JP5747787B2 - Lane recognition device - Google Patents

Description

  The present invention relates to a lane recognition device that recognizes a lane on a road.

  When performing driving support such as lane keeping control (lane keeping) or lane departure warning, it is important to recognize a lane (a pair of left and right white lines). In the lane recognition image processing apparatus described in Patent Literature 1, an image is captured by an imaging unit, the captured image is stored, a lane marking search range is set for the stored image, and a lane marking candidate is determined from the search range. The points are extracted, the coordinates of the lane marking candidate points are converted to the aberration corrected coordinates, the lane marking mathematical model formula is derived by approximating the converted set of lane marking candidate points with a mathematical model formula, and the lane marking mathematical model formula is derived. Outputs the lane recognition result obtained from the model formula.

JP 2006-99541 A

  In the lane recognition as described above, a certain range is given to the search range, and the lane is recognized from the search range with a sufficient margin. This is because the road width varies depending on the travel location, so that the lane can be reliably recognized by covering the minimum width to the maximum width. When the lane is recognized in such a search range having a margin, the lane may be erroneously recognized depending on the traveling scene. For example, when a white line is drawn on the wall on the roadside and the white line on the wall side of the lane is faint and difficult to recognize, there is a reflective tape on the top of the pole that exists in the median strip, etc. The white line on the pole side may be faint and difficult to recognize. If such a white line drawn on the wall or reflective tape exists in parallel to the lane, and it is within the search range, even if it is different from the actual white line position of the running lane, There is a possibility of misrecognizing it as a white line. When the lane is erroneously recognized in this way, it cannot be determined whether or not the lane is erroneously recognized while a white line drawn on the wall, a reflective tape, or the like continues to exist in parallel with the lane. When one of the white lines is erroneously recognized as described above, the lane width is calculated wider than the actual one, and the center position of the lane is also calculated as a position on the erroneous recognition side accordingly. As a result, for example, when the misrecognized lane information is used for driving support such as lane keeping control or lane departure warning, the accuracy of driving support decreases.

  Therefore, an object of the present invention is to provide a lane recognition device that prevents erroneous detection of lane recognition.

  A lane recognition device according to the present invention is a lane recognition device that recognizes a lane on a road, recognizes a lane in which the host vehicle is traveling, and acquires lane information, and the host vehicle is traveling. Compare the lane information recognized by the lane recognition means with the lane recognition information obtained by the lane recognition means and the lane recognition information obtained by the lane recognition means. Determining means for determining whether or not.

  In this lane recognition device, the lane recognition means recognizes the lane in which the host vehicle is traveling, and acquires lane information. A lane consists of a pair of left and right lane markings (white line, yellow line, etc.). The lane information includes, for example, the lane width and curvature (curve radius). This lane information includes information used by the driving support device, for example, the target lateral position in the case of lane keeping control, and the left and right lane marking positions in the case of lane departure warning. . In the lane recognition device, the structure information acquisition means acquires the structure information of the lane in which the host vehicle is traveling. The lane structure information is lane-specific information, and includes, for example, the lane width and curvature (curve radius). In the lane recognition device, the determination means compares the recognized lane information with the lane structure information (true value as lane information), and determines whether the recognized lane information is incorrect (lane recognition means). Whether or not the recognition by is a misrecognition). In this way, the lane recognition device can determine whether or not the lane recognition is misrecognized with high accuracy by comparing the recognized lane information with the true lane structure information. Can be prevented. As a result, when various types of driving assistance are performed using lane information, it is possible to prevent driving assistance using misrecognized lane information.

  In the lane recognition device according to the present invention, the preceding vehicle or / and the following vehicle with respect to the own vehicle is recognized, the other vehicle recognition means for acquiring the lateral position of the preceding vehicle or / and the following vehicle with respect to the own vehicle, and the lane recognition by the determination means. If it is determined that the recognition by the means is a misrecognition, the lane on either side of the lane line on the right side and the lane line on the left side of the lane based on the lateral position of the preceding vehicle and / or the following vehicle recognized by the other vehicle recognition unit It is good also as a structure provided with the correction means which discriminate | determines whether the line was recognized incorrectly and corrects the information of the lane recognized by the lane recognition means according to this discrimination | determination result.

  In this lane recognition device, the other vehicle recognition means recognizes the preceding vehicle and the following vehicle of the own vehicle, and acquires the lateral position of the preceding vehicle and the following vehicle with respect to the own vehicle. The vehicle normally travels near the center of the lane (at least in the lane) by a driver's operation or vehicle control by driving assistance. Therefore, in the lane recognition device, when the determination unit determines that the recognition error is incorrect, the correction unit determines whether the preceding vehicle or the following vehicle recognized by the correction unit is based on the lateral position (at least one lateral position or both lateral positions). Then, it is determined which side of the pair of left and right lane lines of the lane is erroneously recognized, and the recognized lane information is corrected based on the determination result. Thus, in the lane recognition device, by using the relative lateral positional relationship between the preceding vehicle and the succeeding vehicle and the host vehicle, it is possible to determine the lane marking on the side that has been erroneously recognized with high accuracy and the lane information. Can be corrected with high accuracy. When various types of driving assistance are performed using the corrected lane information, high-precision driving assistance can be performed, and the reliability of the vehicle user is improved.

  In the lane recognition device according to the present invention, when the lane position acquisition means for acquiring the position of the lane in which the host vehicle is traveling and the determination means determines that the recognition by the lane recognition means is erroneous recognition, the lane position acquisition means acquires If the left lane is the leftmost lane, it is determined that the left lane is misrecognized. If the lane obtained by the lane position acquisition means is the right lane, the right lane It is good also as a structure provided with the correction means which discriminate | determines that it recognized and corrects the information of the lane recognized by the lane recognition means according to this discrimination | determination result.

  In this lane recognition device, the lane position acquisition means acquires the position of the lane in which the host vehicle is traveling. Here, when there are a plurality of lanes on the road, position information is acquired as to whether the lane in which the host vehicle is traveling is the right lane, the left lane, or any other lane. A lane is misrecognized when there is something similar to a white line on the roadside outside the lane, and the vehicle misrecognizes the left lane or the right lane while driving Probability is high. Therefore, in the lane recognition device, if the determination means determines that the vehicle is misrecognized, the correction means erroneously recognizes the left lane line of the lane when the position of the lane in which the host vehicle is traveling is the leftmost lane. When the position of the lane in which the host vehicle is traveling is the rightmost lane, it is determined that the right lane line is erroneously recognized, and the recognized lane information is corrected according to the determination result. Thus, in the lane recognition device, by using the position of the lane in which the host vehicle is traveling, the lane marking on the side that is erroneously recognized can be determined with high accuracy and the lane information can be corrected with high accuracy. Can do.

  According to the present invention, by comparing the recognized lane information with the true lane structure information, it is possible to determine with high accuracy whether or not the lane recognition is erroneous recognition, and prevent erroneous recognition of the lane.

It is a lineblock diagram of the lane maintenance control device concerning a 1st embodiment. It is a flowchart which shows the flow of a process in the lane keeping control apparatus which concerns on 1st Embodiment. It is a block diagram of the lane keeping control apparatus which concerns on 2nd Embodiment. It is a flowchart which shows the flow of a process in the lane keeping control apparatus which concerns on 2nd Embodiment. It is a block diagram of the lane keeping control apparatus which concerns on 3rd Embodiment. It is a flowchart which shows the flow of a process in the lane keeping control apparatus which concerns on 3rd Embodiment.

  Embodiments of a lane recognition device according to the present invention will be described below with reference to the drawings. In addition, the same code | symbol is attached | subjected about the element which is the same or it corresponds in each figure, and the overlapping description is abbreviate | omitted.

  In the present embodiment, the present invention is applied to a lane recognition function incorporated in a lane keeping control device (lane keeping device) mounted on a vehicle. The lane keeping control device according to the present embodiment recognizes a lane (a pair of left and right white lines) from an image captured by a camera, and performs steering control (electric power) based on a control target (deviation amount between the host vehicle and the lane center). The target steering control amount is output to the steering device). In particular, the lane keeping control device according to the present embodiment determines whether or not the lane recognition by the image is misrecognition, and if it is determined as misrecognition, determines whether the white line on the left or right side is misrecognized. Then, the control target is corrected according to the determination result. In the present embodiment, there are three forms in which the determination method on the misrecognition side and the control target correction method are different. The first embodiment is a form using the lateral position information of the preceding vehicle. Is a form using the lateral position information of the preceding vehicle and the following vehicle, and the third embodiment is a form using the position information of the host vehicle and the position of the lane based on the number of lanes.

  A lane keeping control device 1 according to the first embodiment will be described with reference to FIG. FIG. 1 is a configuration diagram of a lane keeping control device according to the first embodiment.

  The lane keeping control device 1 compares the lane width of the lane recognized from the image of the in-vehicle camera and the lane width of the lane during travel of the host vehicle (lane structure information) acquired from the map information, and the lane recognition by the image is incorrect. It is determined whether or not it is recognized. When the lane recognition by the image is misrecognition, the lane keeping control device 1 determines the white line on the erroneously recognized side using the lateral position of the preceding vehicle recognized from the image of the camera that images the front of the host vehicle, The control target (target lateral position) for maintaining the lane is corrected according to the determination result. Since the vehicle normally travels in the vicinity of the center of the lane by the driver's operation or vehicle control by driving assistance, the misrecognized side can be identified and corrected by using the lateral position of the preceding vehicle or the like. . Note that white line recognition and preceding vehicle recognition are performed using a camera that captures the front of the host vehicle, but white line recognition and subsequent vehicle recognition are performed using a camera that captures the rear of the host vehicle, and discrimination is performed using the lateral position of the subsequent vehicle. And may make corrections.

  The lane keeping control device 1 includes a GPS [Global Positioning System] receiver 10, a front camera 11, a map database 20, a runway information output means 30, a white line recognition means 31, a vehicle recognition means 32, a white line misrecognition detection means 33, a control target. Correction means 34 and vehicle control means 35 are provided. Each means 30, 31, 32, 33, 34, 35 is configured by executing an application program for each means by a computer such as a microcomputer. Moreover, the map database 20 and each means 30, 31, 32, 33, 34, 35 are incorporated in plural or one ECU [Electronic Control Unit].

  In the first embodiment, the front camera 11 and the white line recognizing means 31 correspond to the lane recognizing means described in the claims, and the front camera 11 and the vehicle recognizing means 32 are described in the claims. Corresponding to vehicle recognition means, the GPS receiver 10, the map database 20 and the runway information output means 30 correspond to the structure information acquisition means described in the claims, and the white line misrecognition detection means 33 described in the claims. The control target correction means 34 corresponds to the correction means described in the claims.

  The GPS receiver 10 includes a GPS antenna, a processing device, and the like. The GPS receiver 10 receives a GPS signal from each GPS satellite with a GPS antenna. In the GPS receiver 10, each GPS signal is demodulated by the processing device, and the current position (latitude, longitude) of the host vehicle is calculated based on the demodulated GPS information. Then, the GPS receiver 10 outputs information on the current position of the host vehicle to the road information output means 30. The GPS receiver 10 may be a low-cost general-purpose device, and the current position detection accuracy may be low (for example, about several tens of meters).

  The front camera 11 is attached in front of the vehicle. The front camera 11 captures an area including a road ahead of the vehicle, and acquires the captured image (either a color image or a monochrome image). The front camera 11 outputs information of the captured image to the white line recognition unit 31 and the vehicle recognition unit 32. The front camera 11 has a wide imaging range in the left-right direction, and can sufficiently capture a pair of left and right white lines in a running lane. Note that the front camera 11 may be a camera shared with other systems. The front camera may be a stereo camera composed of a plurality of cameras.

  The map database 20 is a database that stores map information. Map information includes road shape information, road structure information, and the like. The road structure information includes road width, lane width, number of lanes, and curvature (curve radius) of the road (lane). The structure information of the road (lane) is information unique to the road (lane) and is the true value of the information of the road (lane).

  The runway information output means 30 acquires map information around the current position from the map database 20 using information on the current position of the host vehicle input from the GPS receiver 10. Then, the lane information output means 30 acquires information on the lane width Wm of the road on which the host vehicle is traveling from the map information around the current position, and outputs the information on the lane width Wm to the white line misrecognition detection means 33.

  The white line recognition means 31 extracts white line candidate points from the image ahead of the host vehicle input from the front camera 11 by edge processing. From the candidate points, the shape of the white line on the left side and the shape of the white line on the right side are estimated for the host vehicle. At this time, in order to reduce the processing load, a range for extracting candidate points is set, and processing is performed only within the range. The white line recognizing means 31 calculates the distance from the lateral center position of the host vehicle to the left white line and the distance to the right white line, and calculates the distance to the left white line and the distance to the right white line. The lane width Ws, the line passing through the center of the pair of left and right white lines (lane center), and the deviation amount (target lateral position) between the center of the lane and the center of the host vehicle are calculated. Further, if necessary, the white line recognition means 31 calculates a curve radius at the center of the lane and calculates a road curvature from the curve radius. The white line recognizing means 31 outputs the recognized paired white line structure information (lane width Ws, lane center, curvature) and control target (target lateral position) to the white line misrecognition detecting means 33 and the control target correcting means 34. To do. Note that processing for recognizing a white line may be processing other than the edge processing described above.

  The vehicle recognition means 32 recognizes a preceding vehicle traveling ahead of the host vehicle from the image ahead of the host vehicle input from the front camera 11 by pattern matching using a vehicle template. When the preceding vehicle can be recognized, the vehicle recognition means 32 calculates a relative distance Lc from the own vehicle to the preceding vehicle and a relative lateral position Xc of the preceding vehicle center with respect to the own vehicle center. Then, the vehicle recognition unit 32 outputs information related to the preceding vehicle (presence / absence of the preceding vehicle, relative distance Lc, relative lateral position Xc when there is a preceding vehicle, etc.) to the control target correcting unit 34. In addition, as a process which recognizes a vehicle, processes other than said pattern matching process may be sufficient.

  The white line misrecognition detection means 33 subtracts the lane width Wm from the lane width Ws using the lane width Wm of the map information input from the runway information output means 30 and the recognized lane width Ws input from the white line recognition means 31. It is determined whether or not the absolute value of the subtraction value is greater than the threshold value α. The lane width Wm is a value unique to the lane and is a true value. On the other hand, the lane width Ws is a value recognized from the image, and it is not known whether it is a correct value. Therefore, if the lane width Ws is different from the true lane width Wm by an error range or more, the pair of left and right white lines constituting the lane is erroneously recognized, and the lane recognition based on the image can be determined to be erroneous recognition. . The threshold value α is set in consideration of the lane width Wm, the lane error range, and the like. When the absolute value of the subtraction value is larger than the threshold value α, the white line misrecognition detection unit 33 determines that the recognition performed by the white line recognition unit 31 is misrecognition. On the other hand, when the absolute value of the subtraction value is equal to or less than the threshold value α, the white line misrecognition detection unit 33 determines that the recognition performed by the white line recognition unit 31 is normal recognition. In the case of normal recognition, the target lateral position calculated by the white line recognition means 31 is used as it is as a control target.

  In the control target correcting means 34, when the white line misrecognition detecting means 33 determines that the recognition is erroneous, the vehicle recognizing means 32 determines whether or not the preceding vehicle information can be acquired. When the vehicle recognition unit 32 cannot acquire the information on the preceding vehicle (for example, when there is no preceding vehicle, the preceding vehicle exists, but the relative distance Lc and the relative lateral position Xc with the preceding vehicle cannot be acquired normally. ), The control target correction means 34 interrupts the lane keeping control because the target lateral position based on the lane misrecognized using the preceding vehicle information cannot be corrected. On the other hand, when the information on the preceding vehicle can be acquired by the vehicle recognition means 32, the control target correction means 34 determines whether or not the relative distance Lc from the host vehicle to the preceding vehicle is longer than the threshold value β. The threshold value β is a threshold value for determining that the preceding vehicle is not separated as far as the vehicle to be followed reliably. The threshold value β may be variable according to the vehicle speed of the host vehicle.

  When the relative distance Lc to the preceding vehicle is equal to or less than the threshold value β, the preceding vehicle can be trusted (the vehicle may travel following the preceding vehicle), so the control target correcting means 34 sets the relative lateral position Xc of the preceding vehicle as the target. Set as horizontal position. On the other hand, when the relative distance Lc to the preceding vehicle is longer than the threshold value β, the preceding vehicle is not so reliable that the control target correcting means 34 recognizes the relative lateral position Xc of the preceding vehicle by the white line recognition means 31. Determine whether the vehicle is on the right or left side of the lane center. When the relative lateral position Xc of the preceding vehicle is determined to be the right side with respect to the center of the lane, the white line on the left side of the pair of left and right white lines recognized by the white line recognition unit 31 is erroneously recognized, and the white line recognition unit 31 accordingly Since it can be determined that the recognized lane center is on the left side, the control target correcting means 34 uses the white line on the right side recognized by the white line recognizing means 31 as a reference and Wm / 2 (the true lane width of the true lane width). The position on the left side of 1/2) is set as the target lateral position. On the other hand, when the relative lateral position Xc of the preceding vehicle is determined to be the left side with respect to the center of the lane, the white line on the right side of the pair of left and right white lines recognized by the white line recognition unit 31 is erroneously recognized, and the white line recognition unit accordingly Since it can be determined that the center of the lane recognized at 31 is on the right side, the control target correction means 34 uses the white line on the left recognized by the white line recognition means 31 as a reference, and sets the position on the right side of the left white line by Wm / 2. Set as the target lateral position. Then, the control target correcting means 34 outputs the set target lateral position to the vehicle control means 35 as a control target. If the white line misrecognition detection means 33 determines that the recognition performed by the white line recognition means 31 is normal recognition, the target lateral position calculated by the white line recognition means 31 is output as it is as a control target.

  The vehicle control unit 35 uses the target lateral position (control target) input from the control target correction unit 34 to calculate a target steering control amount that makes the target lateral position zero. Then, the vehicle control means 35 outputs the target steering control amount to the electric power steering device.

  With reference to FIG. 1, the operation of the lane keeping control device 1 will be described along the flowchart of FIG. 2. FIG. 2 is a flowchart showing a flow of processing in the lane keeping control device according to the first embodiment. In the lane keeping control device 1, the operation described below is repeated at regular intervals.

  The GPS receiver 10 receives a receivable GPS signal, calculates the current position of the host vehicle using the received GPS signal, and outputs the current position information to the runway information output means 30. Further, the front camera 11 images the front of the host vehicle and outputs the image information to the white line recognition unit 31 and the vehicle recognition unit 32.

  The white line recognition means 31 recognizes a pair of left and right white lines (lanes) from the image of the front camera 11, and calculates the lane width Ws, the target lateral position, etc. (S10). The runway information output means 30 acquires map information around the current position from the map database 20 based on the current position by the GPS receiver 10, and acquires the lane width Wm from the map information (S11). Then, the white line misrecognition detecting means 33 subtracts the lane width Wm from the lane width Ws, and determines whether or not the absolute value of the subtraction value is larger than the threshold value α (S12). When it is determined in S12 that the absolute value of the subtraction value is equal to or less than the threshold value α, the target lateral position calculated by the white line recognition unit 31 is set as the control target.

  When it is determined in S12 that the absolute value of the subtraction value is larger than the threshold value α, the vehicle recognition means 32 recognizes the preceding vehicle from the image of the front camera 11, and if the preceding vehicle can be recognized, information on the preceding vehicle ( The relative distance Lc and the relative lateral position Xc) are calculated (S13). Then, the control target correcting means 34 determines whether or not the preceding vehicle information has been acquired (S14). If it is determined in S14 that the preceding vehicle information has not been acquired, the lane keeping control is interrupted.

  When it is determined in S14 that the preceding vehicle information has been acquired, the control target correcting unit 34 determines whether or not the relative distance Lc to the preceding vehicle is longer than the threshold value β (S15). When it is determined in S15 that the relative distance Lc is longer than the threshold value β, the control target correcting unit 34 determines that the left white line is erroneously recognized when the relative lateral position Xc of the preceding vehicle is on the right side of the lane center. If the right lateral line of Wm / 2 is set as the target lateral position with respect to the right recognition right line as a reference, and the relative lateral position Xc of the preceding vehicle is on the left side with respect to the center of the lane, the right white line is determined to be misrecognized. The position on the right side of Wm / 2 with respect to the correctly recognized left white line is set as the target lateral position (S16). On the other hand, when it is determined in S15 that the relative distance Lc is equal to or less than the threshold value β, the control target correcting means 34 sets the relative lateral position Xc of the preceding vehicle as the target lateral position (S17).

  Then, the vehicle control means 35 calculates a target steering control amount so that the set target lateral position becomes zero, and outputs the target steering control amount to the electric power steering device (S18). In the electric power steering device, a steering torque corresponding to the target steering control amount is applied to the steering mechanism. As a result, the host vehicle travels near the center of the lane and maintains the lane.

  According to this lane keeping control device 1, by comparing the lane width of the lane recognized from the image with the true lane width acquired from the map information, it is possible to determine whether or not the lane recognition by the image is erroneously recognized. It is possible to determine and prevent erroneous recognition of the lane. As a result, when the lane keeping control is performed using the lane information, the lane keeping control can be prevented from being performed using the erroneously recognized lane information.

  Furthermore, according to the lane keeping control device 1, by using the relative lateral position of the preceding vehicle, the white line on the erroneously recognized side can be determined with high accuracy and the target lateral position can be corrected with high accuracy. . Since the lane keeping control is performed using the corrected target lateral position, the highly accurate lane keeping control can be performed, and the reliability of the vehicle user is improved.

  Moreover, according to the lane keeping control apparatus 1, since it can comprise using the low precision GPS receiver 10, cost can be reduced. The accuracy of the current position by the GPS receiver 10 is low. For example, even if information on the lane width at a position several tens of meters away from the actual current position is acquired from the map information, the acquired lane width is the actual current position. Since it is not different from the lane width at the position, it is possible to determine misrecognition with high accuracy even using the lane width at the shifted position.

  Next, the lane keeping control device 2 according to the second embodiment will be described with reference to FIG. FIG. 3 is a configuration diagram of a lane keeping control device according to the second embodiment.

  The lane keeping control device 2 performs the same judgment as the lane keeping control device 1 according to the first embodiment as to whether or not the lane recognition by the image is erroneous recognition. In the first embodiment, the white line on the erroneous recognition side is discriminated and corrected using the lateral position information of one vehicle of the preceding vehicle (or the following vehicle). You may not be traveling. Therefore, when the lane recognition based on the image is erroneous recognition, the lane keeping control device 2 recognizes the preceding vehicle recognized from the image of the camera that images the front of the host vehicle in order to improve the accuracy of discrimination and correction of the white line on the erroneous recognition side. And the lateral position of the following vehicle recognized from the image of the camera that captures the rear side of the host vehicle, the white line on the erroneously recognized side is determined, and a control target for maintaining the lane according to the determination result ( Correct the target lateral position.

  The lane keeping control device 2 includes a GPS receiver 10, a front camera 11, a rear camera 12, a map database 20, a track information output means 30, white line recognition means 31, 36, vehicle recognition means 32, 37, and a white line erroneous recognition detection means 38. , Control target correction means 39 and vehicle control means 35 are provided. Each means 30, 31, 32, 35, 36, 37, 38, 39 is configured by executing an application program for each means by a computer. Further, the map database 20 and each means 30, 31, 32, 35, 36, 37, 38, 39 are incorporated into a plurality or one ECU.

  In the second embodiment, the front camera 11 and the white line recognizing means 31, the rear camera 12 and the white line recognizing means 36 correspond to the lane recognizing means described in the claims, and the front camera 11 and the vehicle recognizing means 32. The rear camera 12 and the vehicle recognizing means 37 correspond to other vehicle recognizing means described in the claims, and the GPS receiver 10, the map database 20 and the runway information output means 30 acquire the structure information described in the claims. The white line misrecognition detection means 38 corresponds to the determination means described in the claims, and the control target correction means 39 corresponds to the correction means described in the claims.

  The rear camera 12 is attached to the rear of the vehicle. The rear camera 12 captures an area including a road behind the vehicle and acquires the captured image. The rear camera 12 outputs information of the captured image to the white line recognition unit 36 and the vehicle recognition unit 37. The rear camera 12 has a wide imaging range in the left-right direction, and can sufficiently capture a pair of left and right white lines in a running lane. Note that the rear camera 12 may be a camera shared with other systems.

  The white line recognizing means 36 performs the same processing as the white line recognizing means 31 described in the first embodiment using the image behind the host vehicle input from the rear camera 12, and recognizes the structure information ( The lane width Ws, lane center, curvature) and control target (target lateral position) are output to the white line misrecognition detection means 38 and the control target correction means 39. Similarly, the white line recognizing means 31 using the image of the front camera 11 and the white line recognizing means 36 using the image of the rear camera 12 are similarly structured information (lane width Ws, lane center, curvature) of a pair of left and right white lines and a control target (target). (Horizontal position) is calculated. Information of either one of the white line recognition unit 31 and the white line recognition unit 36 may be used, or an average value of each information of the white line recognition unit 31 and the white line recognition unit 36 is used. May be. Alternatively, the recognition process may be performed by only one of the white line recognition unit 31 and the white line recognition unit 36.

  The vehicle recognizing unit 37 recognizes a succeeding vehicle traveling behind the host vehicle from the image behind the host vehicle input from the rear camera 12 by pattern matching using a vehicle template. When the succeeding vehicle can be recognized, the vehicle recognizing unit 37 calculates a distance Lcr from the own vehicle to the succeeding vehicle and a relative lateral position Xcr of the succeeding vehicle center with respect to the own vehicle center. Then, the vehicle recognition unit 37 outputs information related to the following vehicle (presence / absence of the following vehicle, distance Lcr, relative lateral position Xcr when there is a following vehicle) to the control target correction unit 39.

  The white line misrecognition detection means 38 basically performs the same processing as the white line misrecognition detection means 33 described in the first embodiment. However, as for the lane width Ws by image recognition used for the determination, one of the white line recognition unit 31 and the white line recognition unit 36 may be used, or each of the white line recognition unit 31 and the white line recognition unit 36 may be used. An average value of the vehicle width or the like may be used.

  In the control target correcting means 39, when the white line misrecognition detecting means 38 determines that it is erroneously recognized, whether or not the vehicle information can be acquired by both the vehicle recognizing means 32 and the vehicle recognizing means 37 (the number of recognized vehicles). Whether there are two or not. When vehicle information cannot be acquired by both the vehicle recognition means 32 and the vehicle recognition means 37 (when the number of recognized vehicles is one or less), the control target correction means 39 uses the first embodiment. The same processing performed by the control target correcting means 34 is performed.

  On the other hand, when the vehicle information can be acquired by both the vehicle recognition unit 32 and the vehicle recognition unit 37 (when the number of recognized vehicles is two), the control target correction unit 39 acquires the vehicle information. It is determined whether the relative distance Lcf to the preceding vehicle is equal to or less than the threshold value β and the relative distance Lcr to the subsequent vehicle acquired by the vehicle recognition unit 37 is equal to or less than the threshold value β. When the relative distance Lcf to the preceding vehicle is equal to or less than the threshold value β and the relative distance Lcr to the following vehicle is equal to or less than the threshold value β, the preceding vehicle and the following vehicle can be trusted. And a line segment connecting the following vehicle Xcr is set as a target lateral position of the host vehicle. For example, the center position in the horizontal direction in the line segment is set as the target horizontal position. Here, since the horizontal position information of the two vehicles is used, even if the horizontal positions of the two vehicles vary with respect to the center of the lane, they are averaged. .

  On the other hand, when the distance Lcf to the preceding vehicle is longer than the threshold value β and / or the distance Lcr to the following vehicle is longer than the threshold value β, the preceding vehicle and the following vehicle are not reliable. When the relative lateral position Xcf of the vehicle and the relative lateral position Xcr of the following vehicle are both on the right side with respect to the center of the lane, the white line on the left side of the pair of left and right white lines recognized by the white line recognition unit 31 (or the white line recognition unit 36) is mistaken. Since it can be determined that the lane center recognized by the white line recognition means 31 is shifted to the left side accordingly, the right white line recognized by the white line recognition means 31 is used as a reference, and Wm / 2 minutes from the right white line. The left position is set as the target lateral position. Further, in the control target correcting means 39, when the relative lateral position Xcf of the preceding vehicle and the relative lateral position Xcr of the following vehicle are both on the left side with respect to the lane center, the right side of the pair of left and right white lines recognized by the white line recognizing means 31 is used. Since the white line is misrecognized and it can be determined that the center of the lane recognized by the white line recognizing unit 31 is shifted to the right side accordingly, the left white line recognized by the white line recognizing unit 31 is used as a reference, and the Wm from the left white line / Set the position on the left side for 2 minutes as the target lateral position. Further, in the control target correcting means 39, when it is determined that the relative lateral position Xcf of the preceding vehicle and the relative lateral position Xcr of the following vehicle are opposite to the center of the lane, the relative distance Lcf to the preceding vehicle and the distance to the following vehicle are determined. The white line on the misrecognition side is determined and corrected based on the relative lateral position Xc of the vehicle having the shorter distance (reliable vehicle) of the relative distance Lcr. This determination and correction method performs the same processing as the processing performed when the relative distance Lc to the preceding vehicle is longer than the threshold value β performed in the control target correcting means 34 according to the first embodiment. The control target correcting means 39 outputs the set target lateral position to the vehicle control means 35 as a control target.

  With reference to FIG. 3, the operation of the lane keeping control device 2 will be described along the flowchart of FIG. 4. FIG. 4 is a flowchart showing a process flow in the lane keeping control apparatus according to the second embodiment. The lane keeping control device 2 repeatedly performs the operations described below at regular intervals.

  The GPS receiver 10 receives a receivable GPS signal, calculates a current position using the received GPS signal, and outputs the current position information to the road information output means 30. Further, the front camera 11 images the front of the host vehicle and outputs the image information to the white line recognition unit 31 and the vehicle recognition unit 32. Further, the rear camera 12 images the rear of the host vehicle and outputs the image information to the white line recognition unit 36 and the vehicle recognition unit 37.

  The white line recognition means 31 recognizes a pair of left and right white lines (lanes) from the image of the front camera 11, and calculates the lane width Ws, the target lateral position, etc. (S20). The white line recognition means 36 recognizes a pair of left and right white lines (lanes) from the image of the rear camera 12, and calculates the lane width Ws, the target lateral position, etc. (S20). Further, the runway information output means 30 acquires map information around the current position from the map database 20 based on the current position by the GPS receiver 10, and acquires the lane width Wm from the map information (S21). Then, the white line misrecognition detection means 38 subtracts the lane width Wm from the lane width Ws, and determines whether or not the absolute value of the subtraction value is larger than the threshold value α (S22). When it is determined in S22 that the absolute value of the subtraction value is equal to or less than the threshold value α, the target lateral position calculated by the white line recognition unit 31 (or the white line recognition unit 36) is set as the control target.

  When it is determined in S22 that the absolute value of the subtraction value is larger than the threshold value α, the vehicle recognition means 32 recognizes the preceding vehicle from the image of the front camera 11, and if the preceding vehicle can be recognized, information on the preceding vehicle ( Relative distance Lcf and relative lateral position Xcf) are calculated (S23). The vehicle recognition unit 37 recognizes the following vehicle from the image of the rear camera 12, and if the following vehicle can be recognized, information (relative distance Lcr, relative lateral position Xcr) of the following vehicle is calculated (S24). Then, the control target correcting means 39 determines whether or not the preceding vehicle information and the following vehicle information can be acquired (S25). When it is determined in S25 that both pieces of vehicle information have not been acquired (when the number of recognized vehicles is one or less), the processing after S14 in the flowchart of FIG. 2 is performed. In the process of S14, it is determined not “whether or not preceding vehicle information can be acquired” but “whether or not preceding vehicle information or subsequent vehicle information can be acquired”. And in the process of S15, S16, S17, it processes using the information of the vehicle which can be acquired.

  When it is determined in S25 that the preceding vehicle information and the following vehicle information can be acquired (when the number of recognized vehicles is two), the control target correcting unit 39 determines that the relative distance Lcf to the preceding vehicle is a threshold value β It is then determined whether or not the relative distance Lcr to the following vehicle is equal to or less than the threshold value β (S26). When it is determined in S26 that the relative distance Lcf is equal to or less than the threshold value β and the relative distance Lcr is equal to or less than the threshold value β, the control target correcting unit 39 connects the relative lateral position Xcf of the preceding vehicle and the relative lateral position Xcr of the subsequent vehicle. The top is set as the target lateral position of the host vehicle (S27). On the other hand, when it is determined in S26 that the relative distance Lcf is longer than the threshold value β or / and the relative distance Lcr is longer than the threshold value β, the control target correcting means 39 determines that the relative lateral positions Xcf and Xcr of the preceding vehicle and the following vehicle are both In the case of the right side with respect to the center of the lane, the left white line is determined to be misrecognized, the position on the left side by Wm / 2 with respect to the right white line of correct recognition is set as the target lateral position, and the relative lateral positions Xcf and Xcr are If both are on the left side of the center of the lane, the right white line is determined to be misrecognized, and the position on the right side of Wm / 2 minutes is set as the target lateral position based on the correctly recognized left white line, and relative to the relative lateral position Xcf. When the lateral position Xcr is opposite to the center of the lane, the white line on the erroneous recognition side is determined based on the relative lateral position Xc of the vehicle having the shorter relative distance Lc, and Wm is determined based on the correctly recognized white line. / Set the position of 2 minutes to the target lateral position (S28).

  Then, the vehicle control means 35 calculates a target steering control amount such that the set target lateral position becomes zero, and outputs the target steering control amount to the electric power steering device (S29). In the electric power steering device, a steering torque corresponding to the target steering control amount is applied to the steering mechanism. As a result, the host vehicle travels near the center of the lane and maintains the lane.

  The lane keeping control device 2 has the same effects as the lane keeping control device 1 according to the first embodiment, and also has the following effects. According to the lane keeping control device 2, the preceding vehicle and the following vehicle before and after the host vehicle are both recognized, and the information on the preceding vehicle and the information on the following vehicle is used to make the white line on the erroneously recognized side higher. The target lateral position can be corrected with higher accuracy while being able to discriminate with accuracy. Further, even when only one of the preceding vehicle and the following vehicle exists, determination and correction similar to those of the lane keeping control device 1 according to the first embodiment are possible.

  Next, a lane keeping control device 3 according to a third embodiment will be described with reference to FIG. FIG. 5 is a configuration diagram of a lane keeping control device according to the third embodiment.

  The lane keeping control device 3 makes the same determination as the lane keeping control device 1 according to the first embodiment as to whether or not the lane recognition by the image is erroneous recognition. In the first embodiment and the second embodiment, when a vehicle does not exist before or after the host vehicle or when a sensor for recognizing the vehicle is not provided, the white line on the erroneously recognized side is identified. The target lateral position cannot be corrected. Therefore, when the lane recognition by the image is misrecognized, the lane keeping control device 3 estimates the lane position where the own vehicle is traveling from the highly accurate position of the own vehicle and the road structure information (number of lanes) of the traveling road, Using the lane position, the misrecognized white line is discriminated, and the control target (target lateral position) for maintaining the lane is corrected according to the discrimination result. In other words, lanes may be misrecognized if they have characteristics similar to white lines on the roadside outside the lane (for example, lines drawn on walls parallel to the road, reflections attached to the median pole) If the vehicle is traveling in the left lane or right lane, there is a high probability of misrecognizing the white line on the left side or the white line on the right side. Can be determined and corrected.

  The lane keeping control device 3 includes a GPS receiver 10, a front camera 11, an inertial sensor 13, a wheel speed sensor 14, a map database 21, a lane information output means 40, a white line recognition means 31, a white line misrecognition detection means 33, a high-precision self A vehicle position estimation unit 41, a host vehicle traveling lane estimation unit 42, a control target correction unit 43, and a vehicle control unit 35 are provided. Each means 40, 31, 33, 41, 42, 43, 35 is configured by executing an application program for each means by a computer. Further, the map database 21 and each means 40, 31, 33, 41, 42, 43, 35 are incorporated into a plurality or one ECU.

  In the third embodiment, the front camera 11 and the white line recognizing means 31 correspond to the lane recognizing means described in the claims, and the GPS receiver 10, the map database 21, and the runway information output means 40 are claimed. The GPS receiver 10, the inertial sensor 13, the wheel speed sensor 14, the map database 21, the road information output means 40, the high-accuracy own vehicle position estimation means 41, and the own vehicle travel lane The estimation means 42 corresponds to the lane position acquisition means described in the claims, the white line misrecognition detection means 33 corresponds to the determination means described in the claims, and the control target correction means 43 corresponds to the claims. It corresponds to the correction means described.

  The inertial sensor 13 is, for example, a gyro sensor. The inertial sensor 13 detects the angular velocity in the turning direction of the host vehicle. Then, the inertial sensor 13 outputs angular velocity information to the high-accuracy own vehicle position estimating means 41.

  The wheel speed sensor 14 is provided on each wheel. The wheel speed sensor 14 detects the rotational speed (wheel speed) of the wheel. The wheel speed sensor 14 outputs the wheel speed information to the high-accuracy own vehicle position estimating means 41.

  The map database 21 is a database that stores map information. The map information includes map information for map matching in addition to the map information stored in the map database 20 according to the first embodiment.

  The runway information output means 40 acquires map information around the current position from the map database 21 using information on the current position of the host vehicle input from the GPS receiver 10. Then, the lane information output means 40 acquires information on the lane width Wm and the number of lanes of the road on which the vehicle is traveling from the map information around the current position, and outputs the information on the lane width Wm to the white line misrecognition detection means 33. Then, the information on the number of lanes is used as the vehicle lane estimation means 42.

  The high-accuracy own vehicle position estimating means 41 estimates the turning angle of the own vehicle by time-integrating the angular velocity in the turning direction of the own vehicle inputted from the inertial sensor 13. The high-accuracy own vehicle position estimating means 41 calculates the vehicle speed of the host vehicle from the wheel speed input from the wheel speed sensor 14 of each wheel, and calculates the travel distance of the host vehicle by integrating the vehicle body speed over time. . Then, the high-accuracy own vehicle position estimating means 41 estimates the relative position of the own vehicle based on the turning angle and the movement distance of the own vehicle. Further, the high-accuracy own vehicle position estimating means 41 estimates the position by hybrid navigation using the relative position of the own vehicle and the absolute position from the GPS receiver 10. Further, the high-accuracy own vehicle position estimating means 41 acquires map information for map matching around the current position from the map database 21 and performs map matching using the estimated position by hybrid navigation and the map information for map matching. Estimate the current position of. The estimated own vehicle position is highly accurate (for example, about 1 m), and has an accuracy sufficient to determine the lane position in which the own vehicle is traveling when a plurality of lanes exist on the travel path. Note that the above-described high-accuracy vehicle position estimation method is merely an example, and other methods may be used. Further, when the host vehicle is equipped with a high-performance navigation device, the navigation device functions as the high-accuracy host vehicle position estimating means 41.

  The own vehicle travel lane estimation means 42 is based on the number of lanes in the travel route of the host vehicle input from the travel path information output means 40 and the current position of the host vehicle with high accuracy input from the high accuracy vehicle position estimation means 41. The position of the lane in which the vehicle is traveling is determined. For example, there is no distinction in the case of 1 lane, right lane or left lane in the case of 2 lanes, right end lane, left end lane or center lane in case of 3 lanes, right end lane or left end in case of 4 lanes Discriminate between lanes and other lanes. Then, the host vehicle travel lane estimating unit 42 outputs the determined lane position to the control target correcting unit 43.

  In the control target correcting means 43, when the white line misrecognition detecting means 33 determines that the recognition is wrong, the position of the lane in which the host vehicle that is input from the own vehicle traveling lane estimating means 42 is traveling is the left lane or the right lane, or Determine if it is a lane other than that. When the left lane is determined (when there is a high possibility that there is something similar to the white line on the left side of the host vehicle), the white line on the left side of the pair of left and right white lines recognized by the white line recognition unit 31 is erroneously recognized. Accordingly, since it can be determined that the center of the lane recognized by the white line recognition means 31 is shifted to the left side, the control target correction means 43 uses the right white line recognized by the white line recognition means 31 as a reference, and The position to the left of Wm / 2 minutes from the white line is set as the target lateral position. When it is determined that the lane is the rightmost lane (when there is a high possibility that there is something similar to the white line on the right side of the vehicle), the white line on the right side of the pair of left and right white lines recognized by the white line recognition means 31 is erroneously recognized. Accordingly, it can be determined that the center of the lane recognized by the white line recognizing means 31 is shifted to the right side. Therefore, the control target correcting means 43 uses the left white line recognized by the white line recognizing means 31 as a reference, and the left The position to the right of Wm / 2 minutes from the white line is set as the target lateral position. If the other lane is determined, the erroneously recognized side cannot be determined, so the control target correcting means 43 does not correct the target lateral position. Then, the control target correcting means 43 outputs the set target lateral position to the vehicle control means 35 as a control target.

  With reference to FIG. 5, the operation of the lane keeping control device 3 will be described along the flowchart of FIG. 6. FIG. 6 is a flowchart showing a flow of processing in the lane keeping control device according to the third embodiment. In the lane keeping control device 3, the operation described below is repeated at regular intervals.

  The GPS receiver 10 receives a receivable GPS signal, calculates the current position using the received GPS signal, and outputs the current position information to the road information output means 40 and the high-accuracy vehicle position estimation means 41. doing. Further, the front camera 11 images the front of the host vehicle and outputs the image information to the white line recognition means 31. The inertial sensor 13 detects the angular velocity in the turning direction of the host vehicle, and outputs the angular velocity to the high-accuracy host vehicle position estimating means 41. Further, the wheel speed sensor 14 of each wheel detects the wheel speed and outputs the wheel speed to the high-accuracy own vehicle position estimating means 41.

  The white line recognition means 31 recognizes a pair of left and right white lines (lanes) from the image of the front camera 11, and calculates the lane width Ws, the target lateral position, etc. (S30). Further, the runway information output means 40 acquires map information around the current position from the map database 21 based on the current position by the GPS receiver 10, and acquires the lane width Wm from the map information (S31). Then, the white line misrecognition detection means 33 subtracts the lane width Wm from the lane width Ws, and determines whether or not the absolute value of the subtraction value is larger than the threshold value α (S32). When it is determined in S32 that the absolute value of the subtraction value is equal to or less than the threshold value α, the target lateral position calculated by the white line recognition unit 31 is set as the control target.

  When it is determined in S32 that the absolute value of the subtraction value is larger than the threshold value α, the high-accuracy own vehicle position estimating means 41 uses the inertial sensor 13 to determine the angular velocity of the vehicle in the turning direction and the wheel speed sensor 14 for each wheel. To estimate the relative position of the vehicle. Then, the high-accuracy own vehicle position estimating means 41 uses the relative position of the own vehicle, the absolute current position by the GPS receiver 10, and the map information for map matching acquired from the map database 21. The current position of the host vehicle is estimated (S33). Further, the runway information output means 40 acquires the number of lanes from the map information around the current position (S34). Then, the host vehicle travel lane estimation means 42 estimates the travel lane of the host vehicle based on the current position of the host vehicle and the number of lanes on the travel path of the host vehicle (S35).

  Then, the control target correcting means 43 determines that the left white line is erroneously recognized when the position of the traveling lane of the host vehicle is the leftmost lane, and the position on the left side by Wm / 2 with respect to the right white line of correct recognition. Is set as the target lateral position, and when the position of the vehicle's lane is the rightmost lane, the right white line is determined to be misrecognized, and the position on the right side of Wm / 2 minutes is determined based on the correctly recognized left white line. The target lateral position is set, and in other cases, the target lateral position is not corrected (S36).

  Then, the vehicle control means 35 calculates a target steering control amount so that the set target lateral position becomes zero, and outputs the target steering control amount to the electric power steering device (S37). In the electric power steering device, a steering torque corresponding to the target steering control amount is applied to the steering mechanism. As a result, the host vehicle travels near the center of the lane and maintains the lane.

  According to the lane keeping control device 3, as in the lane keeping control device 1 according to the first embodiment, the lane recognition by the image is erroneously recognized by comparing the lane width of the image recognition with the lane width of the map information. Whether or not it is recognized can be determined with high accuracy. Furthermore, according to the lane keeping control device 3, by using the traveling lane of the own vehicle estimated from the highly accurate own vehicle position and the number of lanes, the erroneously recognized white line can be discriminated with high accuracy and the target The lateral position can be corrected with high accuracy.

  As mentioned above, although embodiment which concerns on this invention was described, this invention is implemented in various forms, without being limited to the said embodiment.

  For example, although the present embodiment is applied to the lane keeping control device, it may be applied to other devices using lane information (such as a lane departure warning device), or to a lane recognition device that performs only lane recognition. You may apply. In the case of a lane departure warning device, the target lateral position is not corrected, but the position of the white line on the erroneously recognized side is corrected.

  In the present embodiment, a pair of left and right white lines are recognized as lanes. However, in addition to the white lines, other color division lines such as yellow lines drawn on the road surface may be recognized.

  In the present embodiment, the lane is recognized using the camera image. However, the lane may be recognized using other means such as a laser radar. In this embodiment, the preceding vehicle and the following vehicle are recognized using the camera image. However, the preceding vehicle and the like may be recognized using other means such as a laser radar.

  Further, in the present embodiment, it is configured to determine whether or not erroneous recognition is performed using the lane width, but it is determined whether or not erroneous recognition is performed using structure information (for example, curvature, curve radius) of other lanes. May be.

  Further, in the first and second embodiments, the determination is made using the relative distance to the vehicle in order to determine whether the preceding vehicle and the following vehicle are reliable, but the determination may be made using other information. Good. For example, the degree of wobbling in the lateral direction of the vehicle is detected, and the determination is made using the degree of wobbling.

  In the third embodiment, the position of the lane in which the host vehicle is traveling is estimated based on the highly accurate position information of the host vehicle and the number of lanes on the travel path. The position of the lane may be acquired.

  1, 2, 3 ... Lane maintaining control device, 10 ... GPS receiver, 11 ... Front camera, 12 ... Rear camera, 13 ... Inertial sensor, 14 ... Wheel speed sensor, 20, 21 ... Map database, 30, 40 ... Runway Information output means 31, 36 ... White line recognition means, 32, 37 ... Vehicle recognition means, 33, 38 ... White line erroneous recognition detection means, 34, 39, 43 ... Control target correction means, 35 ... Vehicle control means, 41 ... High Accuracy own vehicle position estimating means, 42 ... own vehicle traveling lane estimating means.

Claims (2)

A lane recognition device for recognizing a lane on a road,
Lane recognition means for recognizing the lane in which the host vehicle is traveling and acquiring lane information;
Structure information acquisition means for acquiring structure information of the lane in which the host vehicle is traveling,
A determination unit that compares the lane information recognized by the lane recognition unit with the lane structure information acquired by the structure information acquisition unit, and determines whether the recognition by the lane recognition unit is a recognition error;
Other vehicle recognition means for recognizing a preceding vehicle or / and a following vehicle with respect to the own vehicle and acquiring a lateral position of the preceding vehicle or / and the following vehicle with respect to the own vehicle;
When the determination means determines that the recognition by the lane recognition means is a misrecognition, the right lane line and the left lane line of the lane based on the lateral position of the preceding vehicle or / and the following vehicle recognized by the other vehicle recognition means A correction means for determining which side of the lane marking is erroneously recognized, and correcting information on the lane recognized by the lane recognition means according to the determination result;
A lane recognition device comprising: A lane recognition device for recognizing a lane on a road,
Lane recognition means for recognizing the lane in which the host vehicle is traveling and acquiring lane information;
Structure information acquisition means for acquiring structure information of the lane in which the host vehicle is traveling,
A determination unit that compares the lane information recognized by the lane recognition unit with the lane structure information acquired by the structure information acquisition unit, and determines whether the recognition by the lane recognition unit is a recognition error;
Lane position acquisition means for acquiring the position of the lane in which the host vehicle is traveling;
When the determination means determines that the recognition by the lane recognition means is a misrecognition, it is determined that the left lane line of the lane is erroneously recognized when the position of the lane acquired by the lane position acquisition means is the leftmost lane, When the lane position acquired by the lane position acquisition means is the rightmost lane, it is determined that the right lane marking of the lane is erroneously recognized, and the lane information recognized by the lane recognition means is corrected according to the determination result. Correction means to
A lane recognition device comprising:

Images