JPH03231311A - Automatic traveling device - Google Patents

Abstract

PURPOSE:To perform the optimum traveling control of a vehicle in order to make the vehicle follow a target route by obtaining the target control value with which the vehicle can follow the target route and setting the vehicle to be controlled at the steering angle of the vehicle with the target control value defined as the steering angle correction value respectively. CONSTITUTION:The input image sent from an image pickup part 1 which photographs the surface of the frontward road of a vehicle is differentiated by an image processing part 2 to undergo the detection of its edge. Then the optimum threshold value is automatically set by an automatic threshold value setting circuit of the part 2 and according to the gradation of the input image. Then the edge image is binarized. Based on this processed image, the X-Y coordinates are converted into the rho-theta coordinates. Thus the continuous segment information on the road edge are obtained. A travel enable area recognizing part 3 applies the projection conversion processing to the relevant image and recognizes the area between the continuous road edges as a travel enable area of the vehicle. A target route setting part 4 sets a target route on the recognized road as an optimum traveling route. A control part 5 sets the target control value and obtains the steering angle correction value for traveling of the vehicle on the target road to define that correction value as the target control value for traveling control of the vehicle.

Description

DETAILED DESCRIPTION OF THE INVENTION Technical Field The present invention relates to an automatic driving device that allows a vehicle to automatically travel while searching five driving routes.

Prior Art Recently, an automatic driving device has been developed that searches its own driving route, sets an optimal target route on the route, and controls the vehicle so that the vehicle travels on the target route. has been done.

This type of automatic driving system detects the current running conditions of the vehicle, such as the vehicle speed, steering angle, and yaw rate (angular velocity of change in one direction), and then performs a predetermined operation based on the detected running condition of the vehicle. It is conceivable to perform vehicle travel control by determining a control target amount that allows the vehicle to follow the target route through arithmetic processing, but in this case, in order to improve the ability of the vehicle to follow the target route, Another important issue is what to set as the controlled object and how to determine the control target amount.

-l-like The present invention has been made in consideration of the above points.

When detecting the current running state of the vehicle and determining a control target amount that allows the vehicle to follow and run on the target route based on the detected running state of the vehicle, and controlling the running of the vehicle, An automatic driving system is provided in which the steering angle of the vehicle is set as the control target, the steering angle correction amount is set as the control target amount, and the steering angle is appropriately controlled so that the vehicle always travels on the target route. This is what we provide.

DESCRIPTION OF THE PREFERRED EMBODIMENTS An embodiment of the present invention will now be described in detail with reference to the accompanying drawings.

In the automatic driving device according to the present invention, as shown in FIG.
and an image processing section 2 and 5 that process the image captured by the video camera and extract continuous line segments, etc., and 5. A driveable area such as a road in the direction of travel of the vehicle according to the extracted continuous line segments. L2m, a travelable area recognition unit 3, a target route setting unit 4 that sets a target route for the vehicle within the recognized travelable area, a vehicle speed sensor 6 that detects the vehicle travel speed V, and a vehicle speed sensor 6 that detects the vehicle travel speed V. A yaw rate sensor 7 that detects a yaw rate T as the vehicle travels, and a steering angle sensor 8 that detects a tire angle δ caused by steering the vehicle.
The current running state of the vehicle is detected according to the output of each sensor such as, and based on the detected running state of the vehicle, a target amount of steering control necessary for the vehicle to travel on the target route is determined. and the control unit 5 which is obtained by the calculation process.

It is comprised of a steering control section 9 and a steering drive section 10 that steer the vehicle in accordance with the determined control target amount.

Actually, the image processing section 2. The travelable area recognition section 3, target route setting section 4, and control section 5 are replaced by a microcomputer. It is also possible to include the steering control section 9 in the computer.

In addition to the standard lens, the video camera in the imaging unit 1 is equipped with a telephoto lens or a wide-angle lens to obtain images appropriate to the vehicle speed, road conditions, etc. The vehicle is equipped with multiple special video cameras such as infrared cameras and ultra-high sensitivity cameras, and under computer control, these multiple video cameras are switched as appropriate depending on the vehicle speed, the state of the captured image, etc. It looks like this.

Furthermore, it is also possible to arrange two video cameras with the same imaging characteristics in parallel to obtain binocular stereoscopic images.

Extraction of continuous line segments such as road edges in the image processing section 2 is performed as follows.

First, an image edge detection process is performed by differentiating the input image sent from the imaging unit I that images the road surface in front of the vehicle. The setting circuit automatically sets an optimum threshold value according to the degree of shading of the input image at that time, and binarizes the edge image.

In this case, the input image must be binarized first, and then
Differential processing may be performed for edge detection. Further, instead of performing binarization, multi-value conversion that expresses the shading of the image may be performed.

Next, based on the edge detected and binarized or multivalued processed image 1, line segments on the X-Y coordinates are expressed as ρ-
By performing Hough transformation processing, which is a well-known method that performs gray coordinate transformation on points on the θ coordinate, it is possible to combine continuous point sequences and remove isolated points without continuity. For example, information on continuous line segments of road edges as shown in FIG. 2 is obtained.

Here, 0 is the angle of a perpendicular drawn from a straight line on the XY coordinate to the origin of the coordinate, and ρ is the length of the perpendicular. For example, the line segment on the X-■ coordinate shown in Figure 4 is
As shown in FIG. 5, it appears as point 01 at the ρ-O coordinate.

At this time, edge tracking processing may be performed based on the binarized processed image to find edge portions of the image that have continuity.

In addition, Houg is used to determine the continuity of image edges.
By performing multiple processes such as h-conversion processing and edge tracking processing in parallel, and making comprehensive judgments based on the results of each process, more accurate road edge information can be obtained. . Furthermore, if the above-mentioned process for extracting continuous image edges is performed while growing the region of the input image as the vehicle travels, it is possible to extract road edge information with higher accuracy. Become.

Since the image captured by the video camera in the imaging unit 1 is based on perspective projection, the driveable area recognition unit 3 converts the image of the road edge based on perspective projection as shown in FIG. 2 into the image shown in FIG. Projective transformation processing, which is a known method for converting images of road edges that eliminate the effects of perspective projection, is performed, and the area between consecutive road edges is recognized as a driveable area for a vehicle.

Note that the projective transformation characteristics are set in advance in the driveable area recognition unit 3 according to the perspective projection characteristics of the video camera.

Next, when the road in front of the vehicle that is the recognized drivable area is recognized by the drivable area recognition unit 3, the target route setting unit 4 determines the optimum driving route for the vehicle on the recognized road. A target route is set as follows.

The target route is desirably set to be suitable for the vehicle's driving conditions at that time, taking into consideration the road shape and vehicle speed, as will be described later. It is uniformly certified as follows depending on whether it is narrow or wide.

That is, when the target route setting unit 4 determines that the road is a wide track with a width of a certain level or more, for example, as shown in FIG. A target route ○C is set along the reference edge with a predetermined gap width W of about 1.5 m.

Furthermore, if the target route setting unit 4 determines that the road is a narrow track with a width less than a certain level, the target route is set at the center of the road, although not particularly shown.

The position on the X-Y coordinates of the set target route is then recorded in the internal memory of the target route setting section 4 while being updated sequentially as the vehicle travels. At this time, the scale on the X-Y coordinates is determined by the magnification of the video camera in the imaging section l.

Note that in FIG. 6, point P indicates the current position of the vehicle; for example, the center lower end of the imaging surface by a video camera is P.
The mounting position of the video camera in the vehicle is set in advance so that it becomes a point.

In the figure, the trajectory from point P to point O is determined by the steering control of the vehicle under the control of the control unit 5, as described later, so that the vehicle at point P joins the target route ○C. It shows the driving route until the end. The 0 point is the merging position of the vehicle to the target route OC at that time.

Further, in the present invention, it is also possible to detect the driving state of the vehicle and set an optimal target route for the vehicle on the road in accordance with the detected driving state as described below.

That is, in the target route setting section 4, for example.

The vehicle speed detected by the vehicle speed sensor 6 is read, and if the vehicle speed at that time is within the low speed range below a preset threshold, the road shape is changed as shown in FIG. 7(a). Set the target route OC to follow.

Similarly, if the vehicle speed at that time is within a high speed range exceeding a preset threshold value, the vehicle is traveling on a winding road as shown in FIG. 7(b).

A target route Oc is set within the road with a gentle curvature so that the lateral acceleration acting on the vehicle is reduced as much as possible.

Next, once the target route on the road is set.

In the control unit 5, a control target amount for causing the vehicle to join the target route is determined by the following calculation process.

In this case, in particular, the present invention sets the control target to the steering angle of the vehicle, predicts the future traveling route from the currently detected traveling state of the vehicle, and calculates the difference between the predicted route of the vehicle and the target route. Find the steering angle correction amount for the vehicle to travel on the target route from , and use the steering angle correction amount as the control target amount,
The vehicle is controlled to travel in accordance with the steering angle correction amount.

Specifically, 1. For example, if the current direction of the vehicle is in the Y-axis direction based on the running state of the vehicle, the vehicle arrival position on the X-axis at a certain distance ahead in the Y-axis direction is predicted, and the predicted position and Y
The same X on the target path at a certain distance in the axial direction
The steering angle correction amount is calculated according to the deviation from the target position on the axis.

Now, for example, as shown in FIG. 8, a vehicle 45 at point P
Let us consider the case where the route OC merges with the target route OC.

First, based on the current vehicle speed v (m/s) of the vehicle detected by the vehicle speed sensor 6, t n+ of the vehicle at point P
Distance L (m) on the Y axis after seconds (L=vxtn,)
is calculated, and the deviation xjl between the 0 point on the Y axis, which is the distance away from the point P, and the target route OC, that is, [,. A value proportional to the position on the target route OC is calculated.

Similarly, a predicted route AC at the current steering angle of the vehicle is calculated based on the yaw rate T (rad/s) of the vehicle detected by the yaw rate sensor 7, and a predicted route AC is calculated from the zero point on the Y axis. The AC deviation Xm+, that is, a value proportional to the position on the predicted route AC after tmm is obtained as follows.

Now, when the radius of the arc drawn by the predicted route AC is R,
xm is given by the following equation.

x,,, =R-(R"-(vXtm)z)"
=R-R(1-(vXtm/R)")"Here, if R>vxtll.

x,,,'=R-R(1-(v X trn/R)"
/ 2) = v2 T/2v (3) Note that the sign of the yaw rate T is positive when, for example, 7Il+1 Benji AC is turning left.

Then, the difference e(e
=)cλ-xm), the yaw rate ΔT of the vehicle to be corrected is determined according to the following formula.

ΔT=eX2v/L2 − (4) Next, the tire angle δ of the vehicle at point P detected by the steering angle sensor 8 is taken in, and the tire angle control unit 45JLδ′ for merging the vehicle onto the target route OC It is determined as follows.

Now, we have the relationship shown in Figure 9.

When R>Q.

δ12/R...(5) From equations (2) and (5), δ=(12/v)T...(6) can be obtained. Here, Q is the wheelbase.

Therefore, in view of the relationship in equation (6), the correction amount Δδ of the tire angle according to the yaw 1−ΔT of the vehicle to be corrected is given by a linear equation.

Δδ=(4/v)ΔT (7) Here, the general formula for steering angle with respect to vehicle speed V is Q(1+Kv
”), from equation (7), Δδ=ΔT ((1(
1+Kv'')/v)-(8) is a constant coefficient (stability factor) determined by vehicle characteristics such as tire characteristics and wheel base.

Then, the target tire angle control amount δ' for causing the vehicle to join the target route OC is determined as δ'=δ+Δδ (9).

The steering control section 9 issues a drive command to the steering drive section 10 in accordance with the control target amount δ' given from the control section 5, so that the steering drive section 10 drives the steering appropriately to move the vehicle toward the target path OC. Perform steering to merge.

The above process is repeated at predetermined time intervals on the order of several seconds, thereby making it possible for the vehicle to automatically travel along the intended route oC.

FIG. 10 shows a detailed block configuration example of the automatic traveling device according to the present invention shown in FIG.

Here, the 1 image processing section 2 includes a global line segment extraction section I1.
．． It consists of a local line segment tracing section I2 and a region extraction section 13.

The global line segment extraction unit 11 globally obtains line segments of boundaries of roads with simple shapes such as straight roads. Here, in the case of a straight and simple road such as an expressway, the processing of the though transformation, which is a straight line detection means, is applied to the area where the road is likely to exist, which is obtained from the mounting angle of the video camera in the image carrier 1. By applying this process, a wide range in the image can be processed at once. Even if the processing area is large enough to cover the entire road, if it is a straight road, the boundary line segments can be extracted, and only once!
The Iough conversion process enables high-speed extraction.

In addition, this global line segment extraction unit 11 converts road line segment data up to the previous processing stored in a local map database 18, which converts road line segment data (described later) into vector representations and manages them as local map data. High reliability of road line segment extraction can be obtained by taking the information and using it as a reference for processing by checking connections with road line segments that are expected to appear in the input image for this processing. That's what I do.

The local line segment tracing unit 12 locally obtains line segments of boundary lines of roads with complicated shapes. here,
On roads such as intersections, complex branch roads, and sharp curves, the road boundary lines are locally traced within a minute range in which the road boundary lines can be detected by Rough transformation.

In addition, this local line segment tracing unit 12 uses the processing results and local map data in the global line segment extraction unit 11 described above to obtain initial data such as a search starting point and a search direction. This improves the reliability of minute extraction and reduces processing time.

The area extraction unit 13 extracts the outline of a road area that does not have clear boundaries such as white lines. On asphalt roads that do not have clear boundaries such as white lines, extraction of road line segments based on the boundary between the road and the dirt, grass, etc. on the shoulder of the road becomes unstable. For such roads, region extraction using color features or the like is effective. Here, based on the assumption that the area in front of the vehicle is a road area, the road area is extracted by extracting parts similar to the features within that area. Fundamentally, using a similar method, it is possible to extract a road area based on the road temperature, for example from an infrared image at night.

In addition, the processing performed by the region extraction unit 13 is not only for extracting regions of roads where clear boundaries cannot be obtained, but also for each of the aforementioned global line segment extraction and local line segment extraction processing.
It is also used to prevent guardrails and walls at the edge of the road from being mistakenly recognized as roads.

Furthermore, the processing in this area extraction section 13 is as follows.

High reliability can be obtained by using local map data and comparing it with previous processing results.

In addition, in the configuration shown in FIG. 10, the data management unit 15 serves as the drivable area recognition and target route setting unit 14 (corresponding to the drivable area recognition unit 3 and target route setting unit 4 shown in FIG. 1). ．． Data evaluation section 16. Local map management section 17 and local map database 18. It consists of a control supervisor 19.

Here, a radar device W20 is provided for obtaining external world information other than one image and detecting obstacles in front of the vehicle.

In addition, the distance traveled and the direction of travel of the vehicle are detected.
- Find the position on the Y coordinate by calculation.

A navigation device 21 is provided that displays the determined vehicle position on a road map and outputs predetermined guidance information that has been input in advance when the vehicle reaches a scheduled position on the road map.

The data management section 15 performs time management of a plurality of image processing results in one image processing section 2, and also performs projective transformation processing for projecting an image of a road edge by perspective projection onto a two-dimensional plane. Furthermore, positional deviations on the coordinates due to differences in image acquisition time are corrected by respectively performing coordinate transformation based on a predicted route table calculated by a red road prediction unit 22, which will be described later.

In other words, the execution time of each image process in the image processing unit 2 is not the same and changes depending on the situation, so the results of each process are finally integrated at the same time and coordinates. .

Therefore, the data management unit 15 has means for independently providing a function of measuring the amount of movement of the vehicle from moment to moment using a reference clock and converting the results of each image processing executed asynchronously to the same coordinates at an arbitrary time. I try to take it.

By taking this measure, the results of each image processing performed asynchronously and in parallel can be efficiently processed at the same time.

It can be converted into coordinates, and the same processing such as coordinate transformation can be consolidated, and the load on image processing, distance information detection processing, and communication and calculation processing at the subsequent integrated evaluation stage can be effectively reduced. Communication between each process is simplified.

According to this means, by simply adding the image input time etc. to each image processing result, it is possible to process the rest completely independently. For example, it is possible to process a driving road closer to the front in a short processing cycle! ! Even if it takes some time to recognize the long-distance driving route, by running each of the seven processes in parallel, high-speed automatic driving becomes possible.

In addition, in order to integrate and evaluate multiple image processing results at the same time and coordinates, it is possible to start all the processing at the same time, and then integrate the processing results when all the processing is completed. I can think of it.

In this case, the cycle is constrained by the longest processing time, and furthermore, it becomes necessary to perform close communication between each process in order to align the start times.

In addition, if you have multiple image processing means and distance information acquisition means such as radar equipment [20], the reliability of recognition can be greatly improved by referring to and using the previous driving route recognition results and each other's processing results. can do. however,
The execution time of each of these processes is not the same, but changes depending on the situation, and the coordinate systems handled may also be different. In such a case, the data management unit 15 is designed to efficiently exchange information necessary for each process.

Therefore, the data management unit 15 measures the amount of movement of the vehicle from moment to moment using a reference clock, and outputs the driving route recognition results and other processing results required in each process executed asynchronously at any time. A method is taken to independently provide the function of converting to the coordinates of .

By taking this method, it is possible to efficiently convert time and coordinates in response to various requests for multiple processes, it is also possible to consolidate the same processing such as coordinate conversion, and it is also possible to perform communication and The load on arithmetic processing can be effectively reduced.

In order to refer to and utilize previously processed travel path recognition results and each other's processing results, it is necessary to perform time adjustment and correction based on the amount of movement of the vehicle.

However, each process directly exchanges data and converts time and coordinate systems, which creates a large burden on both communication and calculation time.

According to this means, each processing result is independently
Furthermore, data can be created at any time and in any format as desired, and communication and calculation time are significantly reduced, resulting in faster overall processing cycles.

It efficiently performs functions such as uniformly managing the processing results of multiple image processing means and distance information acquisition means such as radar equipment F120, and converting them into data on the same coordinates at any time in response to various requests. In order to achieve this, the data management section 15 is structured around a database as shown in FIG. 11.

Global line segment extraction unit 11° local line segment tracing unit 12 and area extraction unit 13 in image processing unit 2, radar Y device 20
．． Each data from the route prediction unit 22 in the control unit 5 is
First, 1 is written into the dual port RAM 27 asynchronously. Similarly, local map data and evaluation result data from the data evaluation section 16 are stored in the dual port RAM.
27.

The data management unit 15 periodically updates the dual port i~RAM.
Go read the contents of 27.

At that time, if there is new data output from each section 11 to 13, 20.22, it is sent to the database write port.
28 and stored in each database 29-34. Here, the database 29 is a practical data evaluation database, the database 30 is a global line segment extraction practical database,
Database 31 is for local line segment tracing, database 32 is for region extraction, and database 3 is for local line segment tracing.
3 is used for distance information acquisition, and database 34 is used for route prediction.

The flow of processing at this time is shown in FIG.

Also, as a result of reading the contents of dual port RAM27,
If it is a data request, the data management unit 15 searches each database 29 to 34 for data according to the request through the database readout boat 35, and further processes the projective transformation processing unit 3G, the time 5 position correction processing unit 37, and the inverse The projective transformation processing section 38 appropriately transforms the data into a desired form and writes it into the dual port RAM 39 for output.

Now, for example, if there is a data request from the data evaluation unit 16, the latest image processing result data and distance information 1
The obtained data are read from each database, and the image processing results are projectively transformed into coordinates on the road surface. Furthermore, the image processing result is corrected to the positional relationship with the vehicle at the desired time, and then passed to the data evaluation section 16.

For example, when a request for distance information is made from the image processing unit 2 at a certain time, the corresponding data is read out from a predetermined database, corrected to the position at the requested time, and delivered to the request destination. At that time, especially in the case of a request from the image processing section 2, the processing load on the image processing section 2 is reduced by passing the data that has been inverse projectively transformed to the coordinates on the image memory in the imaging section 1 to the request destination. ing.

The flow of processing at this time is shown in FIG.

In addition, when correcting the image processing result to the positional relationship with the vehicle at the requested time, if the requested time is t2 and the input time of the relevant data such as the image input time is tl, then the first
As shown in FIG. 4, time tl.

The route prediction data at time t2 (position and orientation of the vehicle on the absolute coordinates X-Y at each time) is read from the route prediction result database 34, and each point of the corresponding data at time t1 is set at time t2 using these data. Correct the position on the relative coordinates with the vehicle.

In FIG. 14, Pl represents the position of the vehicle at time t1, DI represents the traveling direction of the vehicle at the P1 position, P2 represents the vehicle position at time t2, and D2 represents the traveling direction of the vehicle at the P2 position.

The flow of processing at this time is shown in FIG.

Further, as for the position correction, as shown in FIG. 16(a), when the relative coordinate at time tl is XY+the relative coordinate at time t2 is xL, /.

As shown in the same figure (b), relative coordinates x-y at time tl
Point p (xx, yl) on the top is converted to point P' (x2°y2) on relative coordinates x'-y' at time t2.

The conversion formula at that time is given by the following formula.

Furthermore, in the configuration shown in FIG. 10, the data evaluation section 16 is
The plurality of projection-transformed road edge data received from the data management unit 15 are evaluated based on their compatibility with the road model and the relationship between the road data, and after removing incorrect data, it is determined to be appropriate. The multiple road edge data obtained are merged onto a single plane.

Here, by using a plurality of image processing results, evaluation result data that compensate for each other's weaknesses is obtained.

Therefore, 1. For example, for roads with drawn white lines, the optimal image processing results are used to extract the white lines, and for roads without white lines, the image processing results for region extraction based on color features, etc. are used. This makes it possible to respond more flexibly to a diversely changing environment.

In addition, here, in order to create reliable road data, each image processing result is verified to determine whether it is appropriate as road data or not under the following simple conditions.

A road has a certain width (road width condition).

The line segments that make up the road are connected smoothly (smoothness condition).

The number of points in the road data indicated by the point sequence must be greater than or equal to a certain value (abnormal point condition 1).

Road data does not exist outside the measurement range (abnormal point condition 2)
).

Road data does not exist outside the road area (abnormal point condition 3)
).

In this way, since the road data is evaluated in terms of points and the area data and line segment data are matched, excellent evaluation results for road recognition can be obtained.

The flow of processing in the data evaluation section 16 is shown in FIG.

Based on the three image processing results after projective transformation whose position was corrected at the same time from the data management unit 15, first, it is determined whether or not it is valid as road data based on the following conditions, and according to the result, one reliable image is determined. Road data is created and output to the local map management section 17.

Note that, at this time, by performing positional correction at the same time, it is possible to easily compare the data that is the result of the three image processings. In addition, projective transformation facilitates geometric evaluation of road data under the condition that the road is flat.

In addition, in the configuration shown in FIG. 1O, the local map management unit 17 converts the road edge data into a vector representation and stores it as local map data in the local map database 18.
Store in.

The module includes a part that expresses the environment in which the vehicle is located in the form of a local map based on data on road markings given from the data evaluation unit 16, data on obstacle positions given from the radar device 20, etc. It consists of a part that manages the created local map.

A local map is a map in which road markings, obstacle positions, etc. are mapped onto the plane coordinates of the ground surface with the vehicle as the origin.

Then, the local map management section 17 includes the image processing section 2,
Data management unit 15, data evaluation unit 16: In response to a map reference request from the navigation device 21, a map of the necessary range is retrieved from the local map database 18, coordinates are converted according to the vehicle position at the current time, and output. do.

In addition, the local map management unit 17 superimposes the line segment of the road edge at the time of current processing on the local map, determines the attributes of the line segment based on the connection relationship with past data, the relative positional relationship with the vehicle, etc. Remove line segments that do not belong to any attribute.

The advantage of having a local map is that it makes it possible to handle outputs from various sensor systems in a unified manner, and
- degree image processing makes it possible to use past data to supplement areas that become invisible due to blind spots in the camera field of view, such as curved corners, and also improves the reliability of the system by having the sensor system refer to the local map. Examples include things that can be improved.

Furthermore, the local map management section 17 creates and maintains a plurality of local maps. As a result, if there is an access request from another component.

This method has the advantage of not having to wait for the completion of the local map that is currently being processed as it is only necessary to refer to the data of the local map that has already been processed, and the advantage that access management becomes easier because exclusive control is no longer required. Motsu.

Furthermore, in the configuration shown in FIG. 10, the control supervisor 19
uses a man-machine interface with the driver to switch between automatic driving and manual driving, and to switch between starting and stopping automatic driving.

The control supervisor 19 also receives the local map, map information from the navigation device T121, and information from the navigation device T121.
Based on outside world information other than images of obstacles detected by the radar device 20, a driving lane is selected, the necessity of avoiding obstacles is examined, and a driveable area is determined. Then, a target route for the vehicle is set within the determined drivable area, and the data of the target route is sent to the control unit 5.
You can learn from '.

In addition, the control supervisor 19 determines the traveling speed of the vehicle by taking into account information such as the visual field distance of the image, the time required for one image processing of the vehicle speed, and the legal speed on the road on which the vehicle is currently traveling obtained from the navigation device 21. The determined running speed and each designation of the running distance until reaching the running speed are given to the control section 5'.

The control section 5' here includes a route prediction section 22, a route control section 23, and a vehicle speed control section 24.

The route prediction unit 22 calculates the momentum of the vehicle body determined by the vehicle speed V, yaw rate T, and tire angle δ detected by each sensor (other factors such as the sideslip angle of the vehicle body detected by the lateral acceleration sensor may also be taken into consideration). Accordingly, a predicted route of the vehicle on the X-Y coordinates (predicted route AC in FIG. 8) is calculated. It also serves as a reference clock for the device by writing the time and coordinates on the shared memory.

The route control unit 23 determines the current vehicle position and the travel of the vehicle according to the data of the target route set within the drivable area given by the control supervisor 19 and the data of the predicted route given from the route prediction unit 22. Considering the condition,
To generate a travel plan route that is realizable and allows a vehicle to smoothly follow a target route. Then, a target value of the steering angle according to the planned travel route is determined, and a control command for five steering angles is given to the steering control section 9 to perform control so that the actual steering angle becomes the target value. that time.

Steering angle control is performed that feeds back the driving state of the vehicle and causes the vehicle to follow the planned travel route.

The flow of processing at this time is shown in FIG.

The vehicle speed control unit 24 plans a pattern of changes in vehicle speed so that the designated speed is achieved over the travel distance instructed by the control supervisor 19.

Then, target values for the throttle opening and brake pressure are determined so that the actual vehicle speed follows the planned vehicle speed change pattern, and control commands for the throttle opening and brake pressure are given to the actuator control section 25, and Under the control, the actuator 26, which is comprised of a scooter motor, a brake solenoid valve, etc., is driven so that the actual throttle opening and brake pressure become the target values.

FIG. 19 shows an example of a block configuration for vehicle speed control, which includes a vehicle speed plan generation section 40, a throttle opening and brake pressure determination section 41, an actuator control section 42, a throttle motor 43, and a brake solenoid valve 4.
It consists of four actuators. In the diagram, 4
5 indicates a vehicle subject to vehicle speed control. The throttle motor 43 is provided with a sensor (not shown) that detects the actual throttle opening Tx according to the amount of motor drive. The brake solenoid valve 44 is provided with a sensor (not shown) that detects the actual brake pressure Bx according to the valve drive amount. In addition, vehicle 45
The actual driving ratio 11Lx, the actual vehicle speed Vx, and the actual acceleration A
Sensors (not shown) are provided to respectively detect x.

The vehicle speed plan generation unit 40 generates a specified travel distance Lr and a specified speed V.
According to each input data of r and the target route O8, a vehicle speed plan is created so that the vehicle reaches the specified speed when traveling the specified distance 1i 1-r on the target route O8.
For example, it is created by fuzzy reasoning. Then, the target vehicle speed Vs is successively set according to the created vehicle speed plan in accordance with the vehicle's actual running pressure 1iLx and the actual vehicle speed Vx.

The throttle opening degree and brake pressure determination section 41 is.

In accordance with the target vehicle speed Vs given from the medium-speed plan generator 40, the target throttle opening is set to follow the target vehicle speed Vs while looking at the running state of the vehicle obtained from the actually detected vehicle speed Vx and acceleration Ax. The brake pressure Ts and the target brake pressure Bx are determined by, for example, fuzzy reasoning.

The actuator control unit 42 controls the throttle motor while feeding back the actual throttle opening Tx and the actual brake pressure Bx so that the target throttle opening Ts and target brake pressure Bs given by the throttle opening and brake pressure determining unit 41 are achieved. 43 and brake solenoid valve 44, respectively.

FIG. 20 shows an example of the configuration of the navigation device 21. For example, it is a photoelectric type that outputs a pulse signal for each unit travel distance according to the rotation of the tires of a vehicle.

A distance sensor 46 of an electromagnetic type or a mechanical contact type,
For example, a yaw rate sensor 47 comprising a gyroscope that outputs a signal proportional to the amount of change in angular velocity in one direction as the vehicle travels.

The number of pulse signals from the distance sensor 46 is counted to measure the traveling distance of the vehicle, and the yaw rate sensor 47
A CPU, a program that determines changes in the traveling direction of the vehicle in accordance with output signals of the vehicle, determines the position on the X-Y coordinates for each unit traveling distance of the vehicle by sequential calculation, and performs centralized control of the entire navigation device. For RO
A signal processing device (computer control unit [) 48 consisting of M and control RAM, etc., sequentially stores the data of the position on the X-Y coordinates that changes every moment obtained by the signal processing device 48, and calculates the current position of the vehicle. A travel trajectory storage device that stores it as finite continuous position information corresponding to! 49
The road map information digitized in advance is file 11.
A storage medium recycling bag [! ! 51, the map surface characteristics are displayed on the screen according to the read map information, and the current position of the vehicle, the previous travel trajectory, and the current location are displayed based on the position data stored in the travel trajectory storage device 49. Display device i! that constantly updates information such as the direction of travel on the same screen!
52 and the signal processing device 48, as well as specifying the selection of the map to be displayed on the display bag 52 and setting the starting point of the vehicle on the displayed map, and An operating device i that can appropriately input guidance information for guiding the vehicle to travel according to the planned travel route! 53.

With this configuration, as shown in FIG. 21, a road map that has been selectively read out is displayed on the screen of the display device 52, and the vehicle can be seen from the starting point set on the map. Signal processing device 4 as the vehicle travels
A display mark Ml indicating the current position of the vehicle on the X-Y coordinates according to the map scale rate preset by 8, a display mark M2 indicating the traveling direction of the vehicle at the current position, and a display mark M2 indicating the direction of travel of the vehicle at the current position, A driving trajectory display mark M3 up to this point is displayed in a simulative manner following the driving state of the vehicle.

In addition, when inputting guidance information for guiding the vehicle to travel according to the planned travel route on the map displayed on the screen by inputting the operation device 53, 1
For example, as shown in Figure 22, target points a, b, e...
・It is also possible to give instructions such as turning right or left at each target point.

And a signal processing device! 48 reads the input guidance information and displays, for example, a display mark Ml of the vehicle's current position.
When comes to a certain distance in front of target point a,
It is designed to emit guidance information such as ``There is a 7-junction ahead, take the fork on the right.''

Therefore, in the automatic traveling system shown in FIG.
For example, when guidance information such as ``Go to the right branch road at the Y-junction ahead'' is given, the driveable area corresponding to the Y-junction ahead is recognized, so the right side of the Y-junction The target route is set within the area corresponding to the branch road.

Furthermore, various guidance information such as the legal speed of each road on the map is registered in the map information storage medium 50 in the navigation device [21].

Guidance information such as the legal speed on the road on which the vehicle is currently traveling is read out and given to the control supervisor 19.

Hiroka As described above, the automatic driving device according to the present invention recognizes the driveable area of the vehicle by image processing an image obtained by capturing an area in the direction of travel of the vehicle using an imaging device attached to the vehicle. means for setting a target route for the vehicle within the drivable area based on the recognition result; means for detecting the driving condition of the vehicle; and means for determining the target route for the vehicle based on the detected driving condition. In a vehicle having a means for determining a control target amount necessary for traveling on a route, and a means for controlling the vehicle to follow the target route according to the determined control target amount, in particular, the control The target is set to the vehicle's steering angle, the future travel route is predicted from the currently detected vehicle travel state, and the vehicle travels on the target route based on the deviation between the vehicle's predicted route and the target route. The steering angle correction amount is determined and the steering angle correction amount is used as the control target amount to control the running of the vehicle. It has the excellent advantage of being able to optimally perform travel control.

[Brief explanation of drawings]

FIG. 1 is a block configuration diagram showing an embodiment of an automatic driving device according to the present invention, FIG. 2 is a diagram showing road line segments obtained by data processing an image captured by a video camera, and FIG. Figure 3 is a diagram showing an image obtained by projective transformation of the image in Figure 2, Figure 4 is a diagram showing line segments on the X-Y coordinates, and Figure 5 is a diagram showing the line segments in Figure 4 by Hough
Figure 6 shows an example of a target route set on a road; Figure 7 shows points on the ρ-θ coordinates when converted;
) and (b) are diagrams showing the target routes set on the road when the vehicle is running at low speeds and at high speeds, Figure 8 is a diagram showing the relationship between the target route and the predicted route of the vehicle, and Figure 9 is a diagram showing the relationship between the target route and the predicted route of the vehicle. A diagram showing the relationship between the steering angle and its turning radius, FIG. 10 is a more detailed block configuration diagram of the automatic traveling device according to the present invention shown in FIG. 1, and FIG. 11 is a block diagram showing an example of the configuration of the data management section. figure. Figure 12 is a diagram showing the flow of processing when the data management unit reads the contents of the dual port RAM, Figure 13 is a diagram showing the flow of processing when the data management unit requests data, and Figure 14 is different. Absolute coordinate at time X-
A characteristic diagram showing the position and orientation of the vehicle on Y, FIG. 15 is a diagram showing the processing flow when correcting the image processing result to the positional relationship of the vehicle at the desired time, and FIGS. 16(a), ( b
) is a diagram showing each coordinate positional relationship when converting a point on the relative coordinates of the vehicle's position at a certain time to a point on the other relative coordinates of the vehicle's position at a different time. Fig. 17 is a diagram showing the flow of processing in the data evaluation unit, Fig. 18 is a diagram showing the flow of processing in the route control unit, Fig. 19 is a block diagram showing a configuration example of the vehicle speed control unit, and Fig. 20 is a navigation diagram. FIG. 21 is a block diagram showing an example of the configuration of the device, FIG. 21 is a diagram showing an example of a display screen of the navigation device, and FIG. 22 is a diagram showing the input state of guidance information on the planned travel route in the navigation device. 1... Imaging unit 2... Image processing unit 3... Drivable area recognition unit 4... Target route setting unit 5... Control unit 6... Vehicle speed sensor 7... Yaw rate sensor 8
... Rudder angle sensor 9 ... Steering control section IO
...Steering drive unit 11...Global line segment extraction unit 12...Local line segment tracing unit 13...Region extraction unit 14...Drivable area recognition and target route setting unit I5... Data management department 16...Data evaluation department
17...Local map management section 18...Local map database 19...Control supervisor 2
o...Radar device 21...Navigation device 2
2...Route prediction unit 23 Route control unit 24...Vehicle speed control unit AC...Predicted route OC...Target route Hatake Gan's agent Kiyoshi Torii Figure ρ Figure 6 (○) (b ) Figure 0 Figure 6 Figure 12 Figure 3 Figure 14 Figure 5 Figure 7 Figure 8 Figure 21 Figure 3 Figure 2

Claims (1)

[Claims] 1. Means for recognizing a driveable area of a vehicle by image processing an image obtained by capturing an area in the traveling direction of the vehicle with an imaging device attached to the vehicle, and the recognition result. means for setting a target route for a vehicle to travel within a drivable area based on the driving conditions; means for detecting a driving condition of the vehicle; It has means for determining a necessary control target amount, and means for controlling the running of the vehicle so as to follow the target route in accordance with the determined control target amount, and the control target is set to the steering angle of the vehicle, The future driving route is predicted from the currently detected driving state of the vehicle, and the amount of steering angle correction for the vehicle to travel on the target route is calculated from the deviation between the predicted route of the vehicle and the target route. An automatic travel device characterized in that a vehicle travel control is performed using a steering angle correction amount as the control target amount. 2. Predict the vehicle's arrival position on the X-axis after a certain distance in the Y-axis direction when the current direction of the vehicle is in the Y-axis direction,
Item 1 above, wherein a steering angle correction amount is determined in accordance with a deviation between the predicted position and a target position on the same X-axis on the target route at a certain distance ahead in the Y-axis direction. Automatic traveling device according to the description.