis as the vehicle width direction and the Z axis as the vehicle front direction, and then output to the object recognition block 43 and the road shape recognition block 45.

  The object recognition block 43 obtains the center position (X, Z) and size (W, D) of the object based on the measurement data converted into the orthogonal coordinates, and changes the center position (X, Z) with time. Based on the above, the relative speed (Vx, Vz) of an obstacle such as a front vehicle with the vehicle position as a reference is obtained. Further, in the object recognition block 43, the object is a stop object from the vehicle speed (own vehicle speed) V output from the vehicle speed calculation block 47 based on the detection value of the vehicle speed sensor 7 and the above obtained relative speed (Vx, Vz). A recognition type as to whether the vehicle is a moving object is obtained, and an object that affects the traveling of the host vehicle is selected based on the recognition type and the center position of the object, and the distance indicator 15 displays the distance. Note that (W, D) indicating the size of the object is (width, depth), respectively. An object model having such data is called a “target model”.

  The sensor abnormality detection block 44 detects whether or not the data obtained by the object recognition block 43 is an abnormal range value. If the data is an abnormal range value, the sensor abnormality display unit 17 displays that fact. Is made. On the other hand, in the road shape recognition block 45, the road shape is recognized based on the measurement data converted into the orthogonal coordinates and the data obtained in the object recognition block 43. Details of this road shape recognition processing will be described later. The data obtained in the road shape recognition block 45 is output to the preceding vehicle determination block 53.

  Further, the steering angle is calculated by the steering angle calculation block 49 based on the signal from the steering sensor 27, and the yaw rate is calculated by the yaw rate calculation block 51 based on the signal from the yaw rate sensor 28. In a curve radius (curvature radius) calculation block 63, a curve radius (curvature radius) R is calculated based on the vehicle speed from the vehicle speed calculation block 47, the steering angle from the steering angle calculation block 49, and the yaw rate from the yaw rate calculation block 51. To do. In the preceding vehicle determination block 53, the curve radius R and the recognition type obtained in the object recognition block 43, the center position coordinates (X, Z), the object size (W, D), and the relative speed (Vx, Vz). And the preceding vehicle is selected based on the road shape data obtained in the road shape recognition block 45, and the distance Z and relative speed Vz for the preceding vehicle are obtained.

  Then, the inter-vehicle distance control unit and the alarm determination unit block 55 include the distance Z to the preceding vehicle, the relative speed Vz, the host vehicle speed Vn, the preceding vehicle acceleration, the object center position, the object width, the recognition type, and the setting state of the cruise control switch 26. Based on the depression state of the brake switch 9, the opening from the throttle opening sensor 11, and the sensitivity setting value by the alarm sensitivity setting unit 25, it is determined whether or not to issue an alarm if it is an alarm determination, and vehicle speed control if it is a cruise determination. Determine the contents. As a result, if an alarm is required, an alarm generation signal is output to the alarm sound generator 13. If the cruise determination is made, a control signal is output to the automatic transmission controller 23, the brake driver 19 and the throttle driver 21 to perform necessary control. When these controls are executed, necessary display signals are output to the distance indicator 15 to notify the driver of the situation.

  Next, the operation | movement concerning recognition of the road shape performed in the vehicle control apparatus 1 comprised as mentioned above is demonstrated according to the flowchart of FIG. In S1000, which is the first step in FIG. 2, distance / angle measurement data is read. This processing is executed by the distance / angle measuring device 5, but the distance / angle measurement data for one scan is taken in. The scan cycle is 100 msec, and data is taken every 100 msec.

  In subsequent S2000, the polar / orthogonal coordinate conversion block 41 converts the distance / angle measurement data from the polar coordinate system to the XZ orthogonal coordinate system, and the object recognition block 43 performs object recognition based on the converted data. The contents of this object recognition are as described above. The recognized object is called a target or a target model.

  In S3000, based on the yaw rate obtained from the yaw rate sensor 28 or the steering angle obtained from the steering sensor 27, an estimated R (curve radius of the own vehicle traveling curve) is calculated. Here, the estimated R is calculated from the steering angle. That is, the estimated R is obtained from estimated R = constant / steering angle.

  Here, the “constant” is a constant depending on the vehicle speed and the vehicle type, and is stored in advance in the curve radius (curvature radius) calculation block 63 in the computer 3 as a map function for each vehicle type for each vehicle type. Yes. Since this function C is generally known as a function for obtaining the curve radius from the steering angle θ, detailed description thereof is omitted. A method for obtaining the estimated R from the yaw rate Ω can be calculated by dividing the vehicle speed V by the yaw rate Ω.

  In S4000, the own lane probability instantaneous value of the target recognized in S2000 is calculated. The own lane probability is a parameter representing the probability that the target is a vehicle traveling on the same lane as the own vehicle. The own lane probability instantaneous value is a value calculated based on detection data at that moment.

  First, the positions of all the targets obtained in the object recognition process (S2000) are converted into positions when traveling on a straight road. When the center position of the original target is (Xo, Zo) and the width in the X-axis direction is Wo, the straight road conversion position (X, Z, W) is obtained by the following conversion formula (see FIG. 3). .

X <-Xo-Zo 2 / 2R [Formula 1]
Z ← Zo [Formula 2]
W ← Wo ... [Formula 3]
R: Prediction R Right curve: Sign positive left curve: Sign negative Note that “^” in Equation 1 means that the numerical value before “^” is raised to the power of the numerical value after “^”. . The same applies to other parts of the specification. Here, the equation of the circle is approximated under the assumption of | X | << | R |, Z. Further, when the distance / angle measuring device 5 is attached at a position away from the vehicle center, the X coordinate is corrected so that the vehicle center becomes the origin. In other words, here, only the X coordinate is substantially converted.

  The center position (X, Z) obtained by converting to a straight path in this way is arranged on the own lane probability map shown in FIG. 4 and the instantaneous own lane probability of each object, that is, the own lane at that time. Find the probability of being in The probability exists because there is an error between the curve curvature radius R obtained from the steering angle and the actual curve curvature radius. Find the instantaneous lane probability.

  In FIG. 4, the horizontal axis is the X axis, that is, the left-right direction of the own vehicle, and the vertical axis indicates the Z axis, that is, the front of the own vehicle. In the present embodiment, a region up to 5 m left and right and 100 m ahead is shown. Here, the areas are area a (own lane probability 80%), area b (own lane probability 60%), area c (own lane probability 30%), area d (own lane probability 100%), and other areas ( The lane probability is 0%. The setting of this area is determined by actual measurement. In particular, the area d is an area set by considering an interruption immediately before the host vehicle.

  The boundary lines La, Lb, Lc, and Ld that divide the regions a, b, c, and d are given by, for example, the following equations 4 to 7. Note that the boundary lines La ′, Lb ′, Lc ′, and Ld ′ are symmetric with respect to the boundary lines La, Lb, Lc, and Ld on the Y axis.

La: X = 0.7 + (1.75−0.7) · (Z / 100) ^ 2 [Formula 4]
Lb: X = 0.7 + (3.5−0.7) · (Z / 100) ^ 2 [Formula 5]
Lc: X = 1.0 + (5.0−1.0) · (Z / 100) ^ 2 [Formula 6]
Ld: X = 1.5 · (1-Z / 60) [Formula 7]
When this is expressed by a general formula, the following formulas 8 to 11 are obtained.

La: X = A1 + B1 · (Z / C1) ^ 2 [Equation 8]
Lb: X = A2 + B2 · (Z / C2) ^ 2 [Formula 9]
Lc: X = A3 + B3 · (Z / C3) ^ 2 [Formula 10]
Ld: X = A4 · (B4-Z / C4) [Formula 11]
In general, the areas are set so as to satisfy the following expressions 12 to 14 from these expressions 8 to 11. The actual value is determined by experiment.

A1 ≦ A2 ≦ A3 <A4 [Formula 12]
B1 ≦ B2 ≦ B3 and B4 = 1 [Equation 13]
C1 = C2 = C3 (C4 has no restriction) [Equation 14]
Note that the boundary lines La, Lb, Lc, La ′, Lb ′, and Lc ′ in FIG. 4 are parabolas from the viewpoint of calculation processing speed, but if the processing speed permits, it is better to represent them by arcs. The boundary lines Ld and Ld ′ are also better represented by a parabola or arc bulging outward if the processing speed permits.

  Next, the straight road conversion position of each target is collated with the own lane probability map of FIG. By comparing with the map in the following manner, the own lane probability instantaneous value P0 is obtained.

Object having even a small area d → P0 = 100%
Object whose center exists in region a → P0 = 80%
Object whose center exists in region b → P0 = 60%
Object whose center exists in region c → P0 = 30%
Object not satisfying all of the above → P0 = 0%
In S5000, the road shape is recognized based on target data that seems to be a delineator provided on the roadside.

  First, a target whose recognition type whose width W is less than 1 m is a stopped object is extracted. Thereby, most of the vehicle, the guide sign, the signboard, etc. can be removed. For each extracted stop target, a point that intersects the X axis is predicted. In calculating the predicted X-axis intersection, a circle passing through the center of the target and having a relative velocity vector as a tangent vector is obtained. If it is assumed that the center of the circle is on the X axis, the circle R is orthogonal to the X axis, and the radius R is uniquely determined. Actually, the following approximate calculation is performed.

Under the assumption of | X | << | R |, Z, when a circle is parabolically approximated, the equation of the circle that passes through the center of the target and is orthogonal to the X axis is
X = Xo + (Z-Zo) ^ 2 / 2R [Formula 15]
It becomes. Further, since the relative velocity vector of the target is a tangent vector of the circle, Equation 15 is
dX / dZ = Vx / Vz [Formula 16]
It can be expressed as. From these two formulas, the radius R is
R = (Z-Zo) .Vz / Vx
(See FIG. 5A). And when Z = 0,
X = Xo−Zo · Vx / 2Vz
Therefore, the predicted X-axis intersection is determined as follows.
Predicted X-axis intersection = Xo−Zo · Vx / 2Vz
When the predicted X-axis intersections of all the stop targets can be calculated in this way, the sign is divided into negative and positive, and the next statistical process is performed. First, the predicted X-axis intersections of all stop targets are simply averaged and used as a temporary average value. Next, all data that is 2 m or more away from the temporary average is eliminated, and the remaining data is averaged again. The data excluded here is not used for road shape recognition.

  The reason for performing such processing is as follows. In addition to the delineator, for example, when an upper signboard cannot be removed and is mixed, an incorrect road shape is recognized. Therefore, by performing such an averaging process, it is possible to exclude those that deviate greatly from the position where the delineator should be present, so that the road shape can be recognized with high accuracy.

  And as shown in FIG.5 (b), the road edge is recognized by interpolating and connecting the remaining stop target about each of the right and left of a road. Here, as for the intersection of the road edge and the X axis, the one at the closest distance (small Z) is selected from each of the left and right edges of the road, and the predicted X axis intersection of the target is the road edge. Used as an intersection with the X axis. The recognized road edge is set in the road edge coordinate table. The road end coordinate table has a left end and a right end for the road, and the X coordinate value of the road end is stored for each distance of 5 m. The distance range is 0m to 150m. The distance of the target constituting the road edge is rounded to a unit of 5 m by dropping the fraction, and data is set in the corresponding table. Tables with no corresponding data are left empty.

  By the above road shape recognition, a road shape R different from the estimated R obtained in S3000 is obtained.

  In S6000, it is determined whether each target is a vehicle on the same lane as the own vehicle based on the road shape recognized in S5000, and “correction value of own lane probability instantaneous value” is calculated according to the result. First, for each target, a basic determination is made as to whether the vehicle is on the same lane as the subject vehicle. Basic judgment consists of the following three.

[Basic judgment 1] This judgment is a judgment when the road edge can be recognized farther than the target, and is judged on the left side and the right side of the road, respectively.
(A) On the road left side figure 6 (a),
Z_MAX ≧ Zo and | ΔXZ = Zo−ΔXZ = 0 | <1.2
When m ◆ Basic judgment 1 (L) result ← 1
Z_MAX ≧ Zo and | ΔXZ = Zo−ΔXZ = 0 | ≧ 2.0
When m ◆ Basic judgment 1 (L) result ← -1
Otherwise ◆ Basic judgment 1 (L) result ← 0
(B) Same as the left side of the road.
Z_MAX ≧ Zo and | ΔXZ = Zo−ΔXZ = 0 | <1.2
When m ◆ Basic judgment 1 (R) result ← 1
Z_MAX ≧ Zo and | ΔXZ = Zo−ΔXZ = 0 | ≧ 2.0
When m ◆ Basic judgment 1 (R) result ← -1
Otherwise ◆ Basic judgment 1 (R) result ← 0
When the result of the basic determination 1 is “1”, it can be almost certainly determined as a preceding vehicle on the own lane, and when it is “−1”, it can be almost certainly determined as a vehicle or roadside object in another lane. . Further, “0” is a case where it is difficult to determine any one or the road edge cannot be recognized.

[Basic determination 2] This determination is performed when the road edge cannot be recognized up to the target position, and is determined on the left and right sides of the road.
(A) On the road left side figure 6 (b),
| ΔXZ = Z # MAX−ΔXZ = 0 | <1.2 m · (Z # MAX / Zo) ^ 2
Or | ΔXZ = Z # MAX−ΔXZ = 0 | <0.3 m ◆ Basic judgment 2 (L) result ← 1
| ΔXZ = Z # MAX−ΔXZ = 0 | ≧ 2.0 m · (Z # MAX / Zo) ^ 2
And | ΔXZ = Z # MAX−ΔXZ = 0 | ≧ 0.3 m ◆ Basic judgment 2 (L) result ← −1
Otherwise ◆ Basic judgment 2 (L) result ← 0
Here, when Z_MAX> Zo / 2 ◇ Basic judgment 2 (L) reliability ← 1 (high)
When Z_MAX ≦ Zo / 2 ◇ Basic judgment 2 (L) Reliability ← −1 (Low)
(B) The basic determination 2 (R) result and the basic determination 2 (R) reliability are calculated by the same method as that on the right side of the road.

In addition, the own vehicle-target curve in FIG. 6B is a curve obtained by connecting the target and the origin with an arc perpendicular to the X axis. The equation of the circle is assumed to be parabolic approximated by the following equation under the assumption of | X | << | R |, Z.
X = Z ^ 2 / R (R: radius)
As can be seen from FIG. 6B, the distance (ΔXZ = Z # MAX) from the recognized farthest point (distance Z_MAX) to the vehicle-target curve in the direction parallel to the X axis is used for the determination. Yes. Due to the parabolic approximation, (Z # MAX / Zo) ^ 2 is multiplied to the determination values 1.2 m and 2.0 m in the case of the basic determination 1 described above.

  When the result of such basic determination 2 is “1”, it can be determined that the possibility of a preceding vehicle on the own lane is high, and when it is “−1”, there is a possibility of a vehicle or roadside object in another lane. It can be judged that it is expensive. Further, the reliability of the determination is expressed in two stages by the basic determination 2 reliability. When the determination result or the determination reliability is “0”, it is difficult to make any determination or the road edge cannot be recognized.

[Basic determination 3] This determination is based on a distance other than Z = Zo and Z_MAX, and is determined on the left and right sides of the road.
(A) Left side of road The following two types of determination are performed.

  [Decision 3a] In FIG. 7A, for all positive numbers i satisfying i · dZ ≦ Zo (dZ = 5m), Z_MAX → i · dZ, the same determination as the basic decision 2 (L) result = 1 To implement.

When the condition of basic judgment 2 (L) = 1 is satisfied for all i,
◆ Basic judgment 3a (L) result ← 1
When there is one or more i that does not satisfy the condition,
◆ Basic judgment 3a (L) result ← -1
When there is no i to be judged ◆ Basic judgment 3a (L) result ← 0
[Decision 3b] In FIG. 7A, for all positive numbers i satisfying i · dZ ≦ Zo (dZ = 5 m), Z_MAX → i · dZ, the same as the basic decision 2 (L) result = −1. Make a decision.

When the condition of basic determination 2 (L) = − 1 is satisfied for all i,
◆ Basic judgment 3b (L) result ← 1
When there is one or more i that does not satisfy the condition,
◆ Basic judgment 3b (L) result ← -1
When there is no i to be judged ◆ Basic judgment 3b (L) result ← 0
(B) The basic determination 3a (R) result and the basic determination 3b (R) result are calculated by the same method as the road right side road left side.

  When the result of the basic determination 3a is “1”, it can be determined that the vehicle is the leading vehicle on the own lane regardless of the distance of the road edge data. When the result is “−1”, depending on the distance, This is a case where it cannot be determined that the preceding car. Further, when the result of the basic determination 3a is “0”, there is no road edge coordinate data at a shorter distance than the target.

  On the other hand, when the result of the basic determination 3b is “1”, it can be determined that the vehicle or roadside object is in another lane regardless of the road edge data at any distance. When “−1”, depending on the distance, it cannot be determined as a vehicle or roadside object in another lane. Further, when the result of the basic determination 3b is “0”, there is no road edge coordinate data at a shorter distance than the target.

Based on the results of the above three basic determinations, the correction value of the own lane probability instantaneous value is calculated according to the following six types of classification. When a plurality of conditions are satisfied, an instantaneous value having a high priority is adopted.
[First classification] When the road edge can be recognized farther than the target and it can be judged as a preceding vehicle on its own lane,
If basic determination 1 (L) result = 1 and basic determination 3a (L) = 1, own lane probability instantaneous value correction value ← 40% Priority: 5
If basic judgment 1 (L) result = 1 and basic judgment 3a (L) = − 1, own lane probability instantaneous value correction value ← 0% Priority: 3
If basic determination 1 (L) result = 1 and basic determination 3a (L) = 0, the own lane probability instantaneous value correction value ← 40% Priority: 2
For the right side of the road, the correction value is calculated in the same manner as the left side of the road.

[Second classification] When the road edge can be recognized farther than the target and can be judged as a vehicle or roadside object in another lane,
When basic determination 1 (L) result = −1, if basic determination 3a (L) = 1, own lane probability instantaneous value correction value ← −40% Priority: 5
When basic judgment 1 (L) result = -1, if basic judgment 3a (L) =-1, own lane probability instantaneous value correction value ← 0% Priority: 3
When basic determination 1 (L) result = −1, if basic determination 3a (L) = 0, the own lane probability instantaneous value correction value ← −40% Priority: 2
For the right side of the road, the correction value is calculated in the same manner as the left side of the road.

[Third classification] When the road edge is not recognized up to the target position and it can be judged as the preceding vehicle on the own lane,
If basic decision 2 (L) result = 1 and basic decision 3a (L) = 1, own lane probability instantaneous value correction value ← 40% Priority: 1
If basic determination 2 (L) result = 1 and basic determination 3a (L) = − 1, own lane probability instantaneous value correction value ← 0% Priority: 1
If basic decision 2 (L) result = 1, basic decision 3a (L) = 0, and basic decision 2 (L) reliability = 1, the own lane probability instantaneous value correction value ← 40% Priority: 1
If basic determination 2 (L) result = 1, basic determination 3a (L) = 0, and basic determination 2 (L) reliability = −1, the own lane probability instantaneous value correction value ← 20% Priority: 1
For the right side of the road, the correction value is calculated in the same manner as the left side of the road.

[Fourth classification] When the road edge is not recognized up to the target position and can be judged as a vehicle or roadside object in another lane,
If basic determination 2 (L) result = -1 and basic determination 3a (L) = 1, own lane probability instantaneous value correction value ← −40% Priority: 1
If basic determination 2 (L) result = -1 and basic determination 3a (L) =-1, own lane probability instantaneous value correction value ← 0% Priority: 1
If basic determination 2 (L) result = -1, basic determination 3a (L) = 0, and basic determination 2 (L) reliability = 1, the own lane probability instantaneous value correction value ← −40% Priority: 1
If basic determination 2 (L) result = -1, basic determination 3a (L) = 0, and basic determination 2 (L) reliability = -1, own lane probability instantaneous value correction value ← −20% Priority: 1
For the right side of the road, the correction value is calculated in the same manner as the left side of the road.

[Fifth classification] When neither the left end of the road nor the right end of the road is recognized and the conditions of the first to fourth classifications are not satisfied, the own lane probability instantaneous value correction value ← 0% Priority: 0
[Sixth Classification] The area shown in the target map 7 (b) determined to be a roadside object on the road edge, that is, an object centered in a 0.5m area from the recognized road edge (left edge and right edge) to the left and right respectively. The mark is determined to be a roadside object, and the following correction value and priority are set.
Self-lane probability instantaneous value correction value ← -70% Priority: 6
As described above, the correction value (Ph) of the own lane probability instantaneous value corresponding to the recognized road shape is calculated, and the calculation result is roughly as follows.

  When the road shape can only recognize a short distance, the correction value is made smaller (as compared to the case where the road shape can be recognized far) (refer to the contents including the result of the basic determination 2). When only a short distance can be recognized, the determination is made assuming a circle (arc) connecting the object and the vehicle position. Therefore, in view of the existence of such an estimation element, it is preferable to relatively reduce the correction value.

  If it is determined that the roadside object is on the road edge, the correction value is increased (see the sixth classification). In addition, since it was determined to be a roadside object on the road edge, there is a high possibility that it exists outside the own lane, so a large negative correction value of −70% was set. For this reason, even if the own lane probability instantaneous value is high, the probability can be lowered by correction, and erroneous selection can be prevented. For example, such a countermeasure is particularly effective when “the vehicle is running on a straight line and the front is already in a curve”.

  The priority is used as follows. In the present embodiment, the determination can be made based on any of the left and right road edges. However, the recognition degree with respect to the left and right ends is different, and there may be a case where there is a difference in the possibility of being on the same lane as the reference. Therefore, in this case, the correction value is calculated based on the determination result with the higher priority. In the above example, in the first classification and the second classification when the road shape can be recognized farther than the object, the priority is 5, 3, and 2, but the road shape is limited to a shorter distance than the object. In the third classification and the fourth classification when they are not recognized, since the priority is 1, the determination result based on the first and second classifications has priority.

  The above classification is performed by combining the result of basic judgment 3 with basic judgment 1 or 2. In this way, since the priority is determined in consideration of the position of each object used when recognizing the road shape, the priority can be determined in consideration of the entire road shape.

  Subsequently, in S6100, it is determined whether or not there is a large difference between the estimated R calculated in S3000 and the road shape R recognized in S5000. This is because if the own lane probability instantaneous value is corrected when the difference between the estimated R and the road shape R is large, the accuracy of the own lane probability instantaneous value may be reduced.

  For example, the estimated R and the road shape R are expected to deviate significantly as shown in FIG. 8A, for example, at a branch point such as an IC exit of an expressway, or at the time of a lane change in FIG. The situation can be considered. In such a situation, if the own lane probability instantaneous value correction value Ph is corrected with respect to the own lane probability instantaneous value, the instantaneous own lane probability may be reduced. Will be recognized, so guard processing is added. The guard process is a process in which correction is not performed when the estimated R and the road shape R are greatly deviated. That is, it is a process in the case where it is determined that “there is a large deviation” in this step.

The specific determination of “is there a big divergence” is performed as follows. First, information on the current segment and its conditions shown in FIG. 9 are prepared in advance as a table. The segment information includes the vehicle speed, estimated R, road shape R, and the like. Each numerical value shown in FIG. 9 is an example, and other numerical values may of course be set.
1) When condition a and condition o are satisfied, judgment is performed in the order of priority i) → iv)
(i) Condition h and Condition d2 and (Condition (1) or Condition (2) or Condition (3)) Condition (1) in which self-lane probability instantaneous value correction value Ph = 0% based on road shape recognition when established (condition) b and condition g) or (condition c and condition f)
Condition (2); (condition d and condition j and condition k) or (condition e and condition i and condition l)
Condition (3); condition m or condition n
ii) When condition t and condition d2 and condition b2 are established, execute A → B below A; when condition v and {condition x or condition u or condition q} is established, the own lane probability instantaneous value correction value Ph = 0% B; if not established, use correction value Ph calculated by road shape recognition
iii) When the condition s and condition w and condition a2 and {condition p or condition r or condition v or condition c2} is satisfied, the self-lane probability instantaneous value correction value Ph = 0% by road shape recognition is set.
iv) Use the own lane probability instantaneous value correction value Ph calculated in S6000 2) If the above 1) is not established, use the own lane probability instantaneous value correction value Ph calculated in S6000. Up to iii) is the case where “there is a large deviation”, and in this case, the process proceeds to step S7000. In this case, the own lane probability instantaneous value correction value Ph = 0%, and the own lane probability instantaneous value is not corrected in the following step S6200.

  On the other hand, 1) iv) and 2) are cases where “there is no significant deviation”. In this case, the process proceeds to S6200.

In the situation as shown in FIG. 8, for example, the conditions a, b, and c shown in FIG.
a) 7000 ≦ | estimated R | and | road shape R | <700
b) | Estimated R | <1000 and 7000 ≦ | Road shape R |
c) | (1 / estimated R × 1000) − (1 / road shape R × 1000) |> 1.5
And when there is any of these conditions a), b) and c), there is a case where “there is a big difference”.

  Returning to the description of FIG. 2, since it is determined in S6100 that “there is no large deviation” in S6100, the own lane probability instantaneous value correction value calculated in S6000 is added to the own lane probability instantaneous value calculated in S4000 for each target. Add Ph. At this time, limit processing is performed with an upper limit of 100% and a lower limit of 0%.

  In S7000, the own lane probability is calculated. In order to calculate the own lane probability, if it is determined in S6100 that “there is no large deviation”, the own lane probability instantaneous value corrected in S6200 is used. On the other hand, if it is determined in S6100 that “there is a large divergence”, the own lane probability instantaneous value calculated (not corrected) in S4000 is used. In other words, this corresponds to adding 0% as the own lane probability instantaneous value correction value Ph to the own lane probability instantaneous value calculated in S4000.

Specifically, filter processing is performed using the following equation. Here, α is a parameter depending on the distance Z, and is obtained using the map of FIG. The initial value of the own lane probability is 0%.
Own lane probability ← Own lane probability previous value × α + Own lane probability instantaneous value × (1-α)
In subsequent S8000, the preceding vehicle is determined. Among targets whose own lane probability calculated in S7000 is 50% or more, the target having the smallest distance Z is determined as the preceding vehicle. In accordance with the distance and relative speed of the target determined to be the preceding vehicle, control is performed so as to keep the distance between the preceding vehicle constant and an alarm is sounded when the preceding vehicle is in danger of collision.

  Thereafter, when road shape recognition is performed again in S5000, a low pass filter is applied to the road shape R based on the own lane probability calculated in S7000, and the estimated R of the vehicle is corrected with the road shape after the filter. Thereby, a more accurate road shape can be recognized.

  As described above, for the own lane probability instantaneous value (S4000) used to calculate the own lane probability (S7000), the estimated R (S3000) calculated by the own vehicle and the road shape R (S5000) calculated from the road shape recognition. )), There is a feature that correction of the instantaneous lane probability instantaneous value (S4000), which is the probability that a recognized target exists on the traveling road, is not performed.

  As a result, as shown in FIGS. 8A and 8B described above, it is possible to obtain the own lane probability corresponding to the demerit scene in which there is a large difference between the estimated R and the road shape R. That is, it can be avoided that the correction of the instantaneous value of the own lane is a disadvantage, and more accurate road shape recognition is possible. In addition, the accuracy of the preceding vehicle selection can be improved. Therefore, a more accurate road shape can be frequently calculated in a situation where the estimated R and the road shape R are expected to greatly deviate.

  Of course, if there is no large difference between the estimated R (S3000) calculated by the own vehicle and the road shape R (S5000) calculated from the road shape recognition, the own lane probability instantaneous value (S4000) is corrected (S6200). This also enables accurate road shape recognition and improves the accuracy of the preceding vehicle selection. In this case, since the magnitude and priority of the correction value are changed to the degree of recognition of the road shape, more appropriate R correction can be performed, and as a result, a more accurate preceding vehicle can be selected.

  In the present embodiment, at least one set of the steering sensor 27 and the steering angle calculation block 49, the yaw rate sensor 28 and the yaw rate calculation block 51 corresponds to the turning detection means, and the curve radius calculation block 63 corresponds to the curve curvature calculation means. To do. The distance / angle measuring device 5 corresponds to radar means, and the polar / orthogonal coordinate conversion block 41 and the object recognition block 43 correspond to object recognition means. The preceding vehicle determination block 53 corresponds to the own lane probability calculation means, the preceding vehicle selection means, the lane identity determination means, and the correction value calculation means, and the road shape recognition block 45 corresponds to the road shape recognition means.

(Other embodiments)
The above-described embodiment is an example, and the present invention is not limited to the above-described embodiment, and other configurations that can realize the contents of the present invention may be employed. For example, although the above embodiment is configured to select the preceding vehicle, a configuration that is purely intended for road shape recognition may be used. Moreover, it can also change like the following (1)-(3).

  (1) In the above embodiment, the determination is made based on the left and right ends of the road, and when the determination results are different, the higher priority is adopted, but only one of the left and right road ends is used as a reference. The determination can also be realized. However, since it is not always possible to recognize the road edge stably, it is preferable to directly recognize the shapes of both the left and right road edges as in the above embodiment.

  (2) In the above embodiment, specific numerical values are given as correction values and priorities, but this is merely an example, and the numerical values can be changed as appropriate.

  (3) Although the distance / angle measuring device 5 using laser light is employed as the “radar means” in the above embodiment, it has already been described that a millimeter wave or the like may be used. For example, when FMCW radar or Doppler radar is used with a millimeter wave, the distance information from the reflected wave (received wave) to the preceding vehicle and the relative speed information of the preceding vehicle can be obtained at one time. The process of calculating the relative speed based on the distance information as in the case of the case becomes unnecessary.

5 Distance / Angle Measuring Device 27 Steering Sensor 28 Yaw Rate Sensor 43 Object Recognition Block 45 Road Shape Recognition Block 49 Steering Angle Calculation Block 51 Yaw Rate Calculation Block 53 Leading Vehicle Determination Block 63 Curve Radius Calculation Block

Claims (3)

The relative position of the recognition target object with respect to the own vehicle is obtained, and based on the curvature of the own vehicle traveling path calculated based on the turning state of the own vehicle and the own vehicle speed and the relative position of the object, the object is In the vehicle road shape recognition method performed by the vehicle road shape recognition device that calculates the own lane probability that is the probability of being on the same lane,
Based on the distance to the object and the angle in the vehicle width direction, recognize the road shape ahead of the vehicle,
Based on the recognized road shape and the degree of recognition, determine the possibility that the object is on the same lane as the vehicle,
Based on the determination result, a correction value for correcting the own lane probability is calculated,
It is determined whether there is a divergence between the curvature of the own vehicle traveling path and the curvature of the recognized road shape. If there is no divergence, the lane probability is corrected with the correction value, but there is a divergence. The vehicle lane probability is not corrected by the correction value. Turn detection means (27, 28, 49, 51) for detecting the turning state of the own vehicle;
Curve curvature calculating means (63) for calculating the curvature of the own vehicle traveling path based on the turning state and the own vehicle speed detected by the turning detection means;
Radar means (5) for irradiating a transmission wave within a predetermined angle range in the vehicle width direction and detecting the distance to the reflecting object and the angle in the vehicle width direction based on the reflected wave;
An object recognition means (43) for obtaining a relative position of the object based on a distance and an angle in the vehicle width direction as a detection result by the radar means;
Based on the curvature of the own vehicle traveling path obtained by the curve curvature calculating means and the relative position of the object calculated by the object recognizing means, the own lane probability that the object is on the same lane as the own vehicle is obtained. A vehicle road shape recognition device comprising: own vehicle lane probability calculation means (53) to be obtained;
The object recognizing means can determine a recognition type as a moving object or a stopped object based on the relative speed of the object and the own vehicle speed,
further,
Road shape recognition means for extracting stop object data effective for recognizing the road shape using the relative position and recognition type of the object obtained by the object recognition means, and recognizing the road shape based on the extracted data (45)
Lane identity determination means (53) for determining the possibility that the object is on the same lane as the vehicle based on the road shape recognized by the road shape recognition means and the degree of recognition thereof;
Correction value calculation means (53) for calculating a correction value for correcting the own lane probability based on a determination result by the lane identity determination means,
The own lane probability calculating means determines whether there is a divergence between the curvature of the own vehicle traveling path and the curvature of the recognized road shape, and when determining that there is no divergence, While correcting with the correction value calculated by the correction value calculating means, when it is determined that there is a deviation, the own lane probability is not corrected with the correction value calculated by the correction value calculating means, A vehicle road shape recognition device. Curve curvature calculation for calculating the curvature of the vehicle traveling path based on the vehicle turning state and the vehicle speed detected by the turning detection means (27, 28, 49, 51) for detecting the vehicle turning state. Means (63) ,
A distance as a detection result by the radar means (5) that irradiates a transmission wave over a predetermined angle range in the vehicle width direction and detects the distance to the reflecting object and the angle in the vehicle width direction based on the reflected wave; An object recognition means (43) for determining a relative position of the object based on the angle in the vehicle width direction and determining a recognition type as a moving object or a stopped object based on the relative speed of the object and the own vehicle speed ;
Road shape recognition means for extracting stop object data effective for recognizing the road shape using the relative position and recognition type of the object obtained by the object recognition means, and recognizing the road shape based on the extracted data (45) ,
Lane identity determination means (53) for determining the possibility that the object is on the same lane as the own vehicle based on the road shape recognized by the road shape recognition means and the degree of recognition thereof .
Correction value calculation means (53) for calculating a correction value for correcting the own lane probability based on the determination result by the lane identity determination means ; and
Based on the curvature of the own vehicle traveling path obtained by the curve curvature calculating means and the relative position of the object calculated by the object recognizing means, the own lane probability that the object is on the same lane as the own vehicle is obtained. And determining whether or not there is a deviation between the curvature of the own vehicle traveling path and the curvature of the recognized road shape, and when it is determined that there is no deviation, the own lane probability is calculated by the correction value calculation means. While correcting with the calculated correction value, if it is determined that there is a divergence , the computer as the own lane probability calculating means (53) that does not correct the own lane probability with the correction value calculated by the correction value calculating means A computer-readable recording medium on which a program for causing the system to function is recorded.

Images