Method and device for detecting the course of a road for vehicles

Abstract

A road course detection method for vehicles, for calculating a lane probability, which is a probability that a detectable object is located in the same lane as a vehicle applying the method, based on a curvature of a road on which the vehicle is traveling and a position of the object relative to the vehicle, wherein the curvature is calculated from a direction change state of the vehicle and a speed of the vehicle, wherein the method comprises the following steps: - determining a probability that the object is located in the same lane as the vehicle, in accordance with a detected road course ahead of the vehicle and a degree of detection, wherein the road course is detected in accordance with a distance to the object and an angle to the object in a vehicle width direction;- Calculating a correction value, which is used to correct the lane probability, based on the determination; and - Determining whether there is a difference between the curvature of the road on which the vehicle is traveling and a curvature of the detected road path, in order to correct the lane probability using the correction value if there is no difference, or not to correct the lane probability using the correction value if there is a difference.

Description

BACKGROUND OF THE INVENTION [Technical Field of the Invention]

[0001] The present invention relates to a method and a device for detecting the course of a road on which a vehicle is driving. [State of the art]

[0002] Methods for calculating an actual lane probability value are known. The term lane probability refers to a probability that a target to be detected in front of the vehicle is present in the same lane. A lane probability actual value is determined from a curve radius of the road on which the vehicle travels to make a determination as to an object to be controlled. The turning radius is obtained from a position (distance and lateral position) of a target in front of the vehicle detected by a sensor, a turning state of the vehicle obtained based on a steering angle and a yaw rate, and a speed of the vehicle. An actual lane probability calculated in this way is corrected and subjected to, for example, a predetermined filtering process to calculate a lane probability based on which a preceding vehicle is selected.

[0003] The turning state of the vehicle usually does not match the shape of the road on which an actual object to be controlled is traveling. As a countermeasure, for example, Patent Document 1 (JP 3417375 B) discloses a method of calculating a lane probability. According to this method, the course of the road on which the vehicle travels is recognized to correct a curve radius R of the direction change of the vehicle and to correct an actual lane probability. The corrected lane probability actual value is subjected to, for example, a predetermined filtering process to calculate a lane probability based on which a preceding vehicle is selected.

[0004] The calculation according to the above prior art method is based on estimating a road edge that coincides with the shape of the road. Consequently, the method is unlikely to make a correct estimate when the vehicle is not traveling along the course of the road, such as when the vehicle is crossing an interchange, ramp, i. H. an entrance or an exit, a slow lane or the like, or when the vehicle such as a bus is going to stop at a bus stop on a highway.

[0005] For example, as in figure 11, a left-turn road on which the vehicle is running may be connected to a right-turn road. In this case, when a curve radius (radius of curvature) R calculated based on the recognized road shape is corrected for a curve radius R calculated on the vehicle side, the corrected curve radius R ends up not agreeing with the shape of the road Match the road on which the vehicle is driving. Consequently, even if the course of the road on the vehicle side is correctly recognized, the correction of the turning radius R in a situation such as that in FIG figure 11 is likely to affect proper road shape recognition.

[0006] Further, according to the method disclosed in Patent Document 1, when recognizing the shape of the road, whether an object is a moving object or a stationary object is determined. Subsequently, the stationary objects are connected and grouped for the detection of a road edge.

[0007] In the above prior art, there may be a problem that roadside objects are not detected because there is a preceding vehicle or the absolute number of roadside objects is small. As a result, a roadside is no longer calculated correctly and with a good frequency, so that the course of the road is probably no longer recognized correctly.

[0008] Further, according to the above prior art, stationary objects other than roadside objects may be determined for clustering, or a base point of clustering may be erroneous. Consequently, grouping of roadside object groups is not likely to be performed with good accuracy. As a result, a roadside is no longer calculated correctly and with a good frequency, so that the course of the road is probably no longer recognized correctly.

[0009] On the other hand, a patent document 2 (JP 3427809 B) proposes a method of detecting a road on which the vehicle is traveling. Specifically, in the method disclosed in Patent Document 2, segmented measurement data is sorted by angle to remove unnecessary data based on segment shape and to remove segments present in the vicinity of moving objects. Then, the segments effective to recognize the course of the road are grouped clockwise and counterclockwise. Furthermore, if a farthest segment appears doubtful, it is removed. Road edges are then detected based on a road edge object group (left) and a road edge object group (right).

[0010] However, according to the above prior art, the base point for grouping roadside objects may sometimes be erroneous, so that grouping of roadside objects is unlikely to be performed with good accuracy. As a result, a roadside is no longer calculated correctly and with a good frequency, so that the course of the road is probably no longer recognized correctly.

[0011] Further, in the above prior art, there may be a problem that roadside objects are not detected because there is a preceding vehicle or the absolute number of roadside objects is small. As a result, a roadside is no longer calculated correctly and with a good frequency, so that the course of the road is probably no longer recognized correctly.

[0012] Further, the shape of the road on which an actual object to be controlled is traveling is unlikely to match the turning state of the vehicle. In this regard, a patent document 3 (JP 2001-328451 A) proposes a method designed to calculate a radius of curvature of a road based on a stationary object when the stationary object exists on the road. Then, a turning radius based on the vehicle is corrected using the turning radius based on the stationary object. The corrected curve radius is used to calculate an actual lane probability, which is then used to calculate a lane probability. Then, based on the lane probability, a preceding vehicle is selected.

[0013] In this way, usually when recognizing the course of a road, stationary objects are recognized and a curve radius is calculated based on the stationary objects. However, it may sometimes be difficult to detect stationary objects because there is a vehicle ahead or the absolute number of stationary objects is small. Because of this, stationary objects do not necessarily help to accurately calculate a curve radius, so the course of the road is not necessarily calculated correctly and with a good frequency. SUMMARY OF THE INVENTION

[0014] The present invention has been made in view of the above-described problems, and an object of the present invention is to provide a road shape recognition method and a road shape recognition device for vehicles capable of calculating a road shape more correctly and at a good frequency.

[0015] In order to achieve the above object, according to a first example, the shape of the road in front of the vehicle is recognized based on a distance to an object and an angle thereof in the vehicle width direction. On the basis of the detected course of the road and the degree of detection, a probability is determined with which the object is in the same lane on which the vehicle is driving. Then, based on the results of the determination, a correction value for correcting a lane probability is calculated. Then, it is determined whether or not there is a difference between a curvature of the road on which the vehicle is running and the curvature of the detected road shape. If there is no difference, the lane probability is corrected with the correction value. If there is a difference, the lane probability is not corrected with the correction value.

[0016] Consequently, in a situation where an estimated R deviates significantly from a road shape R, a lane probability suitable for the situation is obtained. In particular, a correction of a lane probability is prevented from resulting in an error, so as to enable a more correct calculation of the road course with a good frequency.

[0017] Further, in an apparatus for realizing the road shape recognition method for vehicles, it is determined whether or not there is a difference between a curvature of the road on which the vehicle runs and the curvature of the recognized road shape. When it is determined that there is no difference, the lane probability calculated by lane probability calculation means is corrected by a correction value calculated by correction value calculation means. When it is determined that there is a difference, the lane probability calculated by the lane probability calculation means is not corrected by the correction value calculated by the correction value calculation means.

[0018] Consequently, as in the first example, the course of the road is recognized more correctly and more precisely.

[0019] For example, a computer system can realize the functions of a curve radius calculation means, an object recognition means, a lane probability calculation means, a road shape recognition means, a same-lane determination means, and a correction value calculation means. In this case, the functions can, for example, be in the form of programs launched by the computer system. Such programs may be stored on a computer-readable storage medium, such as a magneto-optical disk, CD-ROM, hard disk, or flash memory, and used by loading the programs onto the computer system as needed and launching the loaded programs. Alternatively, the programs may be stored in ROM or backup RAM as a computer-readable storage medium, and the ROM or backup RAM may be integrated into the computer system.

[0020] In order to achieve the above object, according to a second example, a transmission wave is radiated over a predetermined angular range in the vehicle width direction. When recognizing a road shape around the vehicle based on the reflected wave, the following recognition is performed. Specifically, object unit data having at least a distance to each object is acquired in accordance with angles in a vehicle width direction. At the same time, whether each object is a moving object or a stationary object is determined based on the relative speed of the object obtained based on the reflected wave and the speed of the vehicle. Then, based on the results of the object type determination, object unit data effective for recognizing the shape of the road is extracted. Then, based on the object unit data, roadside object group data is formed clockwise and counterclockwise by grouping the data with a monotonically increasing distance as a connection requirement. Then, a roadside is recognized based on the data of the roadside object group formed in this way.

[0021] It is ensured that such a series of processes is repeatedly executed in a predetermined cycle. After extracting the object unit data effective for recognizing the shape of the road, a data adding process is performed to add the object unit data extracted in the previous cycle to the object unit data extracted in the current cycle. to add. Then, a roadside is recognized based on the object unit data obtained through the data adding process.

[0022] In this way, since the object unit data of the previous cycle is added to the object unit data of the current cycle, the amount of data used for road shape recognition is increased. Consequently, the accuracy in recognizing a road edge is improved and further, a correct road shape is calculated with a good frequency in a situation where an absolute number of road edge objects that can be used for road shape recognition is small, such as when road edge objects are not detected because there is a preceding vehicle or when an absolute number of roadside objects is small.

[0023] Further, an apparatus for realizing the road shape recognition method for vehicles according to the second example comprises: data adding means ( 45 ) for adding object entity data obtained from a means ( 45 ) for extracting effective data in the previous cycle, to object unit data read by the means ( 45 ) for extracting effective data in the current cycle after the process executed by the means ( 45 ) to be executed to extract effective data has been executed. In the device, a road edge detection means ( 41 , 43 and 45 ) a roadside based on the object unit data added by the data adding means ( 45 ) can be obtained.

[0024] Consequently, like the first example, the course of the road is calculated more precisely with a frequency.

[0025] A computer system can realize the function of the recognition means of the road shape recognition device for vehicles. For example, the function can be provided in the form of a program launched by the computer system.

[0026] In order to achieve the above object, according to a third example, a transmission wave is radiated over a predetermined angular range in the vehicle width direction. When recognizing a road shape around the vehicle based on the reflected wave, the following recognition is performed. Specifically, object unit data having at least a distance to each object is acquired in accordance with angles in a vehicle width direction. At the same time, whether each object is a moving object or a stationary object is determined based on the relative speed of the object obtained based on the reflected wave and the speed of the vehicle. Then, based on the results of the object type determination, object unit data effective for recognizing the shape of the road is extracted. If a stationary object on the traveling road between the vehicle, i. H. the vehicle having the device, and a preceding vehicle, d. H. an immediately preceding vehicle, or between the immediately preceding vehicle and a vehicle in front of the immediately preceding vehicle, d. H. a second preceding vehicle, data corresponding to the stationary object on the traveling road is removed from the extracted object unit data. Then, on the basis of the object unit data, roadside object group data is formed clockwise and counterclockwise by grouping the data with a distance of monotonous increase as a connection requirement. Then, a roadside is recognized based on the data of the roadside object group thus formed.

[0027] In this way, stationary objects are removed from the object unit data, the stationary objects being positioned on a road on which the vehicle having the apparatus is running or on which an immediately preceding vehicle is running. Consequently, the roadside object group resulting from the grouping of data is approximated to an actual road shape.

[0028] Further, in the apparatus for realizing the road shape recognition method for vehicles according to the third example, when a stationary object on the traveling road between the vehicle, i. H. the vehicle having the device, and a preceding vehicle, d. H. an immediately preceding vehicle, or between the immediately preceding vehicle and a vehicle in front of the immediately preceding vehicle, d. H. a second preceding vehicle, is present, removes data corresponding to the stationary object on the traveling road from the extracted object unit data.

[0029] Consequently, like the first example, the course of the road is calculated more precisely with a good frequency.

[0030] A computer system can realize the function of the recognition means of the road shape recognition device for vehicles. For example, the function can be provided in the form of a program that is started by the computer system.

[0031] In order to achieve the above object, according to a fourth example, a transmission wave is radiated over a predetermined angular range in the vehicle width direction. When recognizing a road shape around the vehicle based on the reflected wave, the following recognition is performed. Specifically, object unit data having at least a distance to each object is acquired in accordance with angles in a vehicle width direction. At the same time, whether each object is a moving object or a stationary object is determined based on the relative speed of the object obtained based on the reflected wave and the speed of the vehicle. Then, based on the results of the object type determination, object unit data effective for recognizing the shape of the road is extracted. Subsequently, from the extracted object unit data, a lateral position of a stationary object located at a position closest to the vehicle in the vehicle width direction is extracted. At the same time, a stationary object is determined as a starting point. The stationary object in this case is located at a position a predetermined distance from the lateral position of the stationary object located closest to the vehicle in the vehicle width direction and at a position where the direct distance to the vehicle is shortest. Subsequently, data of a roadside object group is formed clockwise and counterclockwise by connecting and grouping, from the starting point, the data with a distance of monotonically increasing as a connection requirement. Then, a roadside is recognized based on the data of the roadside object group formed in this way.

[0032] Consequently, a stationary object at a position having a smallest angle with respect to the vehicle width direction and located far from the vehicle is prevented from being used as a grouping connection start point. Further, when the roadside is seen double, grouping is preferably started from an inner stationary object without using a distant stationary object to form a roadside object group. As a result, the accuracy in recognizing a road edge is improved, and further the road shape is calculated more accurately with a good frequency.

[0033] According to the fourth example, when forming data of a roadside object group, a first connection requirement area (a) and a second connection requirement area (b) are determined. The second connection requirement range (b) is smaller than the first connection requirement range and is included in the first connection requirement range. Then, starting from a start point as a base point, a stationary object included in both the first and second connection requirement areas is connected to the start point. Then, the connected stationary object is used as the subsequent base point for connection with a stationary object included in both the first and second connection requirement areas. This is repeated for grouping the stationary objects so as to form roadside object group data.

[0034] In this way, when stationary objects are grouped, the first and second connection requirement areas are determined. Consequently, a stationary object that is included in the first connection requirement range but not in the second connection requirement range is excluded from connection. Accordingly, when stationary objects are compared by angle, a stationary object having a large difference in distance but a small difference in angle is prevented from being preferentially subjected to the comparison and connection. As a result, a grouping connection is more closely approximated to an actual road shape, and further, the accuracy in road edge recognition is improved.

[0035] For example, in recognizing a roadside based on data of one roadside object group, plural data of roadside object groups are formed. Then, an intersection point between a circle which the roadside object group passes through and an axis in the vehicle width direction is calculated for each roadside object group. Subsequently, a roadside is recognized using only the roadside object groups for the intersection points positioned in a range between an intersection point located closest to the vehicle in the vehicle width direction and a point distant by a threshold distance located from the intersection is defined.

[0036] Consequently, the roadside object groups distant from the vehicle in the vehicle width direction are excluded in the detection of a roadside. Accordingly, a calculated mean road edge is recognized as road edge object groups passed on the vehicle side, so that road edge recognition accuracy is improved.

[0037] Further, in an apparatus for realizing the road shape recognition method for vehicles according to the fourth example, effects similar to those described above are brought about.

[0038] A computer system can realize the function of the recognition means of the road shape recognition device for vehicles. For example, the function can be provided in the form of a program that is started by the computer system.

[0039] In order to achieve the above object, according to a fifth example, a transmission wave is radiated over a predetermined angular range in the vehicle width direction. In recognizing a road shape around the vehicle based on the reflected wave, the following road shape recognition is performed. Specifically, object unit data having at least a distance to each object is acquired in accordance with angles in a vehicle width direction. At the same time, whether each object is a moving object or a stationary object is determined based on the relative speed of the object obtained based on the reflected wave and the speed of the vehicle. Then, based on the results of determination as to the object type, object unit data of an immediately preceding vehicle ( 181 ) and a vehicle in second place ahead ( 182 ) with respect to the vehicle having the device ( 180 ) is extracted from under moving objects, and object unit data from reflectors arranged along the road is extracted from under stationary objects. Following this, three items of object unit data, viz. H. the vehicle having the device ( 180 ), the vehicle immediately ahead ( 181 ) and the second vehicle in front ( 182 ), approximates a circle so as to calculate a radius of the circle. A road course is then detected based on the radius of the circle and a line of reflectors.

[0040] In this way, the radius of the circle is calculated using the vehicle immediately ahead ( 181 ) and the vehicle in second place ahead ( 182 ) calculated. The radius of the circle is used for road shape detection to more accurately obtain a road shape with good frequency even when, for example, it is difficult to detect reflectors or when the number of reflectors on the road is originally small.

[0041] Furthermore, an apparatus for realizing a road course recognition method for vehicles according to the fifth example has: a means ( 8 ) for extracting a preceding vehicle, which is designed to obtain object unit data of the immediately preceding vehicle ( 181 ) and the vehicle in second place ahead ( 182 ) with respect to the vehicle having the device ( 180 ) extract under the moving objects; a reflector extractant ( 108) for extracting the object unit data from reflectors arranged along the road among the stationary objects; and an approximation radius calculation means ( 117 ) to approximate the object unit data from three points, i.e. H. the vehicle having the device ( 180 ), the vehicle immediately ahead ( 181 ) and the second vehicle in front ( 182 ), to a circle to calculate the radius of the circle. In the device, the road shape recognition means ( 117 ) the course of the road based on the radius of the circle and a line of reflectors. Consequently, like the first example, the course of the road is calculated more precisely with a good frequency.

[0042] A computer system can realize the function of the recognition means of the road shape recognition device for vehicles. For example, the function can be provided in the form of a program that is started by the computer system.

[0043] In order to achieve the above object, according to a sixth example, a transmission wave is radiated over a predetermined angular range in the vehicle width direction. When recognizing a road shape around the vehicle based on the reflected wave, the following recognition is performed. Specifically, object unit data having at least a distance to each object is acquired in accordance with angles in a vehicle width direction. At the same time, whether each object is a moving object or a stationary object is determined based on the relative speed of the object obtained based on the reflected wave and the speed of the vehicle. Then, based on the results of the object type determination, object unit data of reflectors arranged along the road among stationary objects are extracted. Then, a circle passing through the line of reflectors is approximated based on the object unit data of the reflectors extracted in the extraction process. In this way, a road shape estimation curve radius for use in recognizing the road shape is calculated.

[0044] It is ensured that such a series of processes is repeatedly executed in a predetermined cycle. In extracting object unit data from reflectors among stationary objects, an adding process is performed. In the adding process, from the object unit data of the reflectors extracted in the previous cycle, certain object unit data is added to the object unit data extracted in the extracting process of the current cycle. The specific object unit data is from reflectors arranged within a predetermined range in the radial direction with respect to the road shape estimation curve radius calculated in the recognition process of the previous cycle. Then, in executing the road shape recognition, a road shape estimation curve radius is calculated based on the object unit data of reflectors obtained in the data adding process to be used in the road shape recognition.

[0045] In this way, of the object unit data of reflectors extracted in the previous cycle, those of reflectors located within a predetermined range with respect to the road course estimated curve radius calculated in the previous cycle become the data of the current cycle added to improve the occurrence frequency of reflectors. Further, since the data of reflectors of the previous cycle located within the predetermined range is selected, data of reflectors reflecting the road shape can be used. Consequently, the road shape is calculated more accurately with a good frequency even when it is difficult to detect reflectors because there is a vehicle ahead or when the number of reflectors on the road is originally small.

[0046] Further, in an apparatus for realizing the road shape recognition method for vehicles according to the sixth example, a data adding means ( 108 ) which separates the object unit data of reflectors located within a predetermined range in the radial direction with respect to the road shape estimation curve radius calculated in the previous cycle from the object unit data of reflectors extracted in the previous cycle. is added to the object unit data of reflectors extracted in the current cycle. Further, in the device, the road shape recognition means ( 117 ) designed to calculate the road course estimated curve radius based on the object unit data of reflectors obtained by the data adding means ( 108 ) to be used in road shape recognition. Consequently, the course of the road is calculated more accurately with a good frequency.

[0047] A computer system can realize the function of the recognition means of the road shape recognition device for vehicles. For example, the function can be provided in the form of a program that is started by the computer system.

[0048] It should be noted that the reference numbers in parentheses in this section and in the claims indicate a correspondence to the specific means in the embodiments described below. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] In the attached drawings shows:

[0050] figure 1 is a block diagram showing a configuration of a vehicle control device according to a first embodiment of the present invention;

[0051] figure FIG. 2 is an explanatory diagram showing an outline of a preceding vehicle selecting process; FIG.

[0052] figure 3 is an explanatory diagram showing how to convert target positions into straight road traveling positions;

[0053] figure 4 is an explanatory diagram showing a lane probability map;

[0054] figure 5A is an explanatory map showing an estimated X-axis intersection point;

[0055] figure 5B is an explanatory diagram showing road edge detection;

[0056] figure 6A is an explanatory diagram showing a determination process when roadsides are recognized farther than a target;

[0057] figure 6B is an explanatory diagram showing a determination process in the case where roadsides covering only a distance shorter than the destination are detected;

[0058] figure 7A is an explanatory diagram showing a determination process based on a distance between each target position used for road edge recognition and a target vehicle curve;

[0059] figure 7B is an explanatory diagram showing areas in the vicinity of roadsides;

[0060] figure Fig. 8A is a diagram showing an intersection area such as an exit of a freeway triangle of a freeway;

[0061] figure 8B is a diagram showing a lane change;

[0062] figure Fig. 9 is a diagram showing a relationship between segment information and its requirement;

[0063] figure 10 is an explanatory diagram showing a map of a parameter α for calculating a lane probability;

[0064] figure Fig. 11 is a diagram showing problems in the prior art;

[0065] figure 12 is a block diagram showing a configuration of a vehicle control device according to a second embodiment of the present invention;

[0066] figure 13 is an explanatory diagram showing an outline of a road shape recognition process;

[0067] figure 14A is an explanatory diagram showing how to segment measurement data;

[0068] figure 14B is an explanatory diagram showing a grouping of segment data;

[0069] figure Fig. 15 is an explanatory diagram showing how to deal with a farthest segment in a roadside object group (left);

[0070] figure Fig. 16 is an exemplary diagram showing how to deal with farthest segments that are co-covered by a roadside object group (left) and a roadside object group (right);

[0071] figure Figure 17 is an exemplary diagram illustrating how to recognize a roadside as a collection of line segments;

[0072] figure 18 is an explanatory diagram showing an outline of a road shape recognition process according to a third embodiment of the present invention;

[0073] figure 19 is an explanatory diagram showing stationary objects on a traveling road;

[0074] figure 20 is an explanatory diagram showing an outline of a road shape recognition process according to a fourth embodiment of the present invention;

[0075] figure 21A is an explanatory diagram showing how to segment measurement data;

[0076] figure 21B is an explanatory diagram showing a grouping of segment data;

[0077] figure Fig. 22 is an explanatory diagram showing how to connect stationary objects one by one starting from a stationary object at a starting point;

[0078] figure Fig. 23 is an explanatory diagram showing how to deal with a farthest segment in a roadside object group (left);

[0079] figure Figure 24 is an exemplary diagram illustrating how to deal with farthest segments that are overlapped by a roadside object group (left) and a roadside object group (right);

[0080] figure Figure 25 is an exemplary diagram illustrating how to recognize a roadside as a collection of line segments;

[0081] figure 26A is an explanatory diagram showing how to recognize a roadside using all segments according to a fifth embodiment of the present invention;

[0082] figure 26B is an explanatory diagram showing how to detect a roadside using segments equal to or less than a threshold according to the fifth embodiment;

[0083] figure 27 is a system configuration diagram showing an inter-vehicle control device to which a road shape recognition device according to a sixth embodiment of the present invention is applied;

[0084] figure 28 is a flowchart showing a road shape recognition;

[0085] figure 29 is an explanatory diagram showing how to calculate an approximation R based on the device-equipped vehicle, a preceding vehicle, and a vehicle in front of the preceding vehicle;

[0086] figure 30 shows a flow chart for illustrating a road course recognition;

[0087] figure Fig. 31 is an explanatory diagram showing how to convert a curve radius into a straight road;

[0088] figure Fig. 32 is an exemplary diagram showing how to calculate a curve radius from a delineator;

[0089] figure Fig. 33 is an explanatory diagram showing how to recognize both edges of a lane;

[0090] figure34 is a flowchart showing road shape recognition performed by an inter-vehicle control device to which a road shape recognition device according to a seventh embodiment of the present invention is applied;

[0091] figure 35 is a flowchart showing road shape recognition performed by an inter-vehicle control device to which a road shape recognition device according to an eighth embodiment of the present invention is applied;

[0092] figure 36 is a flowchart showing road shape recognition performed by an inter-vehicle control device to which a road shape recognition device according to a ninth embodiment of the present invention is applied;

[0093] figure 37 is a flowchart showing road shape recognition performed by an inter-vehicle control device to which a road shape recognition device according to a tenth embodiment of the present invention is applied; and

[0094] figure 38 is an explanatory diagram showing how to use stationary object data in the vicinity of a road shape estimation curve radius R calculated in the previous cycle. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0095] Various embodiments of a road shape recognition method and a road shape recognition device for vehicles of the present invention will be described below with reference to the accompanying drawings. [First embodiment]

[0096] A road shape recognition method and a road shape recognition device for vehicles according to a first embodiment of the present invention will be described with reference to FIG figure 1 to figure 10 described.

[0097] figure 1 shows a structure of a vehicle control device 1 to which the road shape recognition device for vehicles is applied. The control device 1 is installed in an automobile, and issues an alarm when obstacles exist in a predetermined situation in an area requiring issuing of an alarm, or controls the vehicle speed in accordance with a preceding vehicle.

[0098] figure 1 shows a system block diagram of the device 1 . The vehicle control device 1 is essentially a computer 3 built up. The computer 3 is basically composed of a microcomputer and has an input / output interface (I / O) and various driving circuits and detection circuits. Since these hardware components are of known type, they will not be described in detail below. Consequently, the computer 3 a CPU (main processor) 3C as an arithmetic unit and a memory 3M as a storage medium.

[0099] The CPU 3C reads various processing programs, described below, and in the memory 3M are saved and executes them. Consequently, the computer cooperates 3 with its peripheral devices to perform different functions included in the functional blocks in the figure 1 are shown. The computer 3 is configured in the same manner in the following embodiments and modifications described later. The memory as such a storage medium may be a magneto-optical disk, a CD-ROM, a hard disk, a flash memory or the like. A ROM (Read Only Memory) or a back-up RAM (Random Access Memory) can be used as the memory.

[0100] The computer 3 receives input signals of various predetermined detection data from a distance / angle measuring device 5 that serves as an obstacle detection unit for a vehicle, a vehicle speed sensor 7 , a brake switch 9 and a throttle position sensor 11. Furthermore, the computer 3 predetermined control signals to an alarm tone generation unit 13 , a distance indicator 15 , a sensor error indicator 17 , a brake control unit 19 , a throttle control unit 21 and an automatic transmission controller 23 .

[0101] The computer 3 comprises: an alarm volume adjustment unit 24 , which adjusts the volume of the alarm, a sensitivity adjustment unit 25 , a cruise control switch 26 , a steering angle sensor 27 that detects an operation amount of a steering wheel (not shown), and a yaw rate sensor 28 . The computer 3 also has a circuit breaker 29 on and starts predetermined processes when the circuit breaker 29 is switched on.

[0102] The distance / angle measuring device 5 has a transmitter / receiver 5a and a distance / angle calculation unit 5b on. The sender / receiver 5a intermittently outputs a laser beam to scan in the forward direction of the vehicle with a predetermined optical axis (central axis) serving as a center and covering a predetermined angular range in the vehicle width direction, and detects reflected light. The distance / angle unit of account 5b detects a distance r to an object in front of the vehicle based on the time required to detect the reflected light. As an alternative to a laser beam, a radio wave such as a millimeter wave or an ultrasonic wave can be used. Further, the way of scanning is not limited to the scanning with the transmitter, but it may be the scanning with the receiver.

[0103] According to this configuration, the computer leads 3 an alarming process to issue an alarm such as when an obstacle is present in a predetermined alarm area for a predetermined time. Such obstacles include a preceding vehicle running or stopping in front of the vehicle, or objects (such as guard rails or posts) at a roadside. At the same time, the computer gives 3 Control signals to the brake control unit 19 , the throttle control unit 21 and the automatic transmission controller 23 to carry out a so-called vehicle-to-vehicle control under which the vehicle speed is controlled in accordance with the conditions of a preceding vehicle.

[0104] An internal structure of the computer is shown below 3 described in terms of its functional control blocks. In particular, the distance / angle calculation unit 5b the distance / angle measuring device 5 data on the distance r and a scanning angle θ. The output data is sent to a coordinate conversion block 41 sent to convert from a polar coordinate to an orthogonal coordinate. Consequently, the coordinate conversion block converts 41 the transmitted data into an orthogonal coordinate at which the center of the laser radar is an origin (0, 0), the vehicle width direction is an X axis, and the vehicle length direction is a Z axis. The converted data is then sent to an object detection block 43 and a road course recognition block 45 sent.

[0105] The object detection block 43 calculates a center position (X, Z) and a size (W, D) of an object based on the measurement data converted into an orthogonal coordinate. The object detection block 43 calculates a relative speed (Vx, Vz) of an obstacle such as a preceding vehicle with respect to the position of the vehicle equipped with the device 1 is equipped, d. H. of the vehicle having the device, based on the temporal changes of the center position (X, Z). Furthermore, the object recognition block recognizes 43 an object type, d. H. whether an object is a stationary object or a moving object. When detecting an object type, the object detection block uses 43a vehicle speed (speed of the vehicle having the device) V calculated by a vehicle speed calculation block 47 is output in accordance with a detection value of the vehicle speed sensor 7 , and also the relative velocity (Vx, Vz) calculated as described above. Then the object detection block dials 43 an object that will affect driving of the vehicle based on the object types each detected as described above and the center positions of the objects, and the object detection block shows 43 the distance to the selected object on a distance indicator 15 on. (W, D), indicating the size of an object, corresponds to (width, depth). A model of an object with such data is referred to as a "target model".

[0106] Then a sensor error detection block detects 44 whether those in the object detection block 43 calculated data has a value that falls within an erroneous range. If the value falls within the error range, the sensor error display shows 17 this accordingly. On the other hand, the road course recognition block recognizes 45 the course of the road based on the measurement data converted into an orthogonal coordinate and the data stored in the object recognition block 43 be calculated. The details of the recognition process of a road shape will be described later. Those in the road course recognition block 45 received data are attached to a block 53 given for determining a preceding vehicle.

[0107] The computer 3 further comprises: a steering angle calculation block 49 , which calculates a steering angle based on a signal from the steering angle sensor 27 calculated, and a yaw rate calculation block 51 , which calculates a yaw rate based on a signal from the yaw rate sensor 28 calculated. Also calculated in the computer 3 , a curve radius (radius of curvature) calculation block 63 a turning radius (radius of curvature) R based on a vehicle speed from the vehicle speed calculation block 47 , a steering angle from the steering angle calculation block 49 and a yaw rate from the yaw rate calculation block 51 . The block 53 to determine a preceding vehicle, selects a preceding vehicle and calculates a distance Z to the preceding vehicle and a relative speed Vz with respect to the preceding vehicle. The choice of a preceding vehicle is based on the turning radius R as well as the object type, the center position coordinate (X, Z), the size (W, D) of the vehicle and the relative speed (Vx, Vz) defined in the object detection block 43 are calculated, and the road shape data stored in the road shape recognition block 45 be obtained.

[0108] The computer 3 further comprises a vehicle-to-vehicle control and alarm output block 55 which makes an alarm determination or a running determination. The alarm or running determination is made based on the distance Z, the relative speed Vz, the device vehicle speed Vn, the preceding vehicle speed, the object center position, the object width, the object type, the setting conditions of the cruise control switch 26 and the degree of braking applied to the brake switch 9 is exerted, and the throttle position, which is from the throttle position sensor 11 is obtained and the sensitivity value obtained from the alarm sensitivity setting unit 25 is determined. When an alarm determination is made, it is further determined whether or not an alarm should be issued. When a driving determination is made, the details of the vehicle cruise control are determined. As a result of the determination, when an alarm is to be issued, the block gives 55 an alarm generation signal to the alarm sound generation unit 13 out. When a driving determination is made, the block returns 55Control signals to the automatic transmission controller 23 , the brake control unit 19 and the throttle drive unit 21 to perform controls as required. When executing these controls, the block returns 55 a required indication signal to the distance indicator 15 to inform the driver of the situation.

[0109] An operation involved in recognizing a road shape performed by the vehicle control device will be described below 1 is executed, which is configured as described above, with reference to the in the figure 2 shown flowchart described. In step S1000, which is an initial step in the figure is 2, distance / angle measurement data, i. H. Distance / angle data, from distance / angle measuring device 5 had read. In particular, the distance / angle measuring device detects 5 Distance / angle data corresponding to one scan. In this case, the sampling cycle is 100 ms, so the distance / angle measuring device 5 distance / angle data collected every 100 ms.

[0110] Then, in subsequent step S2000, the coordinate conversion block converts 41 the distance / angle data in a polar coordinate system to that in an X-Z orthogonal coordinate system. Then the object detection block detects 43 an object based on the converted data. The details of this object detection correspond to the details described above. An object recognized here is referred to as a target or target model.

[0111] In step S3000, an estimated R (estimated curve radius of the curve the vehicle is turning) based on a yaw rate detected by the yaw rate sensor 28 is obtained, or a steering angle obtained from the steering angle sensor 27 is obtained is calculated. Herein, an estimated R is calculated from the steering angle. More specifically, an estimated R is calculated as follows: Estimated R = constant ÷ steering angle

[0112] The “constant” herein depends on the vehicle speed and the vehicle model. Values ​​of the constants are described as mapping functions based on the vehicle speed for each of vehicle models and in the curve radius (radius of curvature) calculation block 63 of the computer 3 saved. Since such a function C is known as a function for calculating a turning radius from a steering angle θ, no detailed explanation will be given below. An estimated R is calculated based on a yaw rate Ω, i.e. H. calculated by dividing a vehicle speed V by a yaw rate Ω.

[0113] In step S4000, a lane probability actual value is calculated for a target recognized in step S2000. The lane probability is a parameter indicating a probability with which the target is a vehicle traveling in the same lane as the device-equipped vehicle. A lane probability actual value is calculated based on detection data of the moment.

[0114] First, positions of all targets obtained through the object recognition process (step S2000) are converted into positions related to straight road driving. When the center position of each target is (Xo, Zo) and the width in the X-axis direction is Wo, a conversion position (X, Z, W) for a straight road is obtained via the following conversion equations (see figure 3). X ← Xo - Zo^2 / 2R (1) Z ← Zo (2) W ← Where (3) R: Estimated R Right curve: positive sign Left curve: Negative sign

[0115] The symbol ^ in equation (1) refers to a value preceding the symbol ^ being raised to the power of the number of times indicated by a value following the symbol ^. The ^ symbol has the same meaning whenever used in the description. A circle equation is herein approximated by assuming that: |X| << |R|, Z

[0116] Furthermore, when the distance / angle measuring device5 is fixed at a position away from the center of the vehicle, corrects the X coordinate such that the center of the vehicle will be an origin. To be more precise, only the X coordinate is actually converted.

[0117] Each center position (X, Z) obtained by the straight road conversion is placed on a lane probability map shown in FIG figure 4 to calculate an actual lane probability of each object, i. H. a probability with which each object is then present in the lane traveled by the vehicle. There is an error between a turning radius (radius of curvature) R calculated from a steering angle and an actual turning radius. Herein, in order to realize control considering the error, an actual lane probability of each object is calculated as a presence probability.

[0118] In the figure 4, the horizontal axis is an X-axis, i. H. the left-right direction of the vehicle, and the vertical axis a Z-axis, i. H. the forward direction of the vehicle. In the present embodiment, an area covering 5 m to both left and right of the vehicle and 100 m ahead of the vehicle is in the figure 4 shown. Here, the area is divided into an area a (lane probability: 80%), an area b (lane probability: 60%), an area c (lane probability: 30%), an area d (lane probability: 100%) and another area (lane probability : 0%). This setting is based on actual measurement data. Specifically, the area d is determined considering that another vehicle may run right in front of the vehicle.

[0119] Lines La, Lb, Lc and Ld dividing the regions a, b, c and d are obtained via the following equations (4) to (7), for example. It should be noted that lines La', Lb', Lc', and Ld' are symmetrical to lines La, Lb, Lc, and Ld with respect to the Y-axis, respectively. La: X = 0.7 + (1.75 - 0.7) * (Z / 100)^2 (4) Lb: X = 0.7 + (3.5 - 0.7) * (Z / 100)^2 (5) Lc: X = 1.0 + (5.0 - 1.0) * (Z / 100)^2 (6) Ld: X = 1.5 * (1 - Z / 60) (7)

[0120] These equations are described by the following general equations (8) to (11). La: X = A1 + B1 (Z / C1)^2 (8) Lb: X = A2 + B2 (Z / C2)^2 (9) Lc: X = A3 + B3 (Z / C3)^2 (10) Ld: X = A4 * (B4 - Z / C4) (11)

[0121] Usually, ranges are determined based on equations (8) to (11) to satisfy the following equations (12) to (14). Values ​​actually used are determined through experimentation. A1 ≤ A2 ≤ A3 < A4 (12) B1 ≤ B2 ≤ B3 and B4 = 1 (13) C1 = C2 = C3 (C4 has no limit) (14)

[0122] The lines La, Lb, Lc, La', Lb', and Lc' are represented as parabolas in consideration of the calculation processing speed, but can be better represented as circular arcs if the calculation processing speed allows. Also, the lines Ld and Ld' can be better represented as parabolas or arcs of circles curved outward if the calculation processing speed allows.

[0123] Then, the position of each target resulting from the straight-road conversion is compared with that in the figure 4 are compared with the lane probability map shown. The comparison with the map is made based on the following points so as to obtain an actual value P0 of the lane probability. Object, which also slightly occupies the area d → P0 = 100% Object with the center in the area a - P0 = 80% Object with the center in the range b → P0 = 60% Object centered in the range c → P0 = 30% Object that does not meet any of the above conditions → P0 = 0%

[0124] In step S5000, a road shape is recognized based on target data that appear to be delineators placed at a roadside.

[0125] First, of the targets, those targets which are stationary objects in terms of an object type and have a horizontal width W of less than 1 m are extracted. Consequently, most of the vehicles, signs, billboards and the like will be removed. An intersection point with the X-axis is then estimated for each of the extracted stationary targets. In calculating such an estimated X-axis intercept for each stationary target, a circle passing the center of the target and having a relative velocity vector as a tangent vector is calculated. Assuming that the center of the circle is on the X-axis, the circle is perpendicular to the X-axis and consequently the radius R is uniquely determined. In practice, the following approximation is made.

[0126] When approximating a circle to a parabola, assuming that |X| << |R|, Z, an equation of a circle passing the center of the target and perpendicular to the X-axis is described as follows. X = Xo + (Z - Zo)^2 / 2R (15)

[0127] Further, since the relative velocity vector of the target is the tangent vector of the circle, the equation (15) is described as follows. dX / dZ = Vx / Vz (16)

[0128] Based on these two equations, a radius R is described as follows (see figure 5A). R = (Z-Zo)*Vz / Vx

[0129] In this case, when Z=0, the following equation can be established. X = Xo - Zo * Vx / 2Vz

[0130] Consequently, an estimated X-axis intercept is calculated as follows.

[0131] Estimated X-axis intercept = Xo - Zo*Vx / 2Vz

[0132] In this way, estimated X-axis intersection points are calculated for all of the stationary targets. The resultant values ​​are then divided into positive sign values ​​and negative sign values ​​and statistically processed at each division as follows. First, the estimated x-axis intersection points from all of the stationary targets are simply averaged to get a preliminary average. Then, those data that deviate by 2 m or more from the preliminary mean are all removed and the remaining data are averaged again. The removed data is not used for the road course recognition.

[0133] The reasons for such a process are as follows. For example, if the data includes data on billboards located high above the ground and this data is not removed, the billboards being different from the delineators, the course of the road may be misrecognized. In this regard, the averaging process described above is designed to remove data that has a significant offset from the positions where delineators should be present. Consequently, the shape of the road is recognized with good accuracy.

[0134] Subsequently, as in figure 5B, the remaining stationary targets are interpolated connected for both the right and left sides of the road to detect the roadsides. Further, of the targets forming a roadside, the one closest to the vehicle or located at the shortest distance from the vehicle (smallest Z) is selected for both the left and right roadsides such that the estimated X Axis intersection of the selected target is used as the intersection between the roadside and the X-axis. The recognized road edges are specified in a road edge coordinate table. The roadside coordinate table is prepared for both the left roadside and the right roadside. A roadside X-coordinate value for every 5 meters is stored in each table. The distance ranges from 0 m to 150 m. Distances to the respective targets that form each roadside are rounded down to the nearest 5 meter unit and determined in the relevant table. If there is no relevant data, the table is left blank.

[0135] As a result of the road shape recognition described above, a road shape R different from the estimated R obtained in step S3000 is obtained.

[0136] In step S6000, it is determined whether individual targets are vehicles traveling on the traffic lane on which the host vehicle is traveling, based on the road shape recognized in step S5000. Then, according to the determination, a “lane probability actual correction value” is calculated. First, basic determinations are made for each target as to whether the target is a vehicle running on the traffic lane on which the vehicle having the device is running. The basic provisions are as follows. [Basic provision 1]

[0137] This is a determination made when a roadside is detected farther than a target. This determination is made for both the left side and the right side of the road. (a) Left side of the road

[0138] In the figure 6A: If ZMAX_Z ≥ and |ΔXZ = Zo - ΔXZ = 0| < 1.2m: ♦ Basic Determination 1 (Left) Result ← 1 If ZMAX ≥ Zo and|ΔXZ = Zo – ΔXZ = 0| ≥ 2.0m: ♦ Basic Determination 1 (Left) Result ← –1 Else: ♦ Basic Provision 1 (Left) Result ← 0 (b) Right side of the road Just left side of the street: If Z_MAX ≥ Zo and |ΔXZ = Zo - ΔXZ = 0| < 1.2m: ♦ Basic determination 1 (right) Result ← 1 If Z_MAX ≥ Zo and |ΔXZ = Zo - ΔXZ = 0| ≥ 2.0m: ♦ Basic Determination 1 (Right) Result ← –1 Else: ♦ Basic determination 1 (right) Result ← 0

[0139] When the basic determination 1 returns “1”, it is determined with a high probability that the target is a preceding vehicle traveling in the same lane, and when the basic determination 1 returns “−1”, it is determined with a high probability that the target is a vehicle traveling in a different lane or a roadside object. When the basic determination 1 is “0”, it becomes difficult to determine whether the target is present on the same lane or on another lane, or otherwise no roadside is recognized. [Basic provision 2]

[0140] This determination is made when a roadside is not recognized up to the point of a destination. This determination is made for both the left and the right side of the road. (a) Left side of the road

[0141] In the figure 6B: If |ΔXZ = Z#MAX - ΔXZ = 0| <1.2m*(Z#MAX / Zo)^2; or If |ΔXZ = Z#MAX - ΔXZ = 0| < 0.3m: ♦ basic determination 2 (left) Result ← 1 If |ΔXZ = Z#MAX - ΔXZ = 0| ≥ 2.0m*(Z#MAX / Zo)^2; and If |ΔXZ = Z#MAX - ΔXZ = 0| ≥ 0.3m: ♦ Basic Determination 2 (Left) Result ← –1 Otherwise: ♦ Basic determination 2 (left) Result ← 0 If Z_MAX > Zo / 2: ♢ Basic determination 2 (left) reliability ← 1 (high) If Z_MAX ≤ Zo / 2: ♢ Basic determination 2 (left) Reliability ← –1 (low) (b) Right side of the road

[0142] Just down the left side of the road, the results and reliabilities of basic determination 2 (right) are calculated.

[0143] In the figure 6B, a vehicle target curve corresponds to a curve connecting a target and the origin via a segment of a circle perpendicular to the X-axis. The circle equation is approximated to a parabola using the following equation on the assumption that |X| << |R|, Z X = Z^2 / R (R: radius)

[0144] How from the figure6B, the determination is based on a distance (ΔXZ = Z#MAX) between a farthest detection point (distance Z_MAX) and a point where a line parallel to the X-axis is extended from the farthest detection point , which intersects the vehicle target curve. Due to the approximation to a parabola, the determination values ​​1.2 m and 2.0 m of basic determination 1 are multiplied by (Z#MAX / Zo)^2.

[0145] When the basic determination 2 results in “1”, it is determined with a high probability that the target is a preceding vehicle traveling in the same lane, and when the basic determination 2 results in “−1”, it is determined with a high probability that the target is a vehicle traveling in a different lane or a roadside object. In basic determination 2, the reliability of the determination is presented in two levels. When the basic determination 2 is “0”, it becomes difficult to determine whether the target is traveling in the same lane or in a different lane, or otherwise no roadside is recognized. [Basic Provision 3]

[0146] This determination is made based on a distance other than Z=Zo, Z_MAX. This determination is made for both the left and the right side of the road. (a) Left side of the road

[0147] The following two determinations are made. [Provision 3a]

[0148] In the figure 7A, all positive numbers i satisfying i*dZ≦Zo (dZ=5m) are determined as Z_MAX→i*dZ and becomes the same determination as the basic determination 2 (left) Result = 1 hit.

[0149] If all i's satisfy the condition of basic determination 2 (left) Result = 1: ♦ Basic determination 3a (left) Result ← 1 If there are one or more i's that do not meet the condition: ♦ Basic determination 3a (left) Result ← -1 If there is no i to determine: ♦ Basic determination 3a (left) Result ← 0 [Provision 3b]

[0150] In the figure 7A, all positive numbers i satisfying i·dZ≦Zo (dZ=5m) are determined as Z_MAX→i·dZ, and the same determination as the basic determination 2 (left) result = −1 is made.

[0151] If all i's satisfy the condition of basic determination 2 (left) result = -1: ♦ Basic determination 3b (left) Result ← 1 If there are one or more i's that do not meet the condition: ♦ Basic determination 3b (left) Result ← -1 If there is no i to determine: ♦ Basic determination 3b (left) Result ← 0 (b) Right side of the road

[0152] The results of the basic determination 3a (right) and the basic determination 3b (right) are calculated in a manner similar to that of the left side of the road.

[0153] If the basic determination 3a is “1”, using roadside data of any distance will result in a determination that the target is a preceding vehicle on the same lane, and if the basic determination 3a is “−1”, the target will not be unconditionally determined as a preceding vehicle in the same lane depending on the distance. When the basic determination 3a is “0”, there is no roadside coordinate data at a closer distance than the destination.

[0154] On the other hand, if the basic determination 3b is "1", the use of roadside data of any distance will result in a determination that the target is a vehicle on a different lane or a roadside object, and if the basic determination 3b is "- 1” does not necessarily determine the target as a vehicle in a different lane or as an object on a roadside. When the basic determination 3b is “0”, there is no roadside coordinate data at a distance shorter than the destination.

[0155] Based on the results of the three basic determinations determined as described above, the following six classifications are provided. A correction value of a lane probability actual value is calculated in accordance with the six classifications. If multiple conditions are met, a higher priority actual value is used. [First Classification]

[0156] When a roadside is detected farther than a target and the target is determined as a preceding vehicle in the same lane: On the left side of the road: If basic determination 1 (left) result = 1 and basic determination 3a (left) = 1; Correction value ← 40% Priority level: 5 If basic determination 1 (left) result = 1 and basic determination 3a (left) = -1; Correction value ← 0% Priority level: 3 If basic determination 1 (left) result = 1 and basic determination 3a (left) = 0; Correction value ← 40% Priority level: 2

[0157] On the right side of the road, the correction values ​​are calculated in a manner similar to the left side of the road. [Second Classification]

[0158] When a roadside is detected farther than a target and the target is determined as a vehicle in a different lane or a roadside object: On the left side of the street: If basic determination 1 (left) result = -1 and basic determination 3a (left) = 1; Correction Value ← -40% Priority Level: 5 If basic determination 1 (left) result = -1 and basic determination 3a (left) = -1; Correction value ← 0% Priority level: 3 If basic determination 1 (left) result = -1 and basic determination 3a (left) = 0; Correction Value ← -40% Priority Level: 2

[0159] On the right side of the road, the correction values ​​are calculated in a manner similar to the left side of the road. [Third Classification]

[0160] When a roadside is not detected up to the position of a target and the target is determined as a vehicle ahead in the same lane: On the left side of the street: If basic determination 2 (left) result = 1 and basic determination 3a (left) = 1; Correction Value ← 40% Priority Level: 1 If basic determination 2 (left) result = 1 and basic determination 3a (left) = -1; Correction value 0% Priority level: 1 If basic determination 2 (left) result = 1 and basic determination 3a (left) = 0 and basic determination 2 (left) reliability = 1; Correction value 40% priority level: 1 If basic determination 2 (left) result = 1 and basic determination 3a (left) = 0 and basic determination 2 (left) reliability = -1; Correction value 20% Priority level: 1

[0161] On the right side of the road, the correction values ​​are calculated in a manner similar to the left side of the road. [Fourth Classification]

[0162] When a roadside is not detected up to the position of a target and the target is determined as a vehicle on a different lane or a roadside object: On the left side of the street: If basic determination 2 (left) result = -1 and basic determination 3a (left) = 1; Correction Value ← 40% Priority Level: 1 If basic determination 2 (left) result = -1 and basic determination 3a (left) = -1; Correction value ← 0% Priority level: 1 If basic determination 2 (left) result = -1 and basic determination 3a (left) = 0 and basic determination 2 (left) reliability = 1; Correction Value ← -40% Priority Level: 1 If basic determination 2 (left) result = -1 and basic determination 3a (left) = 0 and basic determination 2 (left) reliability = -1; Correction Value ← –20% Priority Level: 1

[0163] On the right side of the road, the correction values ​​are calculated in a manner similar to the left side of the road. [Fifth Classification]

[0164] If none of the above conditions are met, for example because a roadside is not detected from either the left side or the right side of the road: Correction value ← 0% Priority level: 0 [Sixth Classification]

[0165] When a destination is determined as a roadside object: The target is determined as an object on a roadside on condition that the center of the target falls within areas included in the figure 7B are shown, i. H. Areas extending 0.5 m to the left and to the right of a detected roadside (left roadside or right roadside). In this case, the following correction value and priority level are determined. Correction Value ← -70% Priority Level: 6

[0166] As described above, a correction value (Ph) of a current value is calculated in accordance with the recognized road shape. The results of the calculation are summarized as follows.

[0167] When a road shape of only a short distance is recognized, a small correction value is calculated (compared to when a road shape that goes farther is recognized) (see correction values, which include the results of basic determination 2). When a road shape is recognized from only a short distance, a determination is made assuming a circle (arc) connecting the object and the position of the vehicle. Therefore, considering that there are assumed elements, a correction value may preferably be relatively low.

[0168] When a target is determined as a roadside object, a larger correction value is calculated (see sixth classification). When the target is determined to be a roadside object, there is a high possibility that it exists outside the lane on which the vehicle is traveling. Consequently, the correction value is determined as a large negative value of -70%. Accordingly, in the case where the lane probability actual value is high, correcting the actual value can reduce the probability so that erroneous selection is prevented. Such processing is particularly effective, for example, when “the vehicle is running on a straight road and the road ahead of the vehicle turns into a curve”.

[0169] The priority level is used as follows. In the present embodiment, a determination can be made with reference to either the left or the right roadside. However, the degree of recognition differs between the left and right roadsides. Consequently, the probability that an object is present in the same lane may differ between the case where the left side of the road is used as a reference and the case where the right side of the road is used as a reference. In such a case, a correction value is calculated based on the determination results showing a higher priority level. According to the examples described above, the priority levels in the first and second classifications, in which the shape of the road is recognized farther than the object, are 5, 3, and 2. In the third and fourth classification, in which the shape of the road is recognized only in is detected at a distance in front of the object, the priority level is 1. Consequently, the results of the determination based on the first and second classifications show a higher priority.

[0170] The classifications set forth above are based on combining the results of basic determination 3 with basic determination 1 or 2. In this way, the priority level is determined comprehensively considering the entire road shape and the positions of the objects used in the road shape recognition .

[0171] In step S6100, it is determined whether or not there is a large difference between the estimated R calculated in step S3000 and the road shape R calculated in step S5000. The reason for this determination is that when the difference between the estimated R and the road shape R is large and the actual value is consequently corrected, the accuracy of the actual value may deteriorate.

[0172] The estimated R can be, for example, in an intersection area such as an exit of a highway, as in figure 8A, or at the time of a lane change, as in figure 8B, clearly deviate from the road course R. In such a situation, applying a correction value Ph to an actual lane probability may deteriorate the actual lane probability.

[0173] Consequently, a road shape, as indicated by the corrected estimated R in the figure 8B are inevitably recognized. In this case, a protection process is added. In the protection process, when the estimated R deviates from the road shape R significantly, the lane probability is not corrected. More specifically, the protection process is executed when it is determined that “there is a large difference”.

[0174] Specifically, a determination as to “whether there is a large difference” is made as follows. figure 9 shows segment information of this time and the conditions in a table prepared in advance. Segment information includes vehicle speed, estimated R, and road course R figure Numerical values ​​shown in Fig. 9 are only examples, and other numeric values ​​can be set as a matter of course. 1) When conditions a and 0 are satisfied, determinations are made in the order of priority i) → iv). i) Correction value based on road shape recognition is determined as Ph = 0% when conditions h and d2 and (condition (1) or (2) or (3)) are satisfied, where condition (1) (conditions b and g) or (conditions c and f); condition (2) is (conditions d and j and k) or (conditions e and i and 1); and condition (3) is condition m or n. ii) The following A → B is applied when conditions t and d2 and b2 are satisfied. A: When condition v and {condition x or u or q} are satisfied, a correction value based on road shape recognition is determined as Ph=0%. B: When condition v and {condition x or u or q} are not satisfied, a correction value Ph calculated in road shape recognition is used. iii) When conditions s and w and a2 and {condition p or r or v or c2} are satisfied, a correction value based on road shape recognition is determined as Ph=0%. iv) A correction value Ph calculated in step S6000 is used. 2) When the conditions of item 1) are not satisfied, a correction value Ph calculated in step S6000 is used.

[0175] In item 1) above, i) to iii) correspond to the case where “there is a large difference”. In this case, processing proceeds to step S7000. Specifically, since the correction value Ph = 0%, the current lane probability value is not corrected in subsequent step S6200.

[0176] On the other hand, the items 1) iv) and 2) correspond to the case that "there is not a large difference". In this case, processing proceeds to step S6200.

[0177] In the situation as they are in the figure 8A or figure 8B, the conditions a, b and c, which are shown in, for example, FIG figure 9 are shown as follows: a) 7000 ≤ |estimated R| and |Street R| < 700 b) |estimated R| < 1000 and 7000 ≤ |road course R| c) |(1 / estimated R × 1000) - (1 / road course R × 1000)| > 1.5

[0178] If any one of the conditions a), b) and c) is met, "there is a large difference".

[0179] The following is referred to again figure2 referenced. In step S6200, for each target, the correction value Ph calculated in step S6000 is added to the current value calculated in step S4000 based on the determination that “there is no large difference” that is taken in step S6100. In this case, a limiting process is performed with upper and lower limit values ​​of 100% and 0%.

[0180] In step S7000, a lane probability is calculated. In the calculation of the lane probability, when it is determined in step S6100 that “there is no large difference”, the current lane probability corrected in step S6200 is used. On the other hand, when it is determined in step S6100 that “there is a large difference”, the actual value (uncorrected) of the lane probability calculated in step S4000 is used. More specifically, this means that a correction Ph of 0% is added to the current lane probability calculated in step S4000.

[0181] In particular, a filtering process is described with reference to the following equation: Lane probability ← previous lane probability × α + actual lane probability × (1 – α) where α is a parameter that depends on the distance Z and using the in the figure 10 shown figure is calculated. An initial value of the lane probability is 0%.

[0182] In subsequent step S8000, a preceding vehicle is determined. Of the targets with a lane probability of 50% or more calculated in step S7000, the target with a minimum distance Z is determined as a preceding vehicle. According to the distance to the target determined as a preceding vehicle and the relative speed of the target, the device 1 controls the vehicle so that the distance to the preceding vehicle is kept constant, or the device 1 issues an alarm when the vehicle There is a risk of colliding with the vehicle in front.

[0183] Subsequently, when the road shape recognition is executed again in step S5000, the road shape R is subjected to low-pass filtering based on the lane probability calculated in step S7000 to correct the estimated R of the vehicle using the filtered road shape. In this way, the course of the road is recognized more accurately.

[0184] As described above, an actual lane probability (step S4000) is used to calculate a lane probability (step S7000). When there is a large difference between the estimated R (step S3000) calculated in the vehicle and the road shape R (step S5000) calculated based on the road shape recognition, the lane probability actual value (step S4000) not corrected.

[0185] Consequently, in a situation where an estimated R deviates significantly from a road shape R, such as in FIGS figure 8A and figure 8B, a lane probability suitable for this situation is obtained. More specifically, the problem that would be caused by the correction of an actual lane probability is avoided, thereby enabling more accurate road shape recognition. Furthermore, accuracy in selecting a preceding vehicle is improved. Accordingly, in a situation where an estimated R would differ significantly from a road shape R, a correct road shape is calculated with high probability.

[0186] Of course, if the estimated R (step S3000) calculated in the vehicle does not deviate much from the road shape R (step S5000) calculated based on the road shape recognition, the lane probability actual value (step S4000) is corrected (Step S6200). Also in this case, the shape of the road is recognized correctly, so that the accuracy in selecting a vehicle ahead is improved. In this case, a correction value and a priority level are changed in accordance with the degree of road shape recognition. Consequently, the R is corrected more appropriately. This results in a preceding vehicle being selected with higher accuracy.

[0187] In the present embodiment, at least one set of one set corresponds to the steering sensor 27 and the steering angle calculation block 49 and a set of the yaw rate sensor 38 and the yaw rate calculation block 51 the direction change detecting means, while the turning radius calculation block 63 corresponds to the curve radius calculation means. Furthermore, the distance / angle measuring device corresponds 5 the radar mean. The polar / orthogonal coordinate conversion block 41 and the object detection block 43 correspond to the object recognition means. The block 53 for preceding vehicle determination corresponds to the lane probability calculation means, the preceding vehicle selection means, the same lane determination means, and the correction value calculation means. The road course recognition block 45 corresponds to the road shape detection means. [Modifications]

[0188] The embodiment described above is provided as an example only. The present invention is not limited to the embodiment described above, but may be constructed in other ways. According to the above embodiment, the device is configured based on selection of a preceding vehicle, for example. Alternatively, the device may be configured simply based on road shape recognition. Alternatively, the configuration can be modified as shown in the following items (1) to (3). (1) In the above embodiment, a determination is made with reference to both the left and right road edges. When the results of the determination are different between the left and right road edges, the one with a higher priority level is used. Alternatively, a determination may be made with reference to either the left or right shoulder of the road. However, the roadsides are not always recognized continuously. Consequently, as in the above embodiment, both the left and right roadsides can be more directly recognized. (2) In the above embodiment, specific numerical values ​​are introduced for use as a correction value or priority level. However, the numeric values ​​are given as examples only and may therefore be changed as appropriate. [Second embodiment]

[0189] A road shape recognition method and a road shape recognition device for vehicles according to a second embodiment of the present invention will be described below with reference to FIG figure 12 to figure 17 described.

[0190] It should be noted that in the second and subsequent embodiments, components that are the same as or similar to those in the already described embodiment are given the same reference numerals to avoid redundant descriptions. When describing a process of recognizing a road shape, excessive omission of overlapping parts of description may make the content difficult to understand. In such a case, therefore, an overlapping description can be made.

[0191] figure 12 is a diagram showing a configuration of a vehicle control device 1 to which the road shape recognition device for vehicles of the present invention is applied. The vehicle control device 1will be described centered on the differences from the first embodiment.

[0192] In the computer 3 the data of a distance r and a scanning angle θ obtained by the distance / angle calculation unit 5b the distance / angle measuring device 5 are output, to the data grouping block 41 sent. The block 41 converts the data of the distance r and the scanning angle θ into an X-Z orthogonal coordinate with the center of the laser radar as the origin (0, 0), the vehicle width direction as the X axis, and the vehicle length direction as the Z axis. The converted data is grouped to form a segment. The process of segment formation will be described later. Object unit data contained in the data grouping block 41 to be segmented are sent to the object detection block 43 and the road course recognition block 45 issued.

[0193] In the object detection block 43 a speed (Vx, Vz) of an obstacle such as a preceding vehicle with respect to the position of the device-carrying vehicle is calculated based on temporal changes in the center position of the object detected by the data grouping block 41 have been issued. Furthermore, the object detection block calculates 43 an object type, d. H. whether the object is a stationary object or a moving object based on the vehicle speed V calculated by the vehicle speed calculation block 47 in accordance with the detection value from the vehicle speed sensor 7 is obtained is output, and based on the relative velocity (Vx, Vz) calculated as described above. Based on the object type and the object's center position, the object detection block chooses 43 an object that would affect driving of the vehicle, and shows the object detection block 43 the distance to the object using the distance indicator 15 on. The reference characters (W, D) indicating the size of an object describe the (width, depth) of the object.

[0194] The components that make up the hardware structure are the same or similar to those shown in FIG figure 1 are shown.

[0195] An operation of the vehicle control device configured as described above will be explained below 1 when recognizing a road course with reference to the in the figure 13 shown flowchart described. In an initial step S1000 of figure 13, distance / angle measurement data, i. H. Distance / angle data read by the distance / angle measuring device 5. In this step, distance / angle data is acquired according to one scan. The sampling period is 100 ms, i. H. distance / angle data is collected every 100 ms.

[0196] In the subsequent step S2000, data (object unit data) is segmented. When segmenting, the data grouping block 41 is executed, the distance / angle data is converted from a polar coordinate system to an orthogonal X-Z coordinate system as described above, and the converted data is grouped to form a segment.

[0197] The segmentation is in the figure 14A. In the present embodiment, a collection of data recognized in the form of dots is integrated to form segment data. The segment data is calculated when the following two conditions are satisfied: a distance ΔX between dots as recognized data in the X-axis direction is less than or equal to 0.2 m; and a distance ΔZ between dots in the Z-axis direction is less than or equal to 2 m. The segment data is ensured to have a sufficient size to have the integrated collection of dots. The segment data corresponds to a rectangular area with two sides parallel to the X-axis and two sides parallel to the Z-axis. Thus, the segment data has center coordinates (X, Z) and two-side data (W, D) indicating the size. The left and right ends of the rectangular area are also included as coordinate data.

[0198] In step S3000, object recognition is performed. Object detection is performed by the object detection block 43 executed. The details correspond to the details described above.

[0199] In step S3100, stationary object data (segment data) obtained in step S3000 of the previous cycle is read to perform a data adding process of adding the read data to the data of the current cycle. The previous cycle here refers to a cycle immediately before the current cycle, the cycle as shown in the flow chart of FIG figure 13 is repeated in a predetermined cycle.

[0200] For example, if it is assumed that the number of stationary objects obtained in the current and the previous cycle is at 38 respectively. 42 lies, the number of stationary objects obtained in the current step gives a sum of both, i.e. H. 80. Consequently, in the current step, a past value is added to the current value.

[0201] Accordingly, in step S4000 and the following steps, both of the data obtained in the previous cycle and the current cycle are dealt with. Specifically, in step S4000 and the subsequent steps, data (segment data) of stationary objects corresponding to two samples is treated, so that the number of data to be used is increased. In this way, the occurrence frequency of stationary objects is increased.

[0202] In the initial cycle of in the figure 13, no previous cycle is available. Consequently, in this case, no previous data is obtained in the step of reading the previous data, so “0” is added to the data obtained in the current cycle. The data obtained in step S3000 in the initial cycle is used in the subsequent cycle as data of the previous cycle.

[0203] Step S4000 and the following steps are performed by the road shape recognition block 45 executed. In step S4000, the center positions of individual segments obtained in step S2000 are converted into a polar coordinate system and sorted by angle.

[0204] In step S5000, the segments that satisfy a connection requirement are grouped clockwise so as to form a roadside object group (left). This is particularly the case with reference to the figure 14B described. In the present embodiment, the course of the road is recognized based on marker posts arranged at a roadside. Consequently, segments such as unnecessary billboards or vehicles other than those of the delineators are first removed. To this end, those segments that meet one of the following distance requirements are removed, considering them as segments such as unnecessary billboards or vehicles. Segments with a large horizontal width W: distance requirements: Horizontal width ≥ 1.2m; and horizontal-to-vertical ratio D / W < 5 Segments in the close range of a moving object relative to an object type: distance requirements: Distance between center positions is ΔX ≤ 2m; and ΔZ ≤ 2m

[0205] Then, the segments remaining after removing the segments that meet the removal requirements are processed clockwise. Specifically, of the remaining segments, those having a distance Z of monotonic increase and satisfying the following connection requirement are processed clockwise as long as such segments exist. During processing, the segments are sequentially connected for grouping to form a roadside object group (left). Connection Requirement:

[0206] Distance between center positions is ΔX ≤ 3.5m; and ΔZ ≦ 55 m Subsequently, when the distance decreases or does not satisfy the above connection requirement although it increases monotonically, a new other roadside object group (left) is formed. Although a single segment can form a roadside object group (left), only those roadside object groups (left) each made up of three or more than three segments are used herein in the detection of a roadside. In the situation according to the figure 14B, as a result of removing the segments that meet the removal requirements and grouping the remaining segments, roadside object groups (left) #1 through #4 are obtained. However, roadside object group (left) No. 1, which is the only roadside object group having three or more than three segments, is used for roadside detection.

[0207] The roadside object group (left) No. 1 selected in this manner has the previous cycle data added in the data adding process of step S3100 described above. More specifically, the roadside object group (left) No. 1 is introduced based on the increased amount of data. Consequently, the accuracy of the roadside object group (left) #1 is improved.

[0208] In the subsequent step S5100, a segment with a largest distance Z, i. H. a farthest segment, among the segments constituting the roadside object group (left) (i.e. roadside object group #1 in this case), is determined as to whether it is on the left side or the right side of the road. The details of the determination will be described later. The farthest segment, if determined to be on the right side of the road, is then removed from the roadside object group (left). figure 15 shows in (a) an example in which a roadside object group (left) is composed of three or more than three segments and a segment on the right side of the road is mixed. As shown in the figure, when segments are grouped clockwise with a monotonic increase in distance Z, no problem arises if only a farthest segment is determined as to whether it actually exists on the left side of the road or Not. This is because it is hard to imagine a situation where a segment immediately before the farthest segment (hereinafter referred to as the second farthest segment) also exists on the right side of the road.

[0209] A farthest segment is determined as follows. The segments, except for the farthest segment, are, as in figure 15 shown by (b) connected via a smoothly curved line. Then, it is determined whether or not the farthest segment is located in the vicinity of the curved line. The curved line is part of a circle passing two points, i.e. H. a shortest segment with a smallest distance Z and the second most distant segment, and perpendicular to the X-axis. Since the circle should be perpendicular to the X-axis, its center should be on the X-axis. Hence, an equation of the circle is introduced from the center of the circle and the two points on the circumference.

[0210] If the distance ΔX in the X-axis direction between the circle and the farthest segment is less than 1.5 m, the farthest segment is determined to exist on the left side of the road and remains in the roadside object group (Left) contain. When the distance ΔX is greater than or equal to 1.5 m, the farthest segment is determined to be on the right side of the road and is removed from the roadside object group (left). The location of the farthest segment can be determined based on the closest or shortest distance between the farthest segment and the circle, ie. H. the length of a normal extending from the farthest segment to the circle. In practice, however, no problem arises if the farthest segment is determined only based on the distance ΔX in the X-axis direction.

[0211] In subsequent steps S6000 and S6100, the processings in steps S5000 and S5100 are executed with left and right being reversed. Specifically, in step S6000, the segments that satisfy the connection requirement are grouped counterclockwise based on the angles obtained in step S4000 so as to form a roadside object group (right). The processing for removing unnecessary billboards or the like is the same as that performed for the left side. The segments remaining after removal, having a distance Z of monotonic increase and satisfying the connection requirement, are processed in a counter-clockwise direction. During processing, the segments are sequentially connected for grouping to form a roadside object group (right). The connection requirement is also the same as that for the left side. Further, like the processing for the left side, only the roadside object groups (right) each composed of three or more than three segments are used for roadside recognition.

[0212] In step S6100, processing similar to that for the left side is applied to the farthest segment in a roadside object group (right). In particular, the farthest segment is determined as to whether it is on the left or the right side of the road. If it is determined to be on the left, the farthest segment is removed from the roadside object group (right). After the roadside object groups (left) and the roadside object groups (right) are obtained in this way, processing proceeds to step S7000. It should be noted that the farthest segment in a roadside object group (left) may be the farthest segment in the roadside object group (right). More precisely, as in the figure 16 shown by (a), a single farthest segment can belong to both the roadside object group (left) and the roadside object group (right), creating a conflict situation. In such a case, the farthest segment, as in the figure 16 shown by (b) is removed. Otherwise, such processing is not performed.

[0213] In step S8000, the roadside on the left and right is recognized based on the roadside object groups (left) and the roadside object groups (right) each made up of three or more than three segments. In the present embodiment, the segments constituting each roadside object group are as shown in FIG figure 17 is interpolated to recognize the left or right edge of the road as a collection of line segments. Further, using the results of interpolation between roadside object group data, an intersection point with the X-axis is calculated, followed by interpolation also up to the intersection point. Consequently, it is ensured that the course of the road starting from the vicinity of the vehicle position is recognized as a collection of line segments.

[0214] According to the vehicle control device 1 of the present embodiment, when executing the road shape recognition process as described above, the stationary object data (segment data) obtained in the previous cycle is added to the stationary object data (segment data) obtained in the current cycle cycle have been obtained. Consequently, in the road shape recognition process, roadside object groups are formed, increasing the amount of usable data. Consequently, a situation where roadside objects are very many is created under the conditions where it is difficult to detect roadside objects such as stationary objects because there is a preceding vehicle or where the number of roadside objects, usable for road shape recognition is small, such as when the number of roadside objects of the road on which the vehicle is running is originally low. More specifically, the amount of data that can be used to form roadside object groups is increased. As a result, the accuracy in detecting road edges is improved. Furthermore, the course of the road is calculated more correctly and with a good frequency.

[0215] The shape of the road recognized on this side is used for determining a preceding vehicle and hence for controlling the inter-vehicle distance and issuing an alarm in the inter-vehicle distance control. Accordingly, a vehicle-to-vehicle distance is favorably controlled and an alarm is favorably issued.

[0216] In the present embodiment, the distance / angle measuring device corresponds to 5 the radar mean. The data grouping block 41 , the object detection block 43 and the road course recognition block 45 of the computer 3 correspond to the detection means. From the blocks 41 , 43 and 45 correspond to the data grouping block 41 and the object detection block 43 the object recognition means, and corresponds to the road shape recognition block 45 the effective data extracting means, the data adding means, the roadside object group forming means, and the roadside recognizing means.

[0217] Furthermore, the processings correspond to those of the data grouping block 41 , the object detection block 43 and the road course recognition block 45 are executed, the recognition process. The processing performed by the data grouping block 41 and the object detection block 43 are executed correspond to the detection process and the processing performed by the road shape recognition block 45 is executed corresponds to the extracting process and the data adding process. [Third embodiment]

[0218] A road shape recognition method and a road shape recognition device for vehicles according to a third embodiment of the present invention will be described below with reference to FIG figure 12, figure 18, figure 14, figure 19 and figure 15 to figure 17 described.

[0219] The vehicle control device 1 to which the road shape recognition device for vehicles according to the present embodiment is applied has hardware components similar to those in FIG figure 12 on.

[0220] The description of the vehicle control device 1 of the present embodiment focuses on the differences from the second embodiment.

[0221] An operation involved in road shape recognition will be described below with reference to that shown in FIG figure 18 will be described. In an initial step S1000 of figure 18, distance / angle data from the distance / angle measuring device 5 is read. In particular, the distance / angle measuring device detects 5 Distance / angle data corresponding to one scan. In this, the sampling cycle is 100 ms and detects the distance / angle measuring device 5 consequently every 100 ms distance / angle data.

[0222] In the subsequent step S2000, data (object unit data) is segmented. When segmenting, the data grouping block 41 is executed, the distance / angle data is converted from a polar coordinate system to an orthogonal X-Z coordinate system as described above, and the converted data is grouped to form a segment.

[0223] The segmentation is in the figure 14A. In the present embodiment, a collection of data recognized in the form of dots is integrated to form segment data. The segment data is calculated when the following two conditions are satisfied: a distance ΔX between dots as recognized data in the X-axis direction is less than or equal to 0.2 m; and a distance ΔZ between dots in the Z-axis direction is less than or equal to 2 m. Each segment data is ensured to have a size enough to include the collection of integrated dots. The segment data corresponds to a rectangular area with two sides parallel to the X-axis and two sides parallel to the Z-axis. Thus, the segment data has a center coordinate (X, Z) and two-side data (W, D) indicating the size. The left and right ends of the rectangular area are also included as coordinate data.

[0224] In step S3000, object recognition is performed. Object detection is performed by the object detection block 43 executed. The details of the object detection correspond to the details described above. A vehicle in front, i. H. an immediately preceding vehicle, and a vehicle in front of the preceding vehicle, d. H. a second preceding vehicle which are moving objects are detected by the object detection.

[0225] In step S3100, a stationary object removal process is performed. Stationary objects may exist on the traveling road between the vehicle having the device and an immediately preceding vehicle, or between the immediately preceding vehicle and a second preceding vehicle. Such stationary objects are, with a high probability, reflectors on the road surface or billboards and consequently with a high probability not roadside objects that contribute to the recognition of the course of the road. For this reason, relevant stationary objects are removed from the stationary objects recognized by the stationary object removal process executed in step S3000.

[0226] Consequently, the presence of an immediately preceding vehicle and a second preceding vehicle are recognized first. Requirements for a vehicle immediately ahead are: tracking flag is set; and lane probability ≥ 70%

[0227] A lane probability is a parameter indicating a probability that a target is a vehicle traveling on the lane on which the device-equipped vehicle is traveling. Since the way in which a lane probability is calculated is known, it will not be discussed in any more detail below. Requirement for a second-leading vehicle: tracking flag is set; and further away than the vehicle in front

[0228] If the requirements for a vehicle immediately ahead or a vehicle second ahead are not met, ie. H. if there is no immediately preceding vehicle or second preceding vehicle, the current step need not be performed. Of course, the requirements for an immediately preceding vehicle or a second preceding vehicle set forth above are only examples, and the manner of determining an immediately preceding vehicle or a second preceding vehicle can be interpreted as appropriate .

[0229] Then, the stationary objects existing between the vehicle having the device and an immediately preceding vehicle and between the immediately preceding vehicle and a second preceding vehicle are removed from the detected objects. figure 19 shows the device-equipped vehicle, an immediately preceding vehicle, a second-preceding vehicle, and stationary objects around the device-equipped vehicle. The circles in the figure 19 show stationary objects such as manhole covers or billboards. The portion with the dotted hatching shows a traveling road of the device-equipped vehicle or the second-preceding vehicle. In this situation, just as they are in the figure For example, as shown in FIG. 19, assume that a stationary object shown by a mark ⊗ exists between the immediately preceding vehicle and the second preceding vehicle. The data (object unit data) corresponding to this stationary object is removed from the segment data.

[0230] Step S4000 and subsequent steps are performed by the road shape recognition block 45 executed. In step S4000, the center positions of the segments obtained in step S2000 are transformed into a polar coordinate system and sorted by angle.

[0231] In step S5000, the segments that satisfy the connection requirement are grouped clockwise based on the angles obtained in step S4000 so as to form the roadside-object group (left). The is in particular with reference to the figure 14B described. In the present embodiment, the course of the road is recognized based on marker posts arranged at the roadside. Consequently, segments such as unnecessary billboards or vehicles other than those of the delineators are first removed. For this purpose, those segments that meet one of the following removal requirements are removed, determining them as segments such as billboards or vehicles. Segments with a large horizontal width W: distance requirements: Horizontal width ≥ 1.2m; and horizontal-to-vertical ratio D / W < 5 Segments in the vicinity of a moving object with respect to an object type: distance conditions: Distance between center positions is ΔX ≤ 2m; ΔZ ≤ 2m

[0232] Then, the segments that remain after removing the segments that meet the removal requirements are processed in a clockwise direction. Specifically, of the remaining segments, those having a distance Z of monotonic increase and satisfying the following connection requirement are processed clockwise as long as such segments exist. During processing, the segments are sequentially connected for grouping to form a roadside object group (left). Connection Requirement: Distance between center positions is ΔX ≤ 3.5m; ΔZ ≤ 55m

[0233] Subsequently, when the distance is decreased or does not satisfy the above connection requirement, although monotonously increasing, a new other roadside object group (left) is formed. Here, although a single segment can constitute a roadside object group (left), only those roadside objects (left) each composed of three or more than three segments are used in the detection of a roadside. In the situation that in the figure 14B, as a result of removing the segments that satisfy the removal conditions and grouping the remaining segments, roadside object groups (left) Nos. 1 to 4 are obtained. However, roadside object group (left) #1, which is the only roadside object group having three or more than three segments, is used for roadside detection.

[0234] Of course, the roadside object group (left) No. 1 selected in this way does not include the stationary objects on the traveling road removed in the stationary object removal process executed in step S3100. More specifically, the stationary objects on the traveling road do not correspond to the “segments satisfying the connection requirement” referred to in the current step and thus are not subjected to the grouping. Consequently, the accuracy of the roadside object group (left) #1 is improved.

[0235] In the subsequent step S5100, a segment with a largest distance Z, i. H. a farthest segment, among the segments constituting the roadside-object group (left) (i.e., roadside-object group #1), is determined as to whether it exists on the left side or the right side of the road. The details of the determination will be described later. The furthest segment, if determined to be on the right side of the road, is then removed from the roadside object group (left).

[0236] figure 15 shows in (a) an example in which a roadside object group (left) is composed of three or more than three segments and a segment on the right side of the road is mixed. As can be seen from the figure, when segments are grouped clockwise with a monotonic increase in distance Z, no problem arises if only a farthest segment is determined as to whether it actually exists on the left side of the road or not. This is because it is hard to imagine a situation where a segment immediately before the farthest segment (hereinafter referred to as the second farthest segment) also exists on the right side of the road.

[0237] A farthest segment is determined as follows. The segments, except for the farthest segment, are, as in figure 15 shown by (b) connected via a smoothly curved line. Then, it is determined whether or not the farthest segment is located in the vicinity of the curved line. The curved line is part of a circle passing two points, i.e. H. a shortest segment with a smallest distance Z and the second most distant segment, and perpendicular to the X-axis. Since the circle should be perpendicular to the X-axis, its center should be on the X-axis. Hence, an equation of the circle is introduced from the center of the circle and the two points on the circumference.

[0238] If the distance ΔX in the X-axis direction between the circle and the farthest segment is less than 1.5 m, the farthest segment is determined to exist on the left side of the road and remains in the roadside object group (Left) contain. When the distance ΔX is greater than or equal to 1.5 m, the farthest segment is determined to be on the right side of the road and is removed from the roadside object group (left). The location of the farthest segment can be determined based on the closest or shortest distance between the farthest segment and the circle, ie. H. the length of a normal extending from the farthest segment to the circle. In practice, however, no problem arises if the farthest segment is determined only based on the distance ΔX in the X-axis direction.

[0239] In subsequent steps S6000 and S6100, the processings in steps S5000 and S5100 are executed with left and right being reversed. Specifically, in step S6000, the segments that satisfy the connection requirement are grouped counterclockwise based on the angles obtained in step S4000 so as to form a roadside object group (right). The processing for removing unnecessary billboards or the like is the same as that performed for the left side. The segments remaining after removal, having a distance Z of monotonic increase and satisfying the connection requirement, are processed in a counter-clockwise direction. During processing, the segments are sequentially connected for grouping to form a roadside object group (right). The connection requirement is also the same as that for the left side. Further, like the processing for the left side, only the roadside object groups (right) each composed of three or more than three segments are used for roadside recognition.

[0240] In step S6100, processing similar to that for the left side is applied to the farthest segment in a roadside object group (right). In particular, the farthest segment is determined as to whether it is on the left or the right side of the road. If it is determined to be on the left, the farthest segment is removed from the roadside object group (right). After the roadside object groups (left) and the roadside object groups (right) are obtained in this way, processing proceeds to step S7000. It should be noted that the farthest segment in a roadside object group (left) may be the farthest segment in the roadside object group (right). More precisely, as in the figure 16 shown by (a), a single farthest segment can belong to both the roadside object group (left) and the roadside object group (right), creating a conflict situation. In such a case, the farthest segment, as in the figure 16 shown by (b) is removed. Otherwise, the farthest segment's data is not removed.

[0241] In step S8000, the roadside on the left and right is recognized based on the roadside object groups (left) and the roadside object groups (right) each made up of three or more than three segments. In the present embodiment, the segments constituting each roadside object group are as shown in FIG figure 17 is interpolated to recognize the left or right edge of the road as a collection of line segments. Further, using the results of interpolation between roadside object group data, an intersection point with the X-axis is calculated, followed by interpolation also up to the intersection point. Consequently, it is ensured that the course of the road starting from the vicinity of the vehicle position is recognized as a collection of line segments.

[0242] According to the vehicle control device 1 of the present embodiment, in the road shape recognition process, the stationary object data is removed from segment data when the stationary objects are on the traveling road between the vehicle having the device and an immediately preceding vehicle or on the traveling road between the immediately preceding vehicle and a vehicle in second place ahead. Consequently, the right or left end position of segment data will not be located on the traveling road on which the device-equipped vehicle or the immediately preceding vehicle is traveling. This causes the roadside feature groups to approximate the actual shape of the road after they have been grouped. As a result, the accuracy in detecting road edges is improved. Furthermore, the course of the road is calculated more precisely and with good frequency.

[0243] The shape of the road recognized on this side is used for determining a preceding vehicle and hence for controlling the inter-vehicle distance and issuing an alarm in the inter-vehicle distance control. Accordingly, a vehicle-to-vehicle distance is favorably controlled and an alarm is favorably issued.

[0244] In the present embodiment, the distance / angle measuring device corresponds to 5 the radar mean. The data grouping block 41 , the object detection block 43 and the road course recognition block 45 of the computer 3 correspond to the detection means. From the blocks 41 , 43 and 45 correspond to the data grouping block 41 and the object detection block 43 the object recognition means, and corresponds to the road shape recognition block 45 effective data extracting means, stationary object removing means, roadside object group forming means, and roadside detecting means. [Modifications]

[0245] The above embodiment can be implemented in various ways without departing from the scope of the present invention. For example, stationary objects need not necessarily exist in both the traveling road area between the apparatus-equipped vehicle and an immediately preceding vehicle and the traveling road area between the immediately preceding vehicle and a second preceding vehicle. If stationary objects are present in at least one of the areas of the traveling road, the data can be removed. [Fourth embodiment]

[0246] A road shape recognition method and a road shape recognition device for vehicles according to a fourth embodiment of the present invention will be described below with reference to FIG figure 12 and figure 20 to figure 25 described.

[0247] The vehicle control device 1 , to which the road shape recognition device for vehicles according to the present embodiment is applied, has hardware components similar to those in FIG figure 12 are constructed.

[0248] The description of the vehicle control device 1 of the present embodiment is focused on the differences from the second embodiment.

[0249] The internal structure of the computer 3 is described using control blocks. In the computer 3 becomes the data of a distance r and a scanning angle θ obtained by the distance / angle calculation unit 5b the distance / angle measuring device 5 are output, to the data grouping block 41 sent. The block 41 converts the data of the distance r and the scanning angle θ into an X-Z orthogonal coordinate with the center of the laser radar as the origin (0, 0), the vehicle width direction as the X axis, and the vehicle length direction as the Z axis. The converted data is grouped and segmented. The process of segment formation will be described later. The object unit data contained in the data grouping block 41 to be segmented are sent to the object detection block 43 and the road course recognition block 45 issued.

[0250] In the object detection block 43 a speed (Vx, Vz) of an obstacle such as a preceding vehicle with respect to the position of the device-carrying vehicle is calculated based on temporal changes in the center position of the object detected by the data grouping block 41 have been issued. Furthermore, the object detection block calculates 43 an object type, d. H. whether the object is a stationary object or a moving object based on the vehicle speed V calculated by the vehicle speed calculation block 47 in accordance with the detection value from the vehicle speed sensor 7is obtained is output, and based on the relative velocity (Vx, Vz) calculated as described above. Based on the object type and the object's center position, the object detection block chooses 43 an object that would affect driving of the vehicle, and shows the object detection block 43 the distance to the object using the distance indicator 15 on. The reference characters (W, D) indicating the size of an object describe the (width, depth) of the object.

[0251] The sensor error detection block 44 detected whether in the object detection block 43 calculated data has a value that falls within an erroneous range. If the value falls within the error range, the sensor error display shows 17 this accordingly. On the other hand, the road course recognition block recognizes 45 the course of the road based on the data of the center positions of the objects obtained from the grouping block 41 have been output, and based on the data contained in the object detection block 43 be calculated. The details of the road shape recognition process will be described later. The data in the road shape recognition block 45 are received are sent to the block 53 is output for determining a preceding vehicle.

[0252] Furthermore, the steering angle calculation block calculates 49 a steering angle based on a signal from the steering angle sensor 27 . The yaw rate calculation block 51 calculates a yaw rate based on a signal from the yaw rate sensor 28 . The curve radius (radius of curvature) calculation block 63 calculates a turning radius (radius of curvature) R based on a vehicle speed calculated by the vehicle speed calculation block 47 is obtained, a steering angle calculated by the steering angle calculation block 49 is obtained and a yaw rate obtained from the yaw rate calculation block 51 is obtained. The block 53 for determining a preceding vehicle selects a preceding vehicle based on the turning radius R as well as the object type, the center position (X, Z), the size of the object (W, D) and the relative speed (Vx, Vz), which are stored in the object detection block 43 are calculated, and the road shape data stored in the road shape recognition block 45 be obtained. Then the block calculates 53 to determine a preceding vehicle, a distance Z to the preceding vehicle and a speed Vz with respect to the preceding vehicle.

[0253] Then the vehicle-to-vehicle control and alarm output block hits 55 an alarm determination or a driving determination based on the distance Z to the preceding vehicle, the relative speed Vz, the speed Vn of the vehicle having the device, the acceleration of the preceding vehicle, the object center position, the object width, the object type, the setting conditions of the cruise control switch 26 and the degree of braking applied to the brake switch 9 is applied, and the throttle position from the throttle position sensor 11 and the sensitivity value set by the sensitivity adjustment unit 25 is determined. When an alarm determination is made, it is further determined whether or not an alarm should be raised. When a driving determination is made, the details of the vehicle cruise control are determined. As a result of the determination, when an alarm is to be issued, the block gives 55 an alarm generation signal to the alarm sound generation unit 13 out. When a driving determination is made, the block returns 55 Control signals to the automatic transmission controller 23 , the brake control unit 19 and the throttle control unit 21 to run controls as needed. When these controls are executed, the block 55 a required indication signal to the distance indicator 15 to inform the driver of the situation.

[0254] An operation involved in recognizing a road shape performed by the vehicle control device will be described below 1 is executed, which is configured as described above, with reference to the in the figure 20 shown flowchart described. In step S1000, which is an initial step in the figure is 20, distance / angle measurement data, i. H. Distance / angle data, from distance / angle measuring device 5 had read. In particular, the distance / angle measuring device detects 5 Distance / angle data corresponding to one scan. In this case, the sampling cycle is 100 ms, so the distance / angle measuring device 5 distance / angle data collected every 100 ms.

[0255] In the subsequent step S2000, data (object unit data) of stationary objects is segmented. When segmenting, the data grouping block 41 is executed, the distance / angle data is converted from a polar coordinate system to an orthogonal X-Z coordinate system as described above, and the converted data is grouped to form segments (roadside object group).

[0256] In particular, as in figure 21A, a start point selection process is performed to select a roadside object group, i. H. segments to generate. In the start point selection process, one of the stationary objects sorted by the angle is selected as a start point for forming segments.

[0257] In particular, after sorting by angle, in a left angle direction: (1) The lateral position of a stationary object which is located innermost with respect to the vehicle (located on the vehicle side) is extracted in the left side area. (2) A stationary object with a lateral position near (1) and a shortest direct distance to the vehicle is used as a starting point.

[0258] Regarding the condition (1), a lateral position is extracted for a stationary object located at a position from which the distance to the vehicle in the vehicle width direction is shortest (closest to the vehicle). Regarding the condition (2), the “direct distance to the vehicle” refers to the shortest distance from the vehicle to the stationary object, not a distance from the vehicle in the vehicle width direction.

[0259] Regarding the right angle direction, left and right in the conditions ( 1 ) and (2) vice versa. Specifically, in the condition (1), the lateral position of a stationary object located innermost with respect to the vehicle in the right-side area is extracted.

[0260] Regarding the right angle direction, for example, as shown on the left in FIG figure 21A, stationary objects numbered from “0” to “8” exist in the right side area with respect to the inclination of 6° from the Z-axis direction with the vehicle centered. Specifically, in the right-side area, the stationary objects numbered "0" to "8" exist in an area that is 1 m to 8 m in the X-axis direction of the vehicle and 30 m in the Z-axis direction. Axis direction of the vehicle covers. This area is indicated by the shading on the left in the figure figure 21A shown.

[0261] Of these stationary objects, the stationary object No. 0 has the smallest angle with respect to the X-axis but has the greatest distance from the vehicle in the X-axis direction. The stationary objects #1 to 8 are within a range from a position of the stationary object #8 at the closest distance (X min) to the vehicle in the X-axis direction to a position 2 m away from the stationary object No. 8 is located remotely, present.

[0262] When the conditions (1) and (2) are applied to this situation, the lateral position in the X-axis direction of the stationary object No. 8 (X min) is extracted (condition (1)). Further, stationary object #1 having a lateral position close to stationary object #8 and having the shortest distance to the vehicle is determined as a starting point (condition (2)). Consequently, a grouping connection is prevented from starting from the stationary object No. 0 located far away from the vehicle. Further, when the roadside is seen twice, grouping is preferably started from an inner roadside line.

[0263] Accordingly, a roadside object group (segments) is formed from the starting point determined as described above. This is with reference to the figure 22 explained. figure 22 shows an example of stationary connection objects arranged in the right angle direction with respect to the vehicle.

[0264] Starting from the stationary object located closest to the vehicle (stationary object #1 located on the left in the figure 21A), as in figure As shown in Fig. 22, the stationary objects included in both a connection requirement area a and a connection requirement area b smaller than the connection requirement area a are successively connected. In particular, the stationary objects are compared by angle. Of the stationary objects determined to fall within the connection requirement area a, those also falling within the connection requirement area b are interrogated and connected. Each stationary object connected in this way is used as a base point for sequentially querying and connecting to the stationary objects included in both of the connection requirement areas a and b.

[0265] In the figure 22, of the two stationary objects located in the connection requirement area a, one also exists in the connection requirement area b. Consequently, the stationary object located in both of the connection requirement areas a and b is connected. Such connection is repeated, and the stationary objects are connected one by one so as to form a roadside object group.

[0266] In the figure 22, the connection required area b is located on the left side (Z-axis side) in the connection required area a. However, this is merely an example of an arrangement of the connection requirement area b with respect to the connection requirement area a. For example, the connection required area b may be located at the center in the vehicle width direction (X-axis direction) in the connection required area a. Consequently, the location of the connection requirement area b in the connection requirement area a can be determined as appropriate.

[0267] The stationary objects as a collection of dots are integrated as described above to obtain segment data. The segment data corresponds to a rectangular area with two sides parallel to the X-axis and two sides parallel to the Z-axis and of a size that includes the integrated collection of dots. Consequently, the segment data, as shown in the right figure 21A has a center coordinate (X, Z) and two-side data (W, D) indicating the size. The coordinates at the left and right ends of the rectangular area are also included as data.

[0268] In step S3000, object recognition is performed. Object detection is performed by the object detection block 43 executed. The details of the object detection correspond to the details described above. Step S4000 and subsequent steps correspond to the processing performed by the road shape recognition block 45 is performed. In step S4000, the center positions of the respective segments obtained in step S2000 are converted into a polar coordinate system and then sorted by angle.

[0269] In step S5000, the segments that meet the requirement are grouped clockwise so as to form a roadside object group (left). This is particularly the case with reference to the figure 21B described. In the present embodiment, the course of the road is recognized based on marker posts placed on the roadside. Consequently, segments such as unnecessary billboards or vehicles other than those of the delineators are first removed. To this end, those segments that meet either of the following two distance requirements are removed, considering them as segments such as from unnecessary billboards or vehicles. Segments with a large horizontal width W: distance condition: Horizontal width ≥ 1.2m; and horizontal-vertical ratio D / W < 5 segments in the vicinity of a moving object with respect to an object type: Removal condition: distance between center positions ΔX ≤ 2 m; ΔZ ≤ 2m

[0270] Then, the segments remaining after removing the segments that meet the removal requirements are processed clockwise. In particular, of the remaining segments, those having a distance Z of monotonic increase and satisfying the following connection requirement are processed clockwise as long as such segments exist. During processing, the segments are sequentially connected for grouping to form a roadside object group (left). Connection Requirement:

[0271] Distance between center positions is ΔX ≤ 3.5m; ΔZ ≤ 55 m Subsequently, when the distance decreases or does not satisfy the above connection requirement although it increases monotonously, a new other roadside object group (left) is formed. Although a single segment can form a roadside object group (left), only those roadside objects (left) each made up of three or more than three segments are used herein in the detection of a roadside. In the situation that in the figure 21B, as a result of removing the segments that satisfy the removal conditions and grouping the remaining segments, roadside object groups #1 to #4 are obtained. However, roadside object group #1, which is the only roadside object group having three or more than three segments, is used for roadside recognition.

[0272] In the subsequent step S5100, a segment with a largest distance Z, i. H. a farthest segment, among the segments constituting the roadside object group (left) (i.e. roadside object group #1 in this case), is determined as to whether it is on the left side or the right side of the road. The details of the determination will be described later. The farthest segment, if determined to be on the right side of the road, is then removed from the roadside object group (left). figure 23 shows in (a) an example in which a roadside object group (left) is composed of three or more than three segments and a segment on the right side of the road is mixed. As shown in the figure, when segments are grouped clockwise with a monotonous increase in distance, no problem arises if only a farthest segment is determined as to whether it actually exists on the left side of the road or Not. This is because it is hard to imagine a situation where a segment immediately before the farthest segment (hereinafter referred to as the second farthest segment) also exists on the right side of the road.

[0273] A farthest segment is determined as follows. The segments, except for the farthest segment, are, as in figure23 shown by (b) connected via a smoothly curved line. Then, it is determined whether or not the farthest segment is located in the vicinity of the curved line. The curved line is part of a circle passing two points, i.e. H. a shortest segment with a smallest distance Z and the second most distant segment, and perpendicular to the X-axis. Since the circle should be perpendicular to the X-axis, its center should be on the X-axis. Hence, an equation of the circle is introduced from the center of the circle and the two points on the circumference.

[0274] If the distance ΔX in the X-axis direction between the circle and the farthest segment is less than 1.5 m, the farthest segment is determined to exist on the left side of the road and remains in the roadside object group (Left) contain. When the distance ΔX is greater than or equal to 1.5 m, the farthest segment is determined to be on the right side of the road and is removed from the roadside object group (left). The location of the farthest segment can be determined based on the closest or shortest distance between the farthest segment and the circle, ie. H. the length of a normal extending from the farthest segment to the circle. In practice, however, no problem arises if the farthest segment is determined only based on the distance ΔX in the X-axis direction.

[0275] In subsequent steps S6000 and S6100, the processings in steps S5000 and S5100 are executed with left and right being reversed. Specifically, in step S6000, the segments that satisfy the connection requirement are grouped counterclockwise so as to form a roadside object group (right) based on the angles obtained in step S4000. The processing for removing unnecessary billboards or the like is the same as that performed for the left side. The segments remaining after removal, having a distance Z of monotonic increase and satisfying the connection requirement, are processed in a counter-clockwise direction. During processing, the segments are sequentially connected for grouping to form a roadside object group (right). The connection requirement is the same as that for the left side. Further, like the processing for the left side, only the roadside object groups (right) each composed of three or more than three segments are used for roadside recognition.

[0276] In step S6100, processing similar to that for the left side is applied to the farthest segment in a roadside object group (right). In particular, the farthest segment is determined as to whether it is on the left or the right side of the road. If it is determined to be on the left, the farthest segment is removed from the roadside object group (right). After the roadside object groups (left) and the roadside object groups (right) are obtained in this way, processing proceeds to step S7000. It should be noted that the farthest segment in a roadside object group (left) may be the farthest segment in the roadside object group (right). More precisely, as in the figure 24 shown by (a), a single farthest segment can belong to both the roadside object group (left) and the roadside object group (right), creating a conflict situation. In such a case, the farthest segment, as in the figure 24 shown by (b) is removed. Otherwise, such processing is not performed.

[0277] In step S8000, the roadside on the left and right is recognized based on the roadside object groups (left) and the roadside object groups (right) each made up of three or more than three segments. In the present embodiment, the segments constituting each roadside object group are as shown in FIG figure25 is interpolated to recognize the left or right edge of the road as a collection of line segments. Further, using the results of interpolation between roadside object group data, an intersection point with the X-axis is calculated, followed by interpolation also up to the intersection point. Consequently, it is ensured that the course of the road starting from the vicinity of the vehicle position is recognized as a collection of line segments.

[0278] In the road shape recognition process performed by the vehicle control device 1 of the present embodiment, as described above, the lateral position of an innermost stationary object (located on the vehicle side) among the extracted stationary objects is extracted. Further, a stationary object close to the lateral position and closest to the vehicle is used as a starting point. Consequently, a grouping connection is prevented from starting from the stationary object located far from the vehicle as a starting point. Further, when the roadside is seen twice, grouping is preferably started from an inner roadside line. As a result, the accuracy in detecting road edges is improved. Furthermore, the course of the road is calculated more precisely and with good frequency.

[0279] Further, in the present embodiment, stationary objects included in both of the connection requirement areas a and b are connected starting from a stationary object that has been determined as a starting point. Since the stationary objects included in the narrow connection requirement area b are connected, the grouping connection forms a course close to the actual course of the road. As a result, the accuracy in recognizing the road edges is improved. Furthermore, thanks to the comparison of the stationary objects by angle, a stationary object with a large difference in distance but with a small difference in angle is prevented from being preferentially connected. Since two connection requirement ranges are provided to double limit the connection requirement, existing performance is maintained.

[0280] The shape of the road thus recognized is used for determining a preceding vehicle and hence for controlling the inter-vehicle distance and for issuing an alarm in the inter-vehicle distance control. Consequently, a vehicle-to-vehicle distance is favorably controlled and an alarm is favorably issued.

[0281] In the present embodiment, the distance / angle measuring device corresponds to 5 the radar mean. The data grouping block 41 , the object detection block 43 and the road course recognition block 45 of the computer 3 correspond to the detection means. From the blocks 41 , 43 and 45 correspond to the data grouping block 41 and the object detection block 43 the object recognition means and corresponds to the road course recognition block 45 the effective data extracting means, the starting point selecting means, the roadside object group forming means, and the roadside recognizing means.

[0282] Further, the connection requirement range a corresponds to the first connection requirement range, and the connection requirement range b corresponds to the second connection requirement range. [Fifth embodiment]

[0283] A road shape recognition method and a road shape recognition device for vehicles according to a fifth embodiment of the present invention will be described below with reference to FIG figure 26A and figure 26B described.

[0284] The vehicle control device 1 , to which the road shape recognition device for vehicles according to the present embodiment is applied, has hardware components similar to those in FIG figure 12 on.

[0285] The description of the vehicle control device 1 of the present embodiment is focused on the differences from the fourth embodiment.

[0286] In the present embodiment, segments (stationary object groups) on the vehicle side are preferably connected to prepare a roadside table.

[0287] For example, if all of the segments are used to calculate an average roadside, the mean of the roadside is calculated as in figure 26A is changed in the vehicle width direction (X-axis direction). Thus, in the present embodiment, vehicle-side (inside) segments are used for grouping. The grouping of the present embodiment corresponds to steps S5000 and S6000 of the first embodiment.

[0288] In particular, as in figure 26B, circles are calculated passing respective segments and the X-axis. Only certain segments are used for grouping. The specified segments have the intersections falling within a range from an innermost intersection to a point away from the innermost intersection by a predetermined threshold. Consequently, the segment that is in the figure 26B located farthest from the vehicle in the X-axis direction is excluded from the grouping. This ensures that a mean roadside to be calculated is recognized as passing through the interior (vehicle-side) segments.

[0289] In the present embodiment, as described above, roadside object groups located more on the vehicle side (inner side) and on the same curved line are preferably used for the grouping. More specifically, a roadside table is preferably prepared using the vehicle-side (inside) segments. Consequently, in a situation such as when a roadside is double-seen due to multiple segments scattered in the vehicle width direction (X-axis direction), the accuracy in detecting a roadside is improved. In this way, the course of the road is calculated more precisely and with a good frequency. [Modifications]

[0290] The present invention is not limited to the configurations of the first to fifth embodiments, but can be modified in various ways without departing from the scope of the present invention. The present invention can be modified, for example, as shown in the following items (1) to (3). (1) In the above-described embodiments, for example, in step S7000, the farthest segment belonging to both the left and right roadside object groups is removed. Alternatively, the farthest segment can be determined as to whether it belongs to the left roadside object group or the right roadside object group with high probability. Then, the farthest segment may be included in the roadside object group to which the farthest segment has a high probability of belonging. Consequently, the farthest segment can be used as effectively as possible. For example, step S5100 or S6100 may be applied to the determination to be made regarding the farthest segment. In particular, a distance ΔX between the farthest segment and the circle passing the left roadside object group, i. H. passes through can be calculated, and a distance ΔX between the farthest segment and the circle passing the right roadside object group can be calculated in the same way. Then the length of the distances ΔX can be compared to include the farthest segment in the roadside object group with the smallest distance ΔX. Alternatively, for example, a neutral zone can be provided between the two circles. The farthest segment can be determined to belong to the roadside object group if it crosses the neutral zone and is located closer to either the left or right roadside object group. (2) The above-described embodiments and the above item (1) are provided from the standpoint of using the farthest segment as effectively as possible. However, the farthest segment may be unreservedly removed in steps S5100 and S6100 from the standpoint of avoiding erroneous determination as much as possible. Consequently, an erroneous determination that would have been made due to the use of the farthest segment is reliably avoided. (3) In the above-described embodiments, the segments constituting a roadside object group are interpolated to recognize a roadside as a collection of line segments. However, a method of interpolation is not limited to this. For example, segments can be interpolated using curved line segments to recognize a roadside as a smoothly curved line. [Sixth Embodiment]

[0291] A road shape recognition method and a road shape recognition device for vehicles according to a sixth embodiment of the present invention will be described below with reference to FIG figure 27 bis figure 33 described.

[0292] the figure 27 bis figure 33 show a system configuration of an inter-vehicle control device 1A , which is applied to the road shape recognition device for vehicles according to the present embodiment.

[0293] As in figure 27 and similar to the configurations that have been described so far, the inter-vehicle control device 1A , in addition to a computer 2A , a distance / angle measuring device 5 , a vehicle speed sensor 7 , a steering angle sensor 27 , a yaw rate sensor 28 , a cruise control switch 26 , a brake switch 9 , a throttle position sensor 11 , an alarm volume adjustment unit 24 , a sensitivity adjustment unit 25 , a circuit breaker 29 , a sensor error indicator 17 , a distance indicator 15 , a brake control unit 19 , a throttle control unit 21 , an automatic transmission controller 23 and an alarm sound generating unit 13 on.

[0294] The computer 2A has an input / output interface (I / O) and various drive circuits and detection circuits. Since these hardware components are of known design, they will not be discussed in any more detail below. The computer 2A has a circuit breaker 29 on. If the computer 2A is turned on, the computer 2A powered to start predetermined processes. The computer 2A executes not only the inter-vehicle distance control described in the present embodiment but also cruise control when a preceding vehicle is not selected to keep the vehicle speed at a certain speed.

[0295] The distance / angle measuring device 5 corresponding to a radar system has a transmitter / receiver 5a and a distance / angle calculation unit 5b on. The sender / receiver 5a outputs a laser beam to scan an area in the forward direction of the vehicle covering a predetermined angular range and detects the reflected light. Furthermore, the transmitter / receiver records 5a a distance to an object in front of the vehicle and a position coordinate of the object based on the time the distance / angle calculation unit takes 5b needed to capture the reflected light. Since such a radar system is known, it will not be discussed in more detail below.

[0296] In addition to laser beams, the distance / angle measuring device 5 use a radio wave such as a millimeter wave or an ultrasonic wave.

[0297] The vehicle speed sensor 7 acquires a signal corresponding to the rotation speed of the wheels.

[0298] The steering angle sensor 27 detects the amount of change in the steering angle of the steering wheel. If the circuit breaker 29 is turned on, a steering angle memory address of a memory is set to “0”. A relative steering angle θ (wheel) is determined by integrating the amount of change in steering angle that is subsequently detected.

[0299] The yaw rate sensor 28 detects a rate of change Ω (rad / s) in vehicle turning angle (yaw angle) about the vertical axis passing through the vehicle's center of gravity.

[0300] The cruise control switch 26 when turned on, causes cruise control to start, under which a vehicle-to-vehicle control process is also executed. In the vehicle-to-vehicle control process, when the vehicle-to-vehicle distance becomes small, the computer 2A determine that the vehicle is at risk of colliding with the vehicle ahead. In this case, the computer allows it 2A the alarm sound generating unit 13 to sound an alarm. The volume of the alarm sound is controlled by the alarm volume adjustment unit 24 Voted. Further, the sensitivity of the alarm is set by the alarm sensitivity setting unit 25 Voted.

[0301] The brake switch 9 detects the degree of braking by the driver. The brake control unit 19 is with the command of the computer 2A activated when it is necessary to avoid a risk to adjust the brake pressure.

[0302] The throttle position sensor 11 detects the position of the throttle valve of the combustion engine.

[0303] According to the results of detecting the throttle position sensor 11 tells the computer 2A the activation of the throttle control unit 21 to adjust the position of the throttle valve and thus the power output of the engine.

[0304] The sensor error display 17 shows an error of the distance / angle measuring device 5 indicated by a sensor error detection block 109 has been recorded. The distance indicator 15 indicates a distance to a preceding vehicle, which is selected through a process to be described later, based on the results of measurements made by the distance / angle measuring device 5 be obtained.

[0305] The automatic transmission controller 23 selects a gear position of the automatic transmission suitable for controlling the speed of the vehicle based on the command from the computer 2A .

[0306] Referring again to the block diagram of the computer 2A referenced.

[0307] The data related to the distance and angle obtained from the distance / angle calculator 5a the distance / angle measuring device 5 are output by an object detection block 108 converted from a polar coordinate to an orthogonal X-Z coordinate with the vehicle at the centre. On the other hand, a signal from the vehicle speed sensor 7 is output in accordance with the rotation speed of the wheels from a vehicle speed calculation block 110 converted into a vehicle speed signal. Based on the vehicle speed signal and the converted orthogonal X-Z coordinate, the object detection block calculates 108 a center position coordinate (X0, Z0), an object width W, a relative speed (VX0, VZ0), and the type of an object type. The object type shows whether a detected object is a moving object or a stationary object. It should be noted that, in the center position coordinate (X0, Z0) of an object, X0 indicates the position of the object in the vehicle width direction, and Z0 indicates the position of the object in the vehicle traveling direction.

[0308] A steering angle calculation block 112 calculates a steering angle θ based on a signal from the steering angle sensor 27 . A yaw rate calculation block 114calculates a yaw rate Ω based on a signal from the yaw rate sensor 28 .

[0309] A curve radius calculation section 116 , which calculates a radius of curvature R of the road traveled by the vehicle, receives input signals of a vehicle speed from the vehicle speed calculation block 110 , a steering angle θ from the steering angle calculation block 112 and a yaw rate Ω from the yaw rate calculation block 114 . Consequently, the turning radius calculation section calculates 116 a turning radius R based on the vehicle speed, the steering angle θ and the yaw rate Ω.

[0310] A lane probability calculation block 119 calculates a lane probability P of a preceding vehicle. When calculating the lane probability P, the lane probability calculation block uses 119 the curve radius R as well as the center position coordinate (X0, Z0), the object width W, the relative speed (VX0, VZ0) and the object type, which are stored in the object detection block 108 be calculated.

[0311] If the object detection block 108 an object is determined as a reflector provided on the road, a road shape recognition block recognizes 117 the shape of the road based on the center position coordinate (X0, Z0) of the reflector and the curve radius R calculated by the curve radius calculation section 116 is obtained. The details of the road shape recognition will be described later.

[0312] A block 118 for determining a preceding vehicle determines a preceding vehicle based on the turning radius R calculated in the turning radius calculation section 116 is calculated, the lane probability P calculated in the lane probability calculation block 119 is calculated, the center position coordinate (X0, Z0), the relative velocity (VX0, VZ0) and the object type, which are in the object detection block 108 are calculated, and the road shape detected in the road shape recognition block 117 is recognized.

[0313] A control block 120 outputs signals for adjusting a distance between the vehicle and a preceding vehicle to the brake control unit 19 , the throttle control unit 21 and the automatic transmission controller 23 . The signals are generated based on the distance Z0 to the preceding vehicle, the relative speed VZ0 in the traveling direction, the setting conditions of the cruise control switch 26 and the degree of braking applied to the brake switch 9 is applied. Furthermore, there is the control block 120 a signal for the distance indicator 15 is required while sending an alarm signal to the alarm sound generating unit as needed 13 outputs so as to inform the driver of the situation.

[0314] The relative speed VZ0 in the traveling direction, which is the only factor used in executing the vehicle-to-vehicle control, is sent to the control block 120 sent.

[0315] Specifically, a process up to the recognition of a road shape obtained by the computer 2A in the inter-vehicle distance control device 1A is carried out, with reference to in the figure 2 and figure 4 shown flow charts described. The current process is executed repeatedly every second.

[0316] In step S100 of figure 2 reads the distance / angle measuring device 5 distance / angle measurement data, i. H. Distance / angle data (object unit data) for an object in front of the vehicle.

[0317] Subsequently, in step S200, a recognition process for the object in front of the vehicle is executed. In the object recognition process, the distance / angle data obtained from the distance / angle measuring device 5are read, converted from a polar coordinate system to an orthogonal coordinate system. Then, based on the converted distance / angle data, a center position coordinate (X0, Z0) of the object, an object width W0, a relative speed (VX0, VZ0), and an object type are calculated. The relative velocity (VX0, VZ0) of the object is calculated based on the temporal changes in the center position coordinate (X0, Z0). For example, when the relative position of the object hardly moves even though the vehicle is ahead, the object can be determined as a moving object. Also, when a distance to the object gradually increases, the object can be determined as a moving object. On the other hand, when the relative position of the object approaches the vehicle at the same rate (absolute value) of the vehicle speed, the object can be recognized as a stationary object. Other objects, such as an object with an insufficient occurrence duration to complete the recognition, are recognized as unidentified objects.

[0318] In step S300, a turning radius R (estimated R) of the vehicle is calculated based on a steering angle θ detected by the steering angle sensor 27 is obtained, or a yaw rate Ω obtained from the yaw rate sensor 28 is won is calculated. Here, the turning radius R is calculated from a steering angle θ using the following equation (1). R = C / θ (1)

[0319] In the equation (1), C describes a constant that depends on a vehicle type and a vehicle speed. Constants C for respective vehicle types and vehicle speeds are used as mapping functions in the curve radius calculation block 116 of the computer 2A saved. Since the constant C is known as a function for calculating a turning radius R from a steering angle θ, it will not be described in detail below. When a turning radius R is calculated from a yaw rate Ω, a vehicle speed V is divided by a yaw rate Ω.

[0320] Subsequently, in step S310, an approximation R is calculated. In step S300 described above, a turning radius R is calculated based on the state of the vehicle. In the current step, an approximation R is made using the data of a preceding vehicle, i. H. an immediately preceding vehicle, and the vehicle directly in front of the preceding vehicle, d. H. of a second preceding vehicle, which are the moving objects detected in step S200. In particular, an approximation R corresponds to a radius of a circle passing through three points, i.e. H. the coordinate of the immediately preceding vehicle, the coordinate of the second preceding vehicle, and the origin (the vehicle having the device) is approximated. Consequently, an immediately preceding vehicle and a second preceding vehicle are selected from the moving objects detected in step S200.

[0321] In order to select an immediately preceding vehicle and a second preceding vehicle from the moving objects, the following requirements are used. A: lane probability ≥ 70% B: lane probability > 50% and acquisition time ≥ 10 s C: Shortest distance D: Xcross < 2 m E: Located further away than an immediately preceding vehicle

[0322] The lane probability of the requirements A and B is calculated in the lane probability calculation block 119 calculated. A lane probability is a parameter describing a probability that a target travels in the same lane as the device-equipped vehicle. Since the method for calculating a lane probability is known, it will not be discussed in more detail below. The detection time in the requirement B is a period during which a moving object is continuously detected.

[0323] Requirement C indicates a moving object with a shortest distance between the moving object and the device-equipped vehicle. The Xcross of the requirement D is calculated in step S414, which will be described later. Specifically, an estimated X-axis Xcross obtained in step S414 of the previous cycle is read for use in step S310 of the current cycle. The previous cycle herein refers to a cycle immediately before the current cycle in the flowchart of FIG figure 28, wherein the cycle is repeatedly executed in a predetermined cycle.

[0324] In the requirements as set forth above, the moving object that satisfies the following requirements is regarded as an immediately preceding vehicle: (A or B) and C and D

[0325] Further, the moving object that satisfies the following requirements is regarded as a second-preceding vehicle: E and D

[0326] The numerical values ​​of conditions A, B and D are given only as examples and can therefore be determined as appropriate.

[0327] Consequently, as in figure 29, two vehicles are shown traveling in the traffic lane on which the device-equipped vehicle is traveling 180 driving, as the vehicle driving directly ahead 181 and as the second-leading vehicle 182 chosen. Then a circle passing three points, i. H. the vehicle 180 , the vehicle 181 and the vehicle 182 , approximated. The radius of the circle is calculated as an approximation R.

[0328] As described above, a “circle passing three points” is approximated. However, it is not absolutely necessary for the circle to pass through the three points; a circle can only be approximated by three points. When calculating an approximation R based on moving objects, it can be used when another vehicle is different from the immediately preceding vehicle 181 and the second vehicle in front 182 differs, is present on the same lane, the other vehicles can be used to calculate the approximation R. This improves the accuracy of the approximation.

[0329] For example, if, as in figure 29, a vehicle 183 driving in the lane to the right of the lane the vehicle is in 180 is selected as a second-leading vehicle, a circle passing through the broken line in FIG figure 29 is shown. The circle shown by the broken line intersects the X-axis at a position away from the vehicle 180 in the vehicle width direction. Consequently, the accuracy of the approximation R is not high. In this regard, an appropriate selection of the second-preceding vehicle enables 182 , which satisfies the conditions set out above, a calculation of the approximation R with quite high accuracy, as indicated by the solid-line circle in FIG figure 29 shown.

[0330] When the vehicle immediately ahead 181 and the second vehicle in front 182 cannot be chosen among the moving objects, d. H. if the above conditions are not met, the approximation R is not required to be calculated in the current step.

[0331] When the approximation R is calculated as described above, for example, an average between the curve radius R calculated in step S300 and the approximation R calculated in the current step is calculated. Then, the mean value is filtered to detect a curve radius R, which is used for road shape recognition, which will be explained below. Of course, the method of applying an approximation R to a curve radius R as described above is only an example and consequently another method can be used.

[0332] On the other hand, when the immediately preceding vehicle 181 and the like cannot be selected and no approximation R is calculated in the current step, the curve radius R calculated in step S300 is used for road shape recognition. On the other hand, when the curve radius R cannot be calculated in step S300, the approximation R per se calculated in the current step can be used as the curve radius R.

[0333] In the subsequent step S400, road shape recognition is carried out. The details are described below with reference to that in figure 30 shown flowchart described.

[0334] First, in step S410, object unit data which appear to correspond to beacons is extracted. In particular, if the object type specified in the object detection block 108 is detected is determined as a stationary object, those stationary objects having a width of less than or equal to 1 m of the stationary object are extracted. Consequently, the stationary objects such as billboards, which have a relatively large width, are removed to extract only reflectors arranged along the road among the stationary objects. Hereinafter the reflectors are referred to as delineators (reflectors embedded in the roadside, commonly known as "cat's eyes").

[0335] In the subsequent step S412, as in figure 31, a center position coordinate (X0, Z0) of each delineator into a center position coordinate (X1, Z1) with respect to running on a straight road, i. H. of driving straight ahead. After the conversion, the delineators with a high coordinate value X1 are removed.

[0336] Specifically, in converting the position of each delineator into the straight-ahead position, the following equation (2) is used. X1 ← X0 - Z0 × Z0 / 2R Z1 ← Z0 (2)

[0337] The approximation of Equation (2) assumes that |X| << |R|, |X| << Z is executed.

[0338] In the current step, if, for example, the coordinate X1 of a delineator after the straight-road conversion satisfies the condition of the following equation (3), the delineator is excluded from those delineators used for recognizing the shape of the road. |Straight Street Conversion X1| > 3.5m (3)

[0339] According to this condition, a value of an area equivalent to the width of the traffic lane on which the device-equipped vehicle runs is set to, for example, 3.0 m. The lane width equivalent value of 3.0 m is set on both the left and right side of the vehicle. Consequently, the condition for extracting only the delineators on the lane lines of the lane in which the vehicle travels is used.

[0340] Assume that the vehicle is traveling along the middle of a lane 3.5 m wide on a straight road. In this case, the vehicle is located at a position X1 = -1.75 m to the left line of the lane on which the vehicle is running and X1 = -5.25 m to the adjacent lane line. On the other hand, the vehicle is at a position X1=1.75 m to the right line of the lane on which the vehicle is running and X1=5.25 m to the adjacent lane line. In this situation, the condition of the equation (3) removes the delineators whose absolute value of the position coordinate X1 in the vehicle width direction after the straight-road conversion is over 3.5 m. Consequently, only those delineators provided along the lines of the lane in which the vehicle travels are extracted.

[0341] When the turning radius R cannot be calculated in the absence of a steering sensor and a yaw rate sensor, the condition of the following equation (4) is used. Since a calculated turning radius R may have an error, the condition of equation (4) can be combined. |X| > 4.0m (4)

[0342] In the equation (4), the lane width equivalent value is set to 4.0 m. The condition of the equation (4) is provided without considering the curve radius R of the road so that the condition corresponds to the condition before the straight-road conversion. Consequently, the condition of Equation (4) is allowed to have a tolerance compared to the condition of Equation (3).

[0343] In step S414, an estimated X-axis intersection point of each beacon is calculated. It will, as in figure 32 shown, a circle 85 is calculated passing a center position coordinate (X0, Z0) of a delineator and, as a tangent vector, a relative velocity vector 80 of the delineator with respect to the vehicle. The estimated x-axis intercept corresponds to an intersection between the circle 85 and the vehicle width direction, i. H. the X-axis, the origin of which is the vehicle. In a calculation of the estimated X-axis intercept, an approximate calculation is performed using the following equations (5) to (8).

[0344] If a circle is approximated by a parabola, assuming that |X| << |R|, |X| << Z, an equation of the circle (described as a function of X and Z) passing through the center of a delineator and perpendicular to the X-axis is described as follows. X = X0 + {(Z - Z0) × (Z - Z0) / 2R} (5)

[0345] A relative velocity vector (VX, VZ) of a delineator, which is a tangent vector, is described as follows. dX / dZ = VX0 / VZ0 (6)

[0346] From Equations (5) and (6), a turning radius R is described as follows. R = (Z - Z0) × VZ / VX (7)

[0347] Consequently, the equation of the circle is transformed as follows. X = X0 + {(Z - Z0) × VX / 2VZ} (8)

[0348] In this case, when Z=0, the value of X corresponds to an estimated X-axis intersection Xcross, which value is described as follows. X = X - Z × VX0 / 2VZ0 (9)

[0349] In this way, the estimated X-axis intersection point Xcross is calculated.

[0350] In calculating the estimated X-axis cross point Xcross, an area in the width direction of the vehicle that cannot be detected by a radar system can be defined. As a result, the course of the road is recognized more accurately. At the same time, the left and right edges of the lane on which the vehicle is traveling are also recognized as follows.

[0351] The estimated X-axis intersection point Xcross is calculated for each of the delineators that remain and are not removed by equations (3) and (4) above. Finally, however, a delineator with a minimum distance (Z0) in the direction of travel of the vehicle for both the left and right sides of the vehicle is selected and used for the following processing.

[0352] Subsequently, in step S416, the left and right edges of the lane on which the vehicle is traveling are recognized using the results up to step S414. First, the signs of the estimated X-axis intersection points Xcross are grouped into a positive group and a negative group. The positive group is recognized as the delineators along the right edge of the lane, while the negative group is recognized as the delineators along the left edge of the lane.

[0353] Subsequently, for both the left and right edges of the traffic lane, the delineators that remain and are not removed by the straight-road conversion in step S412 are respectively calculated via the respective center position coordinates (X0, Z0) in front of the straight-road transformation, in order to recognize the course of the road.

[0354] The present embodiment has been described by taking the case where a row of delineators on both the left and right sides of the vehicle are detected as an example. However, the present invention is not limited to this. For example, a series of delineators may be detected on either the left or right side of the road.

[0355] As described above, according to the road shape recognition device for vehicles of the present embodiment, the following advantages are brought about.

[0356] Stationary objects with a width over a predetermined value, such as 1 m, are removed to extract only the reflectors located along the road. As a result, most of the unnecessary vehicles, signs, billboards and the like are removed, leaving only the delineators 110 to extract. delineator lines 150 and 151 can, as in figure 33 are present on the left side of the vehicle, and delineator lines 160 and 161 can, as in figure 33 may be present on the right side of the vehicle. Even in such a situation, the delineator lines 151 and 161 removed with the lines outside a range defined by a vehicle width equivalent value with respect to the position of the vehicle 180 available. Consequently, only the delineator lines 150 and 160 that exist at the edges of the lane are defined. In this way the delineators 110 who are on different lines 150 and 151 are present are not incorrectly recognized as being present on the same line of delineators, in order to correctly recognize the course of the road.

[0357] Further, a turning radius R of the road on which the vehicle is running is calculated based on the steering angle θ detected by the steering angle sensor 27 is detected and the yaw rate Ω detected by the yaw rate sensor 28 is detected, calculated. Then, based on the curve radius R, the center position coordinate (X0, Z0) of each delineator becomes 110 into the center position coordinate (X1, Z1) with respect to straight running, i.e. H. driving on a straight road, transformed. Subsequently, by the delineators 110 , which result from the transformation, those delineators 110 , which is outside a range defined by a vehicle width equivalent value with respect to the position of the vehicle 180 present are removed in order to recognize the edges of the lane and the course of the road.

[0358] As a result, when the vehicle is cornering, the lines of delineators are prevented from being erroneously recognized by defining the lines of delineators and thus correctly recognizing the course of the road.

[0359] Further, an intersection point in the vehicle width direction is obtained for each delineator line by calculating the estimated X-axis intersection points Xcross. As a result, the delineator lines can be defined in an area that is out of the radar system's detection range, enabling more accurate road shape recognition.

[0360] In the present embodiment, when an approximation R is made in step S310 based on moving objects such as the vehicle immediately ahead 181 , is calculated, a curve radius R is obtained based on the approximation R, and then road shape recognition using the curve radius R is performed. In this way, the estimated X-axis intersection points Xcross are calculated using not only stationary objects but moving objects as well. Consequently, the road shape is calculated more accurately and with good frequency when it is difficult to detect roadside objects such as delineators, such as the vehicle immediately ahead 181 exists in front of the device-equipped vehicle, or when originally only a small number of roadside objects exist along the road on which the device-equipped vehicle is running.

[0361] As for the correspondence between the description of the present embodiment and the claims, the distance / angle measuring device 5 corresponds to the radar means and corresponds to the object recognition block 108 and the road course recognition block 117 the detection means of the present invention. The object detection block108 however, corresponds to the object detecting means, the preceding vehicle extracting means and the reflector extracting means, and the road shape detecting block 117 corresponds to the approximation radius calculation means and the road shape recognition means. [Seventh Embodiment]

[0362] A road shape recognition method and a road shape recognition device for vehicles according to a seventh embodiment will be described below with reference to FIG figure 34 described.

[0363] The one in the figure 27 shown inter-vehicle control device 1A will be further developed. In the device, a road shape recognition process according to a in the figure 34 shown flowchart executed.

[0364] Specifically, in step S420 the figure 34, Objects that look like delineators, i. H. Objects that are candidate delineators extracted using delineators of the previous cycle and the current cycle. This is the same as step S410 of the sixth embodiment.

[0365] In step S421, an estimated X-axis intersection point Xcross of a delineator line is calculated. In this case, a calculation of the estimated X-axis intersection points Xcross the delineators 110 that are present in the same line of delineators are grouped together in the vicinity of a given section. However, statistical processing is performed for such a portion, and a single representative estimated X-axis cross point Xcross is calculated. By calculating such a representative estimated X-axis intersection point Xcross, it can be determined whether or not there are multiple delineator lines, as will be described later. The calculation of the estimated X-axis intercept is the same as that in step S414 of the sixth embodiment.

[0366] In subsequent step S422, it is determined whether or not there are plural lines of guideposts on both the left and right sides of the vehicle.

[0367] When a plurality of representative estimated X-axis intersection points Xcross are calculated in step S421 on both the right and left sides of the vehicle, the delineator line is determined as a plurality and the processing proceeds to step S423.

[0368] On the other hand, when it is determined that there is a single representative estimated X-axis cross point Xcross on both the right and left sides of the vehicle, the processing proceeds to step S425 to recognize the road shape.

[0369] Specifically, since a line of delineators is detected on the right or left side of the vehicle, road shape recognition similar to that in step S416 of the first embodiment is performed based on the detected delineators.

[0370] In step S423, a lane width is calculated. Specifically, when a plurality of delineator lines are determined to be present on both the right and left sides of the vehicle, the lane width is calculated from an interval between adjacent representative estimated X-axis intersection points Xcross. In this case, the lane width can be calculated from an interval between the estimated X-axis cross points Xcross present at the edges of the lane on which the vehicle is traveling.

[0371] In the subsequent step S424, a lane width equivalent value is calculated based on the lane width calculated in step S423 with respect to the delineators subjected to the straight-road conversion in the first embodiment. Subsequently, the delineators are removed at positions above the vehicle width equivalent value. When the vehicle width equivalent value is 3.0m, the delineators at positions over 3.0m from the position of the vehicle 180 away.

[0372] Then, in step S425, the right and left edges of the traffic lane are recognized based on the delineator lines, which are not removed in step S424, so as to recognize the road shape. The road shape recognition is performed in a manner similar to that in step S416 of the first embodiment.

[0373] In the present embodiment, it is determined whether there are multiple delineator lines on both the right and left sides of the vehicle. However, the present invention is not limited to this, but it can be determined whether there are a plurality of delineator lines on either the right or left side of the vehicle.

[0374] In step S422, a plurality of delineator lines are not necessarily detected continuously. Thus, for example, once there are multiple representative estimated X-axis intersection points Xcross within the last three seconds, a delineator line may be determined to be in the plurality.

[0375] As described above, according to the present embodiment, the road shape is correctly recognized despite the change in the lane width.

[0376] For example, when a vehicle is running on an ordinary road, a predetermined area corresponding to the lane of the ordinary road related to the vehicle can be determined, and then delineator lines present in the predetermined area can be extracted. In this case, when the vehicle enters a freeway, the lane width on the freeway will increase so that the delineator that is closest to the vehicle is now present outside the predetermined area. Consequently, it may be difficult to recognize the course of the road in this situation. In this regard, in the present embodiment, a lane width is calculated and then a predetermined range is determined in accordance with the lane width. Consequently, in the above situation, the course of the road is correctly recognized. [Eighth Embodiment]

[0377] A road shape recognition method and a road shape recognition device for vehicles according to an eighth embodiment will be described below with reference to FIG figure 35 described.

[0378] The one in the figure 27 shown inter-vehicle control device 1A will be further developed. In this device, the process of recognizing a road shape is performed in accordance with a figure 35 shown flowchart executed.

[0379] Specifically, in step S430 the figure 35, objects that look like delineators extracted. This extraction also uses the delineator data from the previous cycle. This is the same as step S410 of the sixth embodiment.

[0380] Subsequently, in step S432, an estimated X-axis intersection point Xcross of the delineator line is calculated. This is the same as step S432 of the second embodiment.

[0381] In subsequent step S434, it is determined whether or not there are plural lines of delineators on the left side of the road.

[0382] In making a determination, the delineators with estimated X-axis intersections Xcross of negative sign are considered to be on the left side of the road. In this case, maximum and minimum values ​​of the estimated X-axis intersection points Xcross are calculated. If the difference is greater than or equal to the vehicle width equivalent value of 3.0 m, a delineator line is determined as a plurality and the processing proceeds to step S436.

[0383] In step S436, the delineators 110 with a high absolute value estimated X-axis intersection point Xcross located on the left side of the road. In this case, a step similar to step S412 of the first embodiment is executed.

[0384] Subsequently, in step S438 similar to step S434, it is determined whether or not there are a plurality of delineator lines on the right side of the road. If it is determined that there are multiple lines of delineators, processing proceeds to step S440, in which the delineators 110with a high absolute value estimated X-axis intersection point Xcross on the right side of the road.

[0385] In the subsequent step S442, the left and right edges of the traffic lane are recognized in order to recognize the course of the road. The manner of recognizing the edges of the lane on which the vehicle is traveling is similar to step S416 of the first embodiment.

[0386] In step S434 or S438, a plurality of guidepost lines are not necessarily detected continuously. Thus, for example, once the difference between the maximum and minimum values ​​of the estimated X-axis intersection points Xcross is greater than or equal to the vehicle width equivalent value within the last three seconds, it can be determined that multiple delineator lines are present. [Ninth Embodiment]

[0387] A road shape recognition method and a road shape recognition device for vehicles according to a ninth embodiment will be described below with reference to FIG figure 36 described.

[0388] The one in the figure 27 shown inter-vehicle control device 1A will be further developed. In this device, the process of recognizing a road shape is performed in accordance with a figure 36 shown flowchart executed.

[0389] Specifically, in step S450 the figure 36, Objects that look like delineators extracted. This extraction also uses the delineator data from the previous cycle. This is the same as step S410 of the sixth embodiment.

[0390] In the subsequent step S452, an estimated X-axis intersection point Xcross of each delineator post is calculated. This calculation is the same as step S414 of the first embodiment.

[0391] Subsequently, in step S454, of the marker posts on the left side of the road, those having an estimated X-axis cross point Xcross of a high absolute value are removed. In this case, delineators with an estimated X-axis intersection point Xcross of a negative sign are considered a group of delineators on the left side of the road. Then, in the group on the left side of the road, a minimum absolute value of the estimated X-axis cross point Xcross is calculated. For example, those delineators that meet the condition of equation (9) below are removed because they have an estimated X-axis crosspoint Xcross of a high absolute value. |estimated X-axis intersection Xcross| > Minimum value of |estimated X-axis intersection Xcross| + 2.0m (9)

[0392] Subsequently, in step S456 equal to step S454, beacons having an estimated X-axis cross point Xcross of a high absolute value are removed from the beacons on the right side of the road, and the processing proceeds to step S458.

[0393] In the subsequent step S458, the left and right edges of the traffic lane are recognized so as to recognize the course of the road. The recognition method is the same as that of the first embodiment.

[0394] In step S454 or S456, multiple beacons are not necessarily detected continuously. Consequently, in a calculation of a minimum absolute value of the estimated X-axis intersection points Xcross, for example, the minimum value detected within the last three seconds can be used.

[0395] A minimum value of the estimated X-axis intersection points Xcross is calculated as described above. A value smaller than the lane width (such as 2.0 m) is then added to the minimum value. The delineator line 151 at a position greater than the position resulting from the addition is removed, around the remaining delineator line 150 to extract. Consequently, only the delineator line can 150 , the closest to the vehicle 180 is arranged, are defined, so that a detection of the road is made possible. [Other embodiments]

[0396] The above-described embodiments each show a configuration example of the road shape recognition device and the road shape recognition method. The structure and method are not limited to the details described above, but can be modified in various ways without departing from the scope of the present invention. For example, step S310 for calculating an approximation R does not necessarily have to be carried out after step S300, but can be carried out at any stage in a flow chart, such as that in FIG figure 30, are executed. [Tenth embodiment]

[0397] A road shape recognition method and a road shape recognition device for vehicles according to a tenth embodiment will be described below with reference to FIG figure 37 and figure 38 as well as the figure 27, figure 30, figure 31, figure 33 and figure 34 described.

[0398] The one in the figure 27 shown inter-vehicle control device 1A will be further developed. In this device, the road shape recognition process is performed in accordance with a figure 37 is executed. The rest of the structure is the same or similar to that in the figure 27 structure shown.

[0399] In particular, a process up to the recognition of a road shape, which is performed by the computer 2 in the vehicle-to-vehicle control device 1A is carried out, with reference to in the figure 37 and figure 30 shown flow charts described. The running process is executed repeatedly every 0.1s.

[0400] First, the distance / angle measuring device reads 5 in step S100 of figure 37 Measurement data (object unit data) of the distance / angle between the vehicle and an object in front of the vehicle.

[0401] Then, in step S200, a process for recognizing the object is performed. This process is the same as that described above.

[0402] In subsequent step S300, a turning radius R (estimated R) of the road on which the vehicle travels is calculated based on a steering angle θ detected by the steering sensor 27 is obtained, or a yaw rate Ω obtained from the yaw rate sensor 28 is obtained is calculated. This process is also the same as that described above.

[0403] Subsequently, in the subsequent step S400, road course recognition is carried out. The details are given with reference to that in the figure 30 shown flowchart described.

[0404] First, in step S410, object unit data of objects looking like beacons are extracted. This process is also the same as that described above.

[0405] The current step also uses the data in the vicinity of the estimated road course-R obtained in step 414 of the previous cycle, among the data on the objects (reflectors) looking like beacons extracted in step S410 of the previous cycle. More specifically, from the data of the previous cycle, the data that can be used in the current cycle is used. The "previous cycle" refers to a cycle immediately before the current cycle in the flowchart of FIG figure 2, wherein the cycle is repeatedly executed in a predetermined cycle.

[0406] In short, in step S414, which will be described later, the circle passing through the delineator is approximated by a parabola to calculate a radius of curvature. Then, a point where the circle crosses the X-axis is calculated as an estimated X-axis cross point Xcross from the curve radius. This curve radius is the “estimated road course R”.

[0407] Further, the “close range” in the expression in the close range of the estimated road shape R refers to a predetermined range obtained by increasing and decreasing the R by a predetermined value a in the radial direction. More specifically, the zone range of R ± α corresponds to, for example, as in figure 38, the vicinity of the estimated road course-R. In the predetermined area in the figure 38, a mark “•” indicates the data of the previous cycle and a mark “o” indicates the data of the current cycle. The data of the previous cycle outside the predetermined range will be saved as in figure 38, not used in the current cycle. Consequently, the accuracy in road shape recognition is prevented from deteriorating.

[0408] Accordingly, in the current step, from the data of the previous cycle, the data of beacons located within the predetermined area based on the estimated road shape-R is added to the data of the current cycle. Thus, in the processes after and after step S410 including step S410, beacon data substantially corresponding to two cycles is handled. Consequently, the amount of usable data increases, so that the occurrence frequency of beacons is increased.

[0409] In the subsequent step S412, as in figure 31, the center position coordinate (X0, Z0) of each delineator into the center position coordinate (X1, Z1) with respect to going straight, i. H. Driving on a straight road, transformed. Then those delineators with a high coordinate value X1 are removed after the conversion.

[0410] Specifically, the conversion of the position of each delineator into a position related to going straight is calculated from the following equation (2). X1 ← X0 - Z0 × Z0 / 2R Z1 ← Z0 (2)

[0411] Equation (2) is approximated assuming that |X| << |R|, |X| << Z

[0412] In the current step, for example, when the coordinate X1 after the straight-road conversion of each delineator satisfies the condition of the following equation (3), the delineator is removed from the delineators used for recognizing the shape of the road. |Straight Street Conversion X1| > 3.5m (3)

[0413] This condition is used, for example, to set a vehicle width equivalent value to 3.0 m on both the left and right sides of the vehicle to define an area equivalent to the lane width with respect to the vehicle and only the delineators at the edges of those traveled by the vehicle extract lane.

[0414] When the vehicle is running in the middle of a lane with a width of 3.5 m on a straight road, the vehicle is at X1=−1.75 m from the edge on the left side of the vehicle and at X1=−5.25 m from the edge of the adjacent lane. Further, the vehicle is located at X1=1.75 m from the edge on the right side of the vehicle and at X1=5.25 m from the edge of the adjacent lane. The condition of the equation (3) allows a removal of the delineator posts with the position coordinate X1 of an absolute value over 3.5 m in the vehicle width direction after the straight-road conversion. As a result, only the delineators at the edges of the vehicle's lane are extracted.

[0415] When the turning radius R cannot be calculated due to lack of a steering sensor and a yaw rate sensor, the condition of the following equation (4) is used. Further, since a calculated turning radius R may have an error, the condition of the equation (4) can be combined. |X| > 4.0m (4)

[0416] In the equation (4), the lane width equivalent value is set to 4.0 m. The condition of the equation (4) is provided without considering the curve radius R of the road so that the condition corresponds to the condition before the straight-road conversion. Consequently, the condition of Equation (4) is allowed to have a tolerance compared to the condition of Equation (3).

[0417] Then, in step S414, an estimated X-axis intersection point of each delineator post is calculated. The estimated x-axis intercept refers to an intersection between a calculated circle 85 and the vehicle width direction, i. H. the X-axis, the origin of which is the vehicle. The calculated circle 85 passes the center coordinate (X0, Z0) of the delineator, as in figure 33 shown and has a vector 80 as a tangent vector describing a speed of the delineator with respect to the vehicle.

[0418] The radius R of the circle 85 corresponds to the estimated road course R. More specifically, the estimated road course R is an R calculated using a delineator (reflector) which is a stationary object.

[0419] Similar to the above case, an approximation calculation is applied to the calculation of the estimated X-axis intercept. When the estimated X-axis cross point Xcross is calculated, an area in the width direction of the vehicle where the radar system cannot detect can be defined. This enables a more precise recognition of the course of the road and at the same time a recognition of the left and right edge of the lane in which the vehicle is driving.

[0420] The estimated X-axis intersection point Xcross is calculated for each of the delineators that remain and are not removed by conditions (3) and (4) above. Finally, the delineators chosen on both the left and right sides of the vehicle have a minimum spacing (Z0) in the vehicle's forward direction. The chosen delineators will be used for the following process.

[0421] In step S416, the left and right edges of the lane on which the vehicle is traveling are recognized using the results up to step S414. First, the signs of the estimated X-axis cross points Xcross calculated in step S414 are divided into two groups, i. H. a positive group and a negative group. The positive group markers are recognized as those belonging to the right edge of the lane, while the negative group markers are recognized as those belonging to the left edge of the lane.

[0422] At both the left and right edges of the traffic lane, the markers that remain and are not removed by the straight-road conversion in step S412 are respectively connected via the respective center position coordinates (X0, Z0) before the straight-road conversion, to see the course of the road.

[0423] In this case as well, the data of the beacons detected in the previous cycle in step S410 is used. Consequently, the number of beacons connected in the current cycle is higher than the number of beacons that would be detected in one cycle. More specifically, the occurrence frequency of delineators will be increased. In this way, the course of the road is recognized more accurately.

[0424] The present embodiment shows, by way of example, that delineator post lines are detected both on the right side and on the left side of the vehicle. However, the present invention is not limited to this, but delineator lines can be detected either on the right side or on the left side of the road.

[0425] As described above, the road shape recognition device for vehicles according to the present embodiment has the following advantages.

[0426] Stationary objects with a width above a predetermined value, i. H. 1m, are removed to extract only the reflectors that are placed on the road. Consequently, most of the vehicles, signs, billboards and the like are removed and only become the delineators 110 extracted.

[0427] If, as in figure 34 shown, delineator lines 150 and 151 on the left side of the vehicle and delineator lines 160 and 161 on the right side of the vehicle are the delineator lines 151 and 161removed that are outside the range defined by a lane width equivalent value with respect to the position of the vehicle 180 available. Consequently, only the delineator lines 150 and 160 defined, which are present at the edges of the lane traveled by the vehicle. In this way the delineators 110 standing on different delineator lines 150 and 151 are present are not erroneously recognized as being present on the same line of delineators. In this way, the course of the road is recognized more accurately.

[0428] A turning radius R of the turning curve of the vehicle is calculated from a steering angle θ detected by the steering angle sensor 27 is detected and the yaw rate Ω detected by the yaw rate sensor 28 is detected, calculated. Then, based on the curve radius R, the center position coordinate (X0, Z0) of each delineator becomes 110 into the center position coordinate (X1, Z1) with respect to straight running, i.e. H. driving on a straight road, transformed. Then some delineators 110 from the delineators 110 removed after conversion. These delineators to be removed are at positions outside the range defined by the vehicle width equivalent value with respect to the position of the vehicle 180 present. In this way, the edges of the lane and consequently the course of the road are recognized.

[0429] As a result, when the vehicle is cornering, the course of the road is correctly recognized, preventing erroneous detection when defining the delineator lines.

[0430] Further, calculation of each of the estimated X-axis cross points Xcross results in calculation of a cross point between each delineator line and the vehicle width direction. Consequently, the delineator line is also defined in an area where the radar system cannot detect. In this way, the course of the road is recognized more accurately.

[0431] The present invention has a feature that when executing the road shape recognition, a part of the guideposts extracted in the previous cycle is used in addition to the guideposts of the current cycle. The part of the beacons extracted in the previous cycle corresponds to the data in the vicinity of the estimated road course-R calculated in the previous cycle. Consequently, in a situation where an absolute number of beacons that can be used for road shape recognition is small, a situation where there are a number of beacons is created. The situation where an absolute number of delineators is small may be, for example, the situation where it is difficult to detect delineators since there is a vehicle ahead, or the situation where the number of delineators of the road, on which the vehicle runs is originally small. More specifically, the occurrence frequency of delineators is increased to increase the amount of data usable for calculation in road shape recognition. In this way, a correct course of the road is calculated with a good frequency.

[0432] Further, it is not the case that all of the data of the beacons in the previous cycle are only used unconditionally, but the previous cycle data used in the present embodiment correspond to the beacons of the previous cycle which are within a predetermined range with respect to of the estimated road course R calculated in the previous cycle. Consequently, the data that does not match the road shape is removed from the road shape detection. This means that the course of the road is recognized more correctly.

[0433] As for the correspondence between the description of the present embodiment and the claims, the distance / angle measuring device corresponds 5 the radar means and correspond to the object detection block 108 and the road course recognition block 117 the detection means of the present invention. Of these, the object detection block corresponds 108however, the object recognizing means, the extracting means and the data adding means, and corresponds to the road shape recognizing block 117 the roadside detection device.

[0434] Also corresponds to that of the object detection block 108 The process executed is the acquisition process, the extraction process, and the data addition process, and is the same as that of the road shape recognition block 117 running process the detection process. In addition, the estimated road course R corresponds to the estimated road course curve radius of the present invention. [Other embodiments]

[0435] The tenth embodiment described above is only an example of the configuration of the inter-vehicle distance control device 1 and the road course recognition method. The structure and method are not limited to the details described above, but can be modified in various ways without departing from the scope of the present invention. For example, in the above embodiment, when adding the object unit data of reflectors extracted in the previous cycle to the object unit data of reflectors of the current cycle, the reflectors included in a range R ± α as the near range of the estimated road course R, has been selected. However, this is only an example of the “near range of the estimated road shape-R”. Instead, a predetermined range in the radial direction may be appropriately set with respect to R, such as a range of R+α or a range of R−α.

[0436] Further, the tenth embodiment described above may be modified in accordance with the flow charts described in the seventh, eighth and ninth embodiments shown above. More specifically, the road shape recognition process shown in the flowchart of FIG figure 34, figure 35 or figure 36 can be applied to the tenth embodiment.

[0437] The embodiments described above use a distance / angle measuring device using laser beams as the “radar means”. Alternatively, as already described above, a millimeter wave, for example, can be used. For example, when an FMCW radar or a Doppler radar with a millimeter wave is used, information on the distance to a vehicle ahead from a reflected wave (received wave) and information on the relative speed of the vehicle ahead are obtained at once. This thus eliminates the calculation of a relative speed based on the distance information. QUOTES INCLUDED IN DESCRIPTION

[0438] This list of the documents cited by the applicant was generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Patent Literature Cited

[0439] JP 3417375B

[0003] JP 3427809B

[0009] JP 2001-328451A

[0012]

Claims

[1] A road course detection method for vehicles, for calculating a lane probability, which is a probability that a detectable object is located in the same lane as a vehicle applying the method, based on a curvature of a road on which the vehicle is traveling and a position of the object relative to the vehicle, wherein the curvature is calculated from a direction change state of the vehicle and a speed of the vehicle, the method comprising the following steps: – Determining a probability that the object is located in the same lane as the vehicle, in accordance with a detected road course in front of the vehicle and a degree of detection, wherein the road course is detected in accordance with a distance to the object and an angle to the object in a vehicle width direction; – Calculating a correction value used to correct the lane probability in accordance with the determination; and – Determine whether there is a difference between the curvature of the road on which the vehicle is traveling and a curvature of the detected road path, in order to correct the lane probability using the correction value if there is no difference, or not to correct the lane probability using the correction value if there is a difference. [2] Road course detection device for vehicles, which includes: – a direction change detection device ( 27 , 28 , 49 and 51 ) for detecting a change of direction state of a vehicle equipped with the device; – a curvature calculation tool ( 63) to calculate the curvature of a road on which the vehicle is traveling, from a direction change state of the vehicle, calculated by the direction change detection device, and a speed of the vehicle; – a radar device ( 5 ) for detecting a distance to a reflector and an angle to the reflector in a vehicle width direction based on a reflected wave, which is a reflection of a transmitted wave that is emitted over a predetermined angular range in the vehicle width direction; – an object recognition device ( 43 ) to calculate a relative position of the object based on the distance detected by the radar and the angle in the vehicle's width direction; and – a lane probability calculation tool ( 53) for calculating a lane probability, which is a probability that the object is in the same lane as the vehicle, based on the curvature of the road and the relative position of the object, wherein the curvature is calculated by the curvature calculation means and the relative position is calculated by the object detection means, wherein – the object recognition means is designed to determine an object type as to whether each of the objects is a moving object or a stationary object, in accordance with a relative speed of the object and a speed of the vehicle, the device further comprising: – a road course detection device ( 45) for the detection of a road course in accordance with extracted stationary object data that are effective for the detection of a road course, while the stationary object data are extracted using the relative position of the object and the object type obtained from the object detection means; – a lane matching device ( 53 ) to determine the probability that the object is located in the same lane as the vehicle, in accordance with the road course detected by the road course recognition device and a degree of detection; and – a correction value calculation tool ( 53 ) to calculate a correction value used to correct the lane probability in accordance with the determination made by the lane matching determining means, and – the lane probability calculation means determines whether there is a difference between a curvature of the road on which the vehicle is traveling and a curvature of the detected road path, in order to correct the lane probability using the correction value calculated by the correction value calculation means if it is determined that there is no difference, or not to correct the lane probability using the correction value calculated by the correction value calculation means if it is determined that there is a difference. [3] A computer-readable storage medium that stores programs for operating a computer system as the curvature calculation means, the object recognition means, the lane probability calculation means, a means for selecting a vehicle ahead, the road course detection means, the lane conformity determination means and the correction value calculation means of the road course detection device for vehicles according to claim 2. [4] Road course detection method for vehicles, for detecting a road course around a vehicle applying the method on the basis of a reflected wave which is a reflection of a transmitted wave which is emitted over a predetermined angular range in a vehicle width direction, the method comprising the following steps: – Executing a data acquisition process to acquire object unit data with respect to an angle in a vehicle width direction based on the reflected wave, wherein the object unit data have at least a distance to an object; – Performing an extraction process to extract the object unit data effective for detecting a road course, in accordance with a determination regarding an object type as to whether each of the objects is a moving object or a stationary object, while the determination is made in accordance with a relative velocity of the object obtained on the basis of the reflected wave and a velocity of the vehicle; – Executing a recognition process to detect a roadside in accordance with data from a roadside object group formed by grouping data that exhibit a connection requirement of a monotonically increasing distance, while the grouping is performed in directions both clockwise and counterclockwise based on the object unit data extracted in the extraction process; – Repeatedly executing a sequence of the capture process, the extraction process, and the recognition process in a predetermined cycle; – Executing a data addition process to add the object unit data obtained in the extraction process of the previous cycle to the object unit data obtained in the extraction process of the current cycle, with the data addition process being executed after the extraction process; and – Executing a recognition process to detect the roadside in accordance with the object unit data obtained in the data addition process. [5] Road course detection device for vehicles, which includes: – a radar device ( 5 ) for detecting an object based on a reflected wave, which is a reflection of a transmitted wave emitted over a predetermined angular range in a vehicle width direction; and – a means of identification ( 41 , 43 and 45 ) for detecting a road course in front of a vehicle equipped with the device in accordance with the detection by the radar means ( 5 ) is executed, whereby – the radar device ( 5) Object unit data that have at least a distance to an object, with respect to an angle to the object in a vehicle width direction, are acquired based on the reflected wave, – the recognition device ( 41 , 43 and 45 ) shows: – an object recognition device ( 41 and 43 ) to determine an object type as to whether each of the objects is a moving object or a stationary object, in accordance with a relative velocity of an object obtained on the basis of the reflected wave and a velocity of the vehicle; – an effective data extraction tool ( 45 ) for extracting object unit data that are effective for detecting a road course, wherein the extraction is based on the detection provided by the object detection means ( 41 and 43 ) is executed; – a roadside object group data creation tool ( 45 ) for forming data of a roadside object group by grouping data that satisfy at least one connection requirement of a monotonically increasing distance, while the grouping is performed in directions both clockwise and counterclockwise, based on the object unit data provided by the effective data extraction means ( 45 ) are extracted; and – a roadside detection device ( 41 , 43 and 45 ) to detect a road edge in accordance with the data of the road edge object group in both directions defined by the road edge object group data generation tool ( 45 ) are formed, whereby – a consequence of processes of the radar medium ( 5 ) and the means of identification ( 41 , 43 and 45) is designed to be executed repeatedly in a predetermined cycle, – the recognition device ( 41 , 43 and 45 ) furthermore, a data addition tool ( 45 ) has, to add the object unit data that is provided by the effective data extraction tool ( 45 ) extracted in the previous cycle, to the object unit data extracted by the effective data extraction tool ( 45 ) are extracted in the current cycle, with the addition being performed after the effective data extraction process ( 45 ) has been carried out; and – the roadside detection device ( 41 , 43 and 45 ) is designed to detect the roadside in accordance with the object unit data provided by the data addition tool ( 45 ) will be received. [6] Computer-readable storage medium that stores a program for operating a computer system as the recognition means of the road course recognition device for vehicles according to claim 5. [7] Road course detection method for vehicles, for detecting a road course around a vehicle applying the method on the basis of a reflected wave which is a reflection of a transmitted wave which is emitted over a predetermined angular range in a vehicle width direction, the method comprising the following steps: – Acquisition of object unit data that have at least a distance to an object, with respect to an angle of the object in a vehicle width direction based on the reflected wave; – Extracting the object unit data effective for detecting a road course, in accordance with a determination regarding an object type as to whether each of the objects is a moving object or a stationary object, while the determination is made in accordance with a relative velocity of the object obtained on the basis of the reflected wave and a velocity of the vehicle; – Removal of data corresponding to stationary objects on a traveled road from extracted object unit data, if stationary objects are present either on the traveled road between the vehicle applying the procedure and a vehicle immediately in front, or on the traveled road between the vehicle immediately in front and a vehicle second in front; and – Detecting a road edge in accordance with data from a road edge object group formed by grouping data that exhibit a connection requirement of a monotonically increasing distance, while the grouping is performed in directions both clockwise and counterclockwise, using object unit data that have been removed with data corresponding to the stationary objects on the traversed road. [8] Road course detection device for vehicles, which includes: – a radar device ( 5 ) for detecting an object based on a reflected wave, which is a reflection of a transmitted wave emitted over a predetermined angular range in a vehicle width direction; and – a means of identification ( 41 , 43 and 45) for detecting a road course in front of a vehicle equipped with the device in accordance with the detection by the radar means ( 5 ) is executed, whereby – the radar device ( 5 ) Object unit data that have at least a distance to an object, with respect to an angle to the object in a vehicle width direction, are acquired based on the reflected wave, – the recognition device ( 41 , 43 and 45 ) shows: – an object recognition device ( 41 and 43 ) to determine an object type as to whether each of the objects is a moving object or a stationary object, in accordance with a relative velocity of an object obtained on the basis of the reflected wave and a velocity of the vehicle; – an effective data extraction tool ( 45) for extracting object unit data that are effective for detecting a road course, wherein the extraction is based on the detection provided by the object detection means ( 41 and 43 ) is executed; – a stationary object removal agent ( 45 ) to remove data corresponding to stationary objects on a busy road from extracted object unit data, if stationary objects are present either on the busy road between the vehicle having the device and a vehicle immediately in front of it, or on the busy road between the vehicle immediately in front and a vehicle second in front; – a roadside object group data creation tool ( 45) to form data of a roadside object group by grouping data that satisfy at least one connection requirement of a monotonically increasing distance, while the grouping is performed in directions both clockwise and counterclockwise, using the object unit data derived from the stationary object distance mean ( 45 ) are obtained; and – a roadside detection device ( 41 , 43 and 45 ) to detect a road edge in accordance with the data of the road edge object group in both directions defined by the road edge object group data generation tool ( 45 ) are formed. [9] Computer-readable storage medium that stores a program for operating a computer system as the recognition means of the road course recognition device for vehicles according to claim 8. [10] Road course detection method for vehicles, for detecting a road course around a vehicle applying the method on the basis of a reflected wave which is a reflection of a transmitted wave which is emitted over a predetermined angular range in a vehicle width direction, the method comprising the following steps: – Acquisition of object unit data that have at least a distance to an object, with respect to an angle of the object in a vehicle width direction based on the reflected wave; – Extracting the object unit data effective for detecting a road course, in accordance with a determination regarding an object type as to whether each of the objects is a moving object or a stationary object, while the determination is made in accordance with a relative velocity of the object obtained on the basis of the reflected wave and a velocity of the vehicle; – Determining a smallest stationary object as a starting point, wherein the smallest stationary object is located within a range of a predetermined distance from a lateral position of a stationary object that is located closest to the vehicle in the vehicle width direction, wherein the smallest stationary object is also located at a position with a shortest direct distance from the vehicle, while the lateral position of the stationary object that is located closest to the vehicle in the vehicle width direction is extracted from the extracted object unit data; and – Detecting a roadside in accordance with data from a roadside object group formed by joining and grouping data starting from the starting point, wherein the data to be grouped have a joining requirement of a monotonically increasing distance, while the grouping is performed in directions both clockwise and counterclockwise. [11] Road course recognition method for vehicles according to claim 10, characterized in that the data of the roadside object group is formed by: – Defining a first interconnection requirement area (a) and a second interconnection requirement area (b) that is contained within the first interconnection requirement area and is smaller than the first interconnection requirement area; – Connecting a stationary object to the starting point as a base point, where the stationary object is included in both the first and second connection requirement areas; – Connecting the connected stationary object as a nearest base point to another stationary object that is contained in both the first and second connection requirement areas; and – Grouping stationary objects by repeating the connection in this way. [12] Road course recognition method for vehicles according to claim 10 or 11, characterized in that the road edge is recognized in accordance with the data of the road edge object group, by: – Generating data from multiple roadside object groups; – subsequent calculation of an intersection point of a circle that passes through a roadside object group and an axis in the vehicle width direction, for each of the several roadside object groups; and – Using only one roadside object group corresponding to an intersection point located within a range of a predetermined threshold distance from an intersection point that is closest to the vehicle in the vehicle width direction. [13] Road course detection device for vehicles, which includes: – a radar device ( 5 ) for detecting an object based on a reflected wave, which is a reflection of a transmitted wave emitted over a predetermined angular range in a vehicle width direction; and – a means of identification ( 41 , 43 and 45) for detecting a road course in front of a vehicle equipped with the device in accordance with the detection by the radar means ( 5 ) is executed, whereby – the radar device ( 5 ) Object unit data that have at least a distance to an object, with respect to an angle to the object in a vehicle width direction, are acquired based on the reflected wave, – the recognition device ( 41 , 43 and 45 ) shows: – an object recognition device ( 41 and 43 ) to determine an object type as to whether each of the objects is a moving object or a stationary object, in accordance with a relative velocity of an object obtained on the basis of the reflected wave and a velocity of the vehicle; – an effective data extraction tool ( 45) for extracting object unit data that are effective for detecting a road course, wherein the extraction is based on the detection provided by the object detection means ( 41 and 43 ) is executed; – a starting point selection tool ( 45 ) for determining a smallest stationary object as a starting point, wherein the smallest stationary object is located within a range of a predetermined distance from a lateral position of a stationary object that is located closest to the vehicle in the vehicle width direction, wherein the smallest stationary object is likewise located at a position with a shortest direct distance to the vehicle, while the lateral position of the stationary object that is located closest to the vehicle in the vehicle width direction is extracted from the object unit data obtained by the effective data extraction means ( 45) are extracted; and – a roadside object group data creation tool ( 45 ) to create data of a roadside object group by joining and grouping data, starting from the starting point selected by the starting point selection tool ( 45 ) is determined, wherein the data to be grouped have a link requirement of a monotonically increasing distance, while the grouping is performed in directions both clockwise and counterclockwise; and – a roadside detection device ( 41 , 43 and 45 ) for the detection of a road edge in accordance with data from the road edge object group in both directions defined by the road edge object group data generation tool ( 45 ) are formed. [14] Road course recognition device for vehicles according to claim 13, characterized in that the roadside object group data formation means ( 45) Data of the roadside object group is formed by: – Providing a first connection requirement area and a second connection requirement area, which is contained within the first connection requirement area and is smaller than the first connection requirement area, for the roadside object group data creation tool; – Connecting a stationary object to the starting point as a base point defined by the starting point selection means ( 45 ) is determined, whereby the stationary object is included in both the first and the second connection requirement area; – Connecting the connected stationary object as a nearest base point to another stationary object that is contained in both the first and second connection requirement areas; and – Grouping stationary objects by repeating the connection in this way. [15] Road course detection device for vehicles according to claim 13 or 14, characterized by the fact that – the roadside object group data creation tool ( 45 ) forms data from several roadside object groups; and – the roadside detection device ( 41 , 43 and 45 ) detects the road edge by calculating an intersection point of a circle passing through a road edge object group and an axis in the vehicle width direction for each of the multiple road edge object groups, and by using only one road edge object group corresponding to an intersection point located within a range of a predetermined threshold away from an intersection point that is closest to the vehicle in the vehicle width direction. [16] Computer-readable storage medium that stores a program for operating a computer system as the recognition means of the road course recognition device for vehicles according to claim 13 or 15. [17] Road course detection method for vehicles, for detecting a road course around a vehicle applying the method on the basis of a reflected wave which is a reflection of a transmitted wave which is emitted over a predetermined angular range in a vehicle width direction, the method comprising the following steps: – Acquisition of object unit data that have at least a distance to an object, with respect to an angle of the object in a vehicle width direction based on the reflected wave; – Extracting object unit data from a vehicle immediately ahead ( 181 ) and a second vehicle ahead ( 182) regarding the vehicle applying the procedure ( 180 ) among moving objects, and extracting object unit data from reflectors arranged along a road, among stationary objects, while determining the object type as to whether each of the objects is a moving object or a stationary object, in accordance with a relative velocity of an object obtained on the basis of the reflected wave, and a velocity of the vehicle; – Calculating the radius of a circle by approximating the circle with three points, i.e., the vehicle applying the procedure ( 180 ), the vehicle immediately in front ( 181 ) and the second vehicle in front ( 182 ), using object unit data from the three vehicles; and – Recognizing a road course in accordance with the radius of the circle and a line of reflectors. [18] Road course detection device for vehicles, which includes: – a radar device ( 5 ) for detecting an object based on a reflected wave, which is a reflection of a transmitted wave emitted over a predetermined angular range in a vehicle width direction; and – a means of identification ( 108 and 117 ) for detecting a road course in front of a vehicle equipped with the device, based on the detection by the radar device ( 5 ) is executed, whereby – the radar device ( 5 ) Object unit data that have at least a distance to an object, with respect to an angle to the object in a vehicle width direction, are acquired based on the reflected wave, – the recognition device ( 108and 117 ) shows: – an object recognition device ( 108 ) to determine an object type as to whether each of the objects is a moving object or a stationary object, in accordance with a relative velocity of the object obtained on the basis of the reflected wave and a velocity of the vehicle; – a means ( 108 ) for extracting a preceding vehicle that is designed to retrieve object unit data from an immediately preceding vehicle ( 181 ) and a second vehicle ahead ( 182 ) regarding a vehicle equipped with the device ( 180 ) among the moving objects in accordance with the object type determination made by the object recognition device ( 108 ) is executed, to extract; – a reflector extraction agent ( 108) for extracting object unit data from reflectors arranged along a road, among the stationary objects in accordance with an object type determination by the object recognition means ( 8 ) is executed; – an approximate radius calculation tool ( 117 ) for calculating the radius of a circle by approximating the circle with three points, i.e., the vehicle carrying the device ( 180 ), the vehicle immediately in front ( 181 ) and the second vehicle in front ( 182 ), using object unit data from the three vehicles; and – a road course detection device ( 117 ) for recognizing a road course in accordance with the radius of the circle defined by the approximation radius calculation tool ( 117 ) is calculated, and a line of reflectors that is measured by the reflector extraction agent ( 108) are extracted. [19] A road course detection method for vehicles, for detecting a road course around a vehicle applying the method on the basis of a reflected wave which is a reflection of a transmitted wave which is emitted over a predetermined angular range in a vehicle width direction, the method comprising the following steps: – Executing a data acquisition process to acquire object unit data that have at least a distance to an object with respect to an angle of the object in a vehicle width direction based on the reflected wave; – Executing an extraction process to extract the object unit data of reflectors arranged along a road, among stationary objects in accordance with an object type determination, while the object type determination is carried out to determine whether each of the objects is a moving object or a stationary object, in accordance with a relative velocity of an object obtained on the basis of the reflected wave, and a velocity of the vehicle; – Executing a recognition process to detect a road course using a road course estimation curve radius, while the road course estimation curve radius is calculated by approximating a circle passing through a line of reflectors based on object unit data of the reflectors extracted in the extraction process; – repeated execution of a sequence of the capture process, the extraction process, and the recognition process in a predetermined cycle; – Executing a data addition process to add the object unit data of reflectors extracted in the extraction process of the previous cycle to the object unit data of reflectors extracted in the extraction process of the current cycle, wherein the object unit data of reflectors extracted in the extraction process of the previous cycle correspond to those that are arranged within a predetermined area in a radial direction, with respect to the road course estimation curve radius calculated in the detection process of the previous cycle, wherein the data addition process is executed in the extraction process; and – Executing a detection process to detect a road course using a road course estimation curve radius, while the road course estimation curve radius is calculated in accordance with object unit data from reflectors obtained in the data addition process. [20] Road course detection device for vehicles, which includes: – a radar device ( 5 ) for detecting an object based on a reflected wave, which is a reflection of a transmitted wave emitted over a predetermined angular range in a vehicle width direction; and – a means of identification ( 108 and 117 ) to detect a road course in front of the vehicle equipped with the device, based on the detection by the radar device ( 5 ) is executed, whereby – the radar device ( 5) Object unit data that have at least a distance to an object, with respect to an angle to the object in a vehicle width direction, are acquired based on the reflected wave, – the recognition device ( 108 and 117 ) shows: – an object recognition device ( 108 ) to determine an object type as to whether each of the objects is a moving object or a stationary object, in accordance with a relative velocity of the object obtained on the basis of the reflected wave and a velocity of the vehicle; – an extraction agent ( 108 ) to extract object unit data from reflectors arranged along a road among the stationary objects in accordance with the object type determination; and – a road course detection device ( 117) for detecting a road course using a road course estimation curve radius, while the road course estimation curve radius is calculated by drawing a circle passing through a line of reflectors, based on the object unit data of reflectors provided by the extraction agent ( 108 ) are extracted, approximated, – a consequence of processes of the radar medium ( 4 ) and the means of recognition ( 108 and 117 ) is designed to be executed repeatedly in a predetermined cycle, – the extraction agent ( 108 ) a data addition tool ( 108) exhibits, to add the object unit data of reflectors extracted in the extraction process of the previous cycle to the object unit data of reflectors extracted in the extraction process of the current cycle, wherein the object unit data of reflectors extracted in the extraction process of the previous cycle correspond to those arranged under object unit data of reflectors extracted in the previous cycle within a predetermined area in a radial direction with respect to the road course estimation curve radius calculated in the previous cycle, and – the road course recognition device ( 117 ) detects a road course using a road course estimation curve radius, while the road course estimation curve radius is calculated in accordance with the object unit data of reflectors provided by the data addition tool (108 ) will be received.