JP4019854B2 - Travel route generator - Google Patents

Description

[0001]
BACKGROUND OF THE INVENTION
The present invention relates to a technical field of a travel route generation device for guiding a vehicle that automatically follows a steering by performing automatic steering along a certain route.
[0002]
[Prior art]
2. Description of the Related Art Conventionally, an automatic steering vehicle that travels following a route by automatically steering a route given by white line recognition or the like is known. However, when these automatic steering vehicles lose the route guidance means, acquisition of reference information for performing automatic steering control stops.
[0003]
For example, Japanese Patent Application Laid-Open No. 10-31799 describes a technique for detecting a distance to a side wall in addition to a CCD camera that detects a white line, and performing steering control along the side wall when a white line cannot be detected. Has been. However, when steering control is performed along the side wall, only the lateral deviation is observed, and it is not possible to know in advance whether the side wall is continuous. Is desired.
[0004]
[Problems to be solved by the invention]
When white line recognition by a CCD camera is used as a route for guiding an automatically steered vehicle as described above, the white line line-of-sight depends on the lighting condition of the road, especially at night, and it is difficult to look far away. When the road surface is wet, the recognition of the white line tends to be interrupted due to the reflection of illumination, and the route may be intermittently lost.
[0005]
The present invention has been made paying attention to the above-mentioned problems, and its purpose is to feature landscape features that can be easily recognized at night or when the road surface is wet when the route for automatic steering is lost. By estimating the route based on the vehicle, automatic steering control can be continued when the state of losing sight of the route continues intermittently, and an allowance time that the driver can handle even when the state of losing sight of the route continues is secured. An object of the present invention is to provide a travel route generation device that can do this.
[0006]
[Means for Solving the Problems]
In order to achieve the above object, in the present invention, an imaging means for imaging a landscape in front of the host vehicle, and an imaging landscape White line inside A specified route recognizing means for recognizing a specified route from the vehicle and an automatic steering control means for controlling the steering angle of the host vehicle based on the recognized default route. In the vehicle travel route generation device to perform, Said In the scenery Placed away from the road surface Landscape vector sequence recognition means for recognizing a vector sequence connecting characteristic points as a landscape vector sequence, and route reproduction data calculation means for calculating a position / direction relationship between the recognized landscape vector sequence and a default route as route reproduction data; A default route recognition judging means for judging whether or not the default route has been recognized, and a position of the default route based on the recognized landscape vector sequence and the calculated route reproduction data when it is judged that the default route is not recognized. A predetermined route estimation means for estimating the posture is provided.
[0007]
【The invention's effect】
Therefore, in the present invention, even if the vehicle that is performing the automatic steering based on the recognized default route loses sight of the automatic steering route, the estimated route is supplied, so that the route is lost. Thus, the automatic steering control can be continued when the vehicle continues intermittently, and a margin time that can be handled by the driver can be ensured even when the state of losing sight of the route continues.
[0008]
DETAILED DESCRIPTION OF THE INVENTION
(First embodiment)
First, the configuration will be described.
FIG. 1 is a configuration diagram illustrating a vehicle to which the travel route generation device of the first embodiment is applied.
[0009]
In the figure, 1 is a CCD camera, 2 is an image processing CPU, 3 is a GPS antenna, 4 is map data, 5 is a gyro sensor, 6 is a wheel speed sensor, 7 is a steering angle sensor, 8 is a vehicle travel information detection device, 9 is a navigation system device, 10 is an automatic steering controller, 11 is a vehicle control CPU, and 12 is a travel route generating device.
[0010]
The CCD camera 1 recognizes a predetermined route ahead through the image processing CPU 2 and calculates a deviation of the yaw angle and position of the vehicle with respect to the default route. Although not particularly shown, normally, these values measured by the image processing CPU 2 are transmitted to the vehicle control CPU 11, and a command is output to the automatic steering controller 10 so as to follow the predetermined route.
[0011]
The predetermined route and the landscape vector are recognized by the landscape vector sequence recognition means 14 in the travel route generation device 12 by a series of processes of the CCD camera 1 and the image processing CPU 2. The default route recognizing means 13 in the travel route generating device 12 may be configured to be supplied by a known device having a different configuration from the CCD camera 1 and the image processing CPU 2.
[0012]
Information acquired by the GPS antenna 3 and the map data 4 is supplied to the route reproduction data calculation means 15 of the travel route generation device 12 through the navigation system device 9 and the shape of the road ahead and the approximate position on the road.
[0013]
The gyro sensor 5, the wheel speed sensor 6, and the steering angle sensor 7 measure the yaw angle change and vertical / lateral displacement of the host vehicle by the host vehicle travel information detection device 8 and supply the measured value to the travel route generation device 12. The vehicle control CPU 11 is a part that controls various logics necessary for the automatic steering control vehicle.
[0014]
The route reproduction data calculation unit 15 calculates the position / direction relationship between the landscape vector sequence recognized by the landscape vector sequence recognition unit 14 and the default route recognized by the default route recognition unit 13 as route reproduction data, and the default route. Output to the estimation means 16. Then, the estimated route estimation means 16 estimates a default route based on the landscape vector sequence and the route reproduction data, and outputs it to the vehicle control CPU 11.
[0015]
Next, the operation will be described.
[Travel route generation control processing]
The travel route generation control process by the travel route generation device 12 will be described based on the flowchart of FIG.
[0016]
In step SA1, driving of the travel route generation device 12 is started.
[0017]
In step SA2, the center of the white line imaged by the CCD camera 1 is set as the travel route, and the travel route normally recognized by the default route recognition means 13 is recognized as the default route.
[0018]
In step SA3, the feature vector sequence recognition means 14 performs feature points Gn-2, Gn-1, Gn, Gn + 1, Gn + 2,..., Gn + such as street lamps in the landscape as shown in FIG. Detect n three-dimensional position. The detection of these three-dimensional positions may be obtained by trigonometry with a stereo camera, but may be obtained in a pseudo manner by the difference between two frames imaged by a monocular camera as shown below.
[0019]
For example, a method for obtaining the position of the same street lamp shown in FIGS. 4 and 5 based on the imaging result at time t and the imaging result at time t + 1 will be described.
[0020]
For the sake of clarity, the deviation from the default route of the vehicle at time t and the yaw angle are both 0, and the vehicle coordinate system has the x-axis in the vertical direction and the x-axis in the horizontal direction as the y-axis. Explain the equal situation.
[0021]
In FIG. 4, the position of the vehicle at time t and the position at time t + 1 are shown with reference to the coordinate system at time t. The values to be obtained are the horizontal distance y at time t and the vertical distance x with the streetlight at time t + 1.
[0022]
θ1 is an angle around the yaw angle with the streetlight at time t, and is obtained from the lens specifications of the CCD camera 1 in correspondence with the point where the streetlight is recognized, as shown in FIG.
[0023]
θ2 is an angle around the yaw angle with the streetlight at time t + 1, and is obtained in the same manner as θ1.
[0024]
θ3 is obtained by a yaw rate sensor mounted on the own vehicle based on a change in yaw angle at time t and time t + 1 of the own vehicle. Note that if multiple streetlights are recognized, θ3 captures several streetlight points that are considered far away from the point sequence extending toward the vanishing point (for example, Gn + 10, Gn + 11, Gn + 12, At time t and t + 1, the streetlight point where the difference in yaw angle found on the imaging screen between Gn + 10 and Gn + 11 and between Gn + 11 and Gn + 12 does not change within the imaging resolution) You may obtain | require by seeing the amount moved simultaneously within the imaging screen during t + 1.
[0025]
X corresponds to the distance traveled in the x-axis direction at time t at time t and time t + 1. If θ2 is sufficiently small, it may be obtained by approximating it as the distance traveled, but ΔX / cos Δθ is integrated by the yaw angle change Δθ and the travel distance ΔX every minute time from time t to time t + 1. Calculate by value. However, when clear steering is performed, it is considered that the vehicle is moving in a circular motion, and it is widely known by approximating a steady circular turning state and every minute time using a steering angle, lateral G, yaw rate sensor, etc. It may be obtained by a vehicle motion.
[0026]
The lateral distance Yoffset is the amount of movement in the y-axis direction at times t and t + 1. It is detected by the amount of movement in the y-axis direction of the host vehicle in the default route during the default route recognition, and relative to the vehicle route during the default route non-recognition. It is obtained by calculating Xysin θy one by one from the yaw angle θy and the travel distance Xy of the vehicle.
[0027]
These values are expressed by the following equations (1) and (2).
Tanθ1 = y / (X + x) (1)
Tan (θ2 + θ3) = (y−Yoffset) / x (2)
By solving this simultaneous equation, x and y can be obtained.
[0028]
Similarly in the height direction, the following equations (3) and (4) are established from FIG. The height z and the distance x are obtained, and θz1 and θz2 are the angles in the pitch angle direction obtained from the vertical position in the imaging screen at times t and t + 1, respectively. It should be noted that the effect of obtaining θz1 and θz2 may be eliminated by referring to the pitch rate variation of the vehicle and the streetlights Gn + 10, Gn + 11, Gn + 12 that are sufficiently far away as in the case of the pitch rate sensor. .
tanθz1 = z / (X + x) (3)
tanθz2 = z / x (4)
By solving these simultaneous equations, x and z can be obtained.
[0029]
Note that x must be the same as that calculated when determining the position in the lateral direction y of the streetlight and that calculated when determining the position in the height direction z of the streetlight. Depending on the positional relationship, when θ1 and θ2 in Equations (1) and (2) are close to 0, the change between time t and t + 1 is not sufficient, so when obtaining x, (1), Compare the magnitude of θ1 and θ2 in Eq. (2) with the absolute angle of θz1 and θz2 in Eqs. (3) and (4), and give priority to x calculated using the larger angle. Used for. For example, when the streetlight is almost directly above the vehicle, θ1 and θ2 are close to 0, and it is determined that θz1> θ1 and θz2> θ2, and are calculated by equations (3) and (4) using θz1 and θz2. In other words, y is obtained by substituting x into the equations (1) and (2).
[0030]
Streetlights Gn, Gn + 1, Gn + 2, ... recognized in this way, vertical position xGn, xGn + 1, xGn + 2, xGn + 3, ..., xGn + n , And lateral positions yGn, yGn + 1, yGn + 2, yGn + 3,..., YGn + n and heights zGn, zGn + 1, zGn + 3,.
[0031]
On the still screen, the yaw angle difference and pitch angle difference between street lights Gn and Gn + 1, between Gn + 1 and Gn + 2, between Gn + 2 and Gn + 3, and so on change smoothly and continuously. By identifying whether or not it is a continuous streetlight row, and the intervals of the vertical distances in the x direction of each streetlight Gn, Gn + 1, Gn + 2,. The height is equal Premise As an example, a streetlight model may be provided in the screen to improve accuracy.
[0032]
Specifically, Gn, Gn + 1, Gn + 2 and some of the streetlights closest to the vehicle of X and z calculated from the streetlight and the amount of change in the y direction according to the detected distance of the predetermined route (or the amount of change obtained from the map data 4 obtained from the navigation system device 9) are expressed as Gn, Gn + 1. , Gn + 2, Gn + 3,..., Gn + n are applied to correct variations due to the recognition accuracy of the detected streetlights.
[0033]
If many street lights are recognized as Gn, Gn + 1, Gn + 2, Gn + 3, ..., Gn + n, Gn, Gn + 1, Gn closest to the vehicle among the street lights The values of x and z calculated from +2 and several street lamps are slightly changed as shown in FIG. 7, and the position error from the street lamp sequence in the static screen is added as shown in FIG. It is also possible to identify the exact value of x by obtaining the above. In addition, by comparing the lateral displacement to the streetlight vector caused by the curvature change in the y-axis direction with the streetlight row on the screen, the curvature of the streetlight row can be adjusted by performing curvature matching with the streetlight vector in the same way as when white lines are recognized. You may identify.
[0034]
For streetlights Gn, Gn + 1, Gn + 2, ... recognized in this way, the vector connecting Gn and Gn + 1 is GVn, and the vector connecting Gn + 1 and Gn + 2 is GVn + 1. In this way, streetlight vectors GVn, GVn + 1, GVn + 2, GVn + 3,..., Gvn + n are obtained as shown in FIG.
[0035]
Returning to FIG. 2, in step SA4, the route reproduction data calculation means 15 obtains the xyz position, yaw angle, and pitch angle of each streetlight vector with the default route, and stores the route difference matrix as route reproduction data.
[0036]
In addition, a vector difference matrix indicating the relationship between the vectors GVn and GVn + 1 and the relationship between GVn + 1 and GVn + 2 following the streetlight Gn closest to the host vehicle is obtained and stored.
[0037]
This vector difference matrix is always based on the vector difference matrix from GVn + 1 when one end of GVn falls outside the imaging range so that there is a streetlight vector in the vertical direction of the vehicle, that is, in the x-axis direction 0. Used for virtual GVn reproduction. Reproduction of the streetlight vector outside the imaging range is performed one by one using the vector difference matrix and the streetlight vector within the imaging range.
[0038]
In step SA5, the path reproduction difference vector calculated in the previous step is stored for use in processing in the next routine and thereafter.
[0039]
In step SA6, based on the difference matrix indicating the relationship between the default route obtained as route reproduction data in step SA4 and the streetlight vector for the recognized streetlight vectors GVn, GVn + 1,... Thus, by reproducing the default route and comparing the vehicle position from the streetlight with the reproduced default route, the vehicle position deviation and the yaw angle with respect to the default route are calculated.
[0040]
In step SA7, it is determined on the basis of the signal sent from the default route recognition means 13 whether the automatically steered vehicle has lost the default route for performing the follow-up control (whether it has been lost or not). If the route is not lost, the process proceeds to step SA8, and if lost, the process proceeds to step SA9.
[0041]
In step SA8, based on the predetermined route, the route information, the positional deviation with respect to the route of the own vehicle, and the yaw angle are output as usual.
[0042]
In step SA9, the route information (curve start point and curvature information) obtained in step SA6, the lateral position deviation with respect to the host vehicle, and the yaw angle are output. For example, when only the lateral position deviation closest to the host vehicle is recognized by the default route recognizing unit 13, for example, the yaw angle of the estimated route and the route information are supplied to the lateral position deviation of the default route. Only information that can no longer be recognized may be supplied.
[0043]
Next, the effect will be described.
In the travel route generation device of the first embodiment, the effects listed below can be obtained.
[0044]
(1) When the default route was lost, the default route was reproduced based on the difference matrix indicating the relationship between the default route and streetlight vector obtained as route reproduction data for the recognized streetlight vector. Since the vehicle position shift and yaw angle with respect to the default route are calculated, the vehicle that is performing the automatic steering based on the default route that recognizes the position / direction relationship with the recognized default route is Even if you lose track of the route, the estimated route is supplied, so if the state of losing sight of the route continues intermittently, automatic steering control can be continued, and the driver can respond even if the state of losing sight of the route continues A sufficient margin can be secured.
[0045]
(2) For recognized streetlight vectors Gn, Gn + 1, Gn + 2, ..., GVn is the vector connecting Gn and Gn + 1, GVn + 1 is the vector connecting Gn + 1 and Gn + 2. Thus, the streetlight vectors GVn, GVn + 1, GVn + 2, GVn + 3, ..., Gvn + n were determined. These streetlight vectors GVn, GVn + 1, GVn + 2, GVn + 3, ..., Gvn + n are the nighttime with uneven lighting of the default route or the situation where the light is reflected when the road surface is wet, etc. It is possible to recognize the default route even in a state where the recognition of the predetermined route is ceaseless, and it is possible to supply effective route estimation information while the route is lost because the streetlight and the like can be looked far over the extension in the imaging scene. Streetlight vectors GVn, GVn + 1, GVn + 2, GVn + 3, ..., Gvn + n lost the path because the object to be imaged is not easily blocked by an obstacle and is not easily removed from the imaging range. At this time, the vehicle can be easily guided away from the predetermined route.
[0046]
(3) When multiple streetlight vector sequences are recognized, the distance from the default route, the continuity of streetlight vector changes, and the degree of coincidence of the shape with the default route are evaluated, and the results are multiplied. Based on the above, the suitability of each streetlight vector string is calculated, and the priority order is given to the streetlight vectors in the order of high suitability, thereby reducing the recognition of streetlight trains drawn into unnecessary branches. A streetlight line in a shelter space installed on one side But Even in the case of being drawn in, it is possible to select a streetlight row having a shape closest to the default route, such as a streetlight row parallel to the default route along the center of the road or the opposite lane.
[0047]
(4) The default route is reproduced based on the difference matrix indicating the relationship between the default route obtained as route reproduction data for the recognized streetlight vector and the streetlight vector, and from the streetlight to the reproduced default route. By comparing the vehicle position with the vehicle position, the vehicle position shift and the yaw angle with respect to the predetermined route are calculated. Therefore, it is possible to estimate a continuous route when the default route is lost.
[0048]
(Second embodiment)
Since the configuration of the travel route generation device of the second embodiment is the same as that of the first embodiment, description thereof is omitted.
[0049]
Next, the operation will be described.
[Travel route generation control processing]
Next, the travel route generation control processing by the travel route generation device 12 will be described based on the flowchart of FIG.
[0050]
Steps SB1 and SB2 are the same as SA1 and SA2 in FIG.
[0051]
In step SB3, the landscape vector string recognition means 14 recognizes street lamp vector strings GVn, GVn + 1, GVn + 2,..., GVn + n along a predetermined route constituted by continuous street lamp vectors. When there are a plurality of streetlight vector sequences, the plurality of streetlight vector sequences ... 1GVn + n, ..., 2GVn + n, ..., nGVn + n are recognized.
[0052]
In step SB4, with respect to the streetlight vectors GVn, GVn + 1,... Already stored in step SB3, whether or not there is a situation in which streetlights related to the vectors are concealed by a signboard or the like as shown in FIG. Determine. If concealment has occurred, the process proceeds to step SB5, and if concealment has not occurred, the process proceeds to step SB6.
[0053]
In step SB5, the most easily recognizable point, such as the edge of a signboard that conceals a streetlight vector, or the center point of a round sign, is recognized as a temporary streetlight point GVn + atemp. Then, the temporary streetlight vector, the difference matrix between the temporary streetlight vectors shown in FIG. 13, and the route difference matrix between the temporary streetlight vector and the default route are obtained, and processing is performed in the same manner as the streetlight vector.
[0054]
When the signboard is recognized, the positional relationship between the streetlight that is expected to be concealed and the streetlight concealment such as the signboard is detected before the streetlight is concealed, and the position of the signboard and other recognizable streetlights By reproducing a streetlight hidden from the relationship, the streetlight vector may be virtually recognized to proceed with the process.
[0055]
In step SB6, among the streetlight vectors recognized in step SB3, recalculation is performed in order to keep the calculation processing related to the streetlight vectors already calculated in the previous loop with high accuracy, so that recognition with improved recognition accuracy can be performed. If the information is stored, the stored information is updated.
[0056]
Step S B In case 7, multiple streetlight vector strings are recognized, the distance from the default route, the continuity of the streetlight vector change, and the degree of coincidence with the default route are evaluated, and the result of multiplying these evaluation results Based on the above, the fitness of the streetlight vector sequence is calculated. Then, priorities are assigned to the streetlight vectors in the order of high fitness.
[0057]
When the map data 4 is recognized, the streetlight vector sequence may be prioritized based on the road shape of the extension of the predetermined route and the recognition result of the streetlight vector sequence.
[0058]
In step SB8, it is determined whether or not the recognized streetlight vector sequence is a section for which a route difference vector with respect to the default route is obtained. The process proceeds to step SB9 for the streetlight vector for which the route difference vector is obtained, and the process proceeds to steps S12 and S18 for the streetlight vector for which the route difference vector is not obtained.
[0059]
Steps SB9, SB10, and SB11 are the same as steps SA4, SA5, and SA6 in FIG.
[0060]
In step SB12, the streetlight vector sequence obtained in step SB3 is collated with the map data 4, and the degree of coincidence is obtained. As for the degree of match, the degree of match is recalculated only for the section for which the route difference vector with the default route cannot be calculated for the streetlight vector sequence that has a certain degree of fit in the route for which the suitability is calculated in Step SB7. Alternatively, it may be the result calculated for the streetlight vector sequence having the highest degree of coincidence.
[0061]
In step SB13, if the degree of coincidence obtained in step SB12 is high, the process proceeds to step SB14 because the reliability of the streetlight vector is high. If the degree of coincidence is low, the process proceeds to step SB20 because the reliability of the streetlight vector is low.
[0062]
In step SB14, it is determined whether or not the route selected as a result of matching in step SB12 is a straight line. If it is a straight line, the process proceeds to step SB17. If it is not a straight line, the process proceeds to step S15.
[0063]
In step BS15, the corner starting point on the navigation map is obtained by obtaining the point having the highest correlation by shifting the streetlight row selected in step SB12 and the map data 4 back and forth further and taking the cross-correlation. To detect.
[0064]
In step SB16, the one having higher accuracy is selected from the road curvature information estimated from the street lamp train and the curvature information included in the map data 4. Specifically, the streetlight recognition interval and the navigation position point sequence data interval are compared, and the narrower one is selected as having higher accuracy. However, if the map data 4 has previously known information corresponding to road curvature information, the map data 4 is selected.
[0065]
In step SB17, using the route reproduction data calculated in step SB10, the default route is reproduced by the streetlight vector that is the recognized route estimation target.
[0066]
In step SB18, it is confirmed that z described in step SA3 is constant. Premise Assuming that the height of the road.
[0067]
In step SB19, if a swell is detected in step SB18, it is determined that the top of the uphill is approaching, and then the process proceeds to step SB20. If it is detected that there is a bulge in the height of the road, the pitch angle of the CCD camera 1 is set so as not to lose sight of the result of imaging of the white line and street lamp by the CCD camera 1 when passing through the bulge. It is stored as reference information to be adjusted.
[0068]
In step SB20, it is determined whether or not a predetermined route has been secured by combining the predetermined route and the estimated route. The travel route here refers to information on the minimum lateral position deviation and yaw angle. Regarding the estimated route, at least one streetlight vector or a correction vector based on a signboard / sign can always stay in the screen completely. Say. If the travel route is secured, proceed to step SB22; otherwise, proceed to step SB21.
[0069]
In step SB21, since it is predicted that the route will be lost, an alarm is issued to the driver. Note that the alarm may be changed according to the remaining distance (or remaining time), and in order to take an appropriate response to the remaining distance (remaining time), a command may be issued to the automatic steering controller 10 that controls automatic steering.
[0070]
Steps SB22, SB23, and SB24 are the same as steps SA7, SA8, and SA9 in FIG.
[0071]
Next, the effect will be described.
In the travel route generation device of the second embodiment, the effects listed below can be obtained in addition to the effects of the first embodiment.
[0072]
(5) Since the streetlight vector is compared with the map data 4, it is possible to select and determine a highly reliable landscape vector, including the streetlight vector of the default route and the unmatched part, so when the default route is lost In addition, the accuracy of route estimation of the unmatched portion is improved.
[0073]
(6) The streetlight vector sequence is collated with the map data 4 to obtain the degree of coincidence. The shape of the landscape vector and the curvature of the map data 4 for the streetlight vector sequence having the highest degree of coincidence with the road curvature information estimated from the streetlight sequence Since it is decided to select the more accurate information, the road curvature information that works effectively to ensure the followability of the automatically steered vehicle even when the route ahead is lost is supplied continuously and accurately. Is possible.
[0074]
(7) The most recognizable point of the concealment that conceals the streetlight vector is recognized as the temporary streetlight point GVn + atemp, and the route difference matrix between the temporary streetlight vector and the default route is obtained, so the landscape vector Even if the image is concealed, the once-recognized landscape vector can be geometrically reproduced, and the estimated route based on the landscape vector can be reproduced continuously even when the default route is lost.
[0075]
(8) Since the predetermined height of the recognized streetlight vector is stored and the height of the road is estimated, the pitch angle of the CCD camera 1 is set in anticipation of a change in the vehicle pitch angle. It is possible to adjust, and it is possible to prevent the scene vector from being out of the imaging range at the top of the undulation where it is difficult to recognize the white line.
[0076]
(9) When it is predicted that the travel route will be lost, the driver is warned and the command is transmitted to the automatic steering controller 10. Therefore, the automatic steering operation is canceled before the route supply is actually stopped. It is possible to secure a margin time for prompting.
[0077]
(Other embodiments)
As mentioned above, although embodiment of this invention has been described based on 1st Example and 2nd Example, the concrete structure of this invention is not limited to this Embodiment, The summary of invention is shown. Design changes and the like within a range that does not deviate are also included in the present invention.
[Brief description of the drawings]
FIG. 1 is a configuration diagram illustrating a vehicle to which a travel route generation device according to a first embodiment is applied.
FIG. 2 is a flowchart showing a flow of a travel route generation control process of the first embodiment.
FIG. 3 is a perspective view illustrating a method for detecting feature points such as street lamps in a landscape.
FIG. 4 is an explanatory diagram showing a streetlight vector calculation method.
FIG. 5 is an explanatory diagram showing a streetlight vector calculation method.
FIG. 6 is an explanatory diagram showing lens specifications of a CCD camera.
FIG. 7 is an explanatory diagram showing a streetlight vector calculation method.
FIG. 8 is an explanatory diagram showing a streetlight vector calculation method.
FIG. 9 is an explanatory diagram showing a method of reproducing a default route.
FIG. 10 is an explanatory diagram showing a method of reproducing a default route.
FIG. 11 is a flowchart showing a flow of a travel route generation control process of the second embodiment.
FIG. 12 is an explanatory diagram showing a method of reproducing a default route when a streetlight is hidden by a signboard.
FIG. 13 is an explanatory diagram showing a method of reproducing a default route when a streetlight is hidden by a signboard.
[Explanation of symbols]
1 CCD camera
2 Image processing CPU
3 GPS antenna
4 Map data
5 Gyro sensor
6 Wheel speed sensor
7 Steering angle sensor
8 Vehicle travel information detection device
9 Navigation system device
10 Automatic steering controller
11 Vehicle control CPU
12 Travel route generator
13 Default route recognition means
14 Landscape vector sequence recognition means
15 Route reproduction data calculation means
16 Default route estimation means

Claims (9)

Imaging means for imaging a landscape in front of the vehicle;
A prescribed route recognition means for recognizing a prescribed route from a white line in a captured landscape;
Automatic steering control means for controlling the steering angle of the vehicle based on the recognized default route;
With
In a travel route generation device for a vehicle that performs follow-up traveling by automatic steering along a traveling road,
Recognizing landscape vector string recognizing means vector sequence as landscape vector sequence connecting the placed characteristic points located away from the road surface in the captured scene,
Route reproduction data calculating means for calculating the position / direction relationship between the recognized landscape vector sequence and the default route as route reproduction data;
A default route recognition judging means for judging whether or not a default route has been recognized;
A default route estimation means for estimating a position / orientation of the default route based on the recognized landscape vector sequence and the calculated route reproduction data when it is determined that the default route is not recognized;
A travel route generating device characterized by comprising: In the travel route generation device according to claim 1,
The landscape vector sequence recognizing means limits an area to be recognized based on a running road, captures a characteristic point sequence arranged in a predetermined route extension direction in an imaged landscape, and connects a line connecting the point sequences. A travel route generation device characterized by being a means for recognizing a landscape vector sequence. In the travel route generation device according to claim 1 or 2,
The route reproduction data calculation means is a processing target when a plurality of landscape vector sequences extending in the direction of a default route of the host vehicle exist in parallel and the recognized landscape vector sequence has a high degree of fitness with the default route. A travel route generation apparatus, characterized in that the travel route generation device is configured to calculate a position / posture relationship between each landscape vector in the selected landscape vector sequence and a predetermined route. In the travel route generation device according to any one of claims 1 to 3,
When it is determined that the default route is not recognized, the default route estimation means estimates the default route based on the recognized landscape vector sequence and the calculated route reproduction data, and also determines the lateral position of the vehicle relative to the default route. A travel route generation device characterized by being means for estimating a deviation and a yaw angle. In the travel route generation device according to any one of claims 1 to 4,
A navigation system for identifying the shape of the road around the vehicle and the position of the vehicle on the road;
Landscape vector trust that compares the obtained road shape with the shape of the recognized landscape vector sequence and determines whether the landscape vector sequence extending beyond the recognized default route is along the running road Sex determination means;
A travel route generating device characterized by comprising: In the travel route generation device according to claim 5,
The landscape vector reliability determination means detects a corner start point based on the result of collating the shape of the landscape vector with the information on the shape of the road ahead, and recognizes it as curvature information of the curve following the detected corner start point destination. A travel route generation device, characterized in that it is a means for selecting the one with higher accuracy from the curvature information estimated from the interval between the characteristic points and the curvature information estimated from the information on the shape of the road ahead. The travel route generation device according to any one of claims 1 to 6,
The landscape vector sequence recognition means includes a landscape vector concealment detection unit that detects a landscape vector concealment that blocks a landscape vector sequence,
A landscape vector concealment feature point recognition unit that temporarily recognizes a characteristic point of the detected landscape vector concealment as a calculation target point of the landscape vector sequence;
A travel route generating apparatus comprising: In the travel route generation device according to any one of claims 1 to 7,
The landscape vector sequence recognition means stores a predetermined height of a recognized landscape vector , and a road undulation shape estimation unit that estimates a road undulation shape based on the information,
An imaging angle adjustment unit that adjusts the imaging angle of the imaging means based on the estimated road undulation shape;
A travel route generating apparatus comprising: In the travel route generation device according to any one of claims 1 to 8,
The predetermined route estimation means predicts that the supply of the travel route will stop based on the recognition sight distance of the default route and the sight distance combined with the recognition status of the estimated route by the landscape vector, and transmits the result to the automatic steering control means A travel route generating apparatus characterized by comprising:

Images