JP2018009833A - Vehicle self-localization device and vehicle self-localization method - Google Patents

Abstract

PROBLEM TO BE SOLVED: To provide a vehicle self-localization device and vehicle self-localization method capable of suppressing the deterioration of accuracy concerning self-localization of an own vehicle on a map.SOLUTION: An ECU 20 includes: a movement amount calculation section 23 for tracking change of a relative position with an own vehicle at a feature point as reference and calculating a first movement amount of the own vehicle on the basis of the tracked change of the relative position; a reliability setting section 24 for setting reliability of the first movement amount; a correction value calculation section 25 for calculating a vehicle speed correction value to correct an output of a vehicle speed detection section 35 mounted on the own vehicle on the basis of the first movement amount when reliability is equal to or more than a threshold; a first localization section 26 for self-localizing the own vehicle on a map on the basis of the first movement amount when reliability is equal to or more than the threshold; and a second localization section 27 for calculating a second movement amount with the use of an output of the vehicle speed detection section after correction by the vehicle speed correction value when reliability is less than the threshold, and self-localizing the own vehicle on a map on the basis of the calculated second movement amount.SELECTED DRAWING: Figure 1

Description

  The present invention relates to a vehicle position specifying device and a vehicle position specifying method for specifying a vehicle position on a map.

  2. Description of the Related Art Conventionally, an own vehicle position specifying device that specifies an own vehicle position on a map is known. For example, the own vehicle position specifying device calculates the movement amount of the own vehicle by integrating the output of the vehicle speed sensor included in the own vehicle, and specifies the own vehicle position on the map based on the calculated movement amount.

  There is also known an own vehicle position specifying device for specifying an own vehicle position on a map based on a change in relative position of an object existing around the own vehicle with respect to the own vehicle. In Patent Document 1, a change in the relative position of an object based on the own vehicle is calculated based on an imaging result of a stereo camera included in the own vehicle, and a movement amount of the own vehicle is calculated based on the calculated change in the relative position. calculate. And the own vehicle position on a map is specified using this movement amount.

Japanese Patent Laid-Open No. 11-51650

  The amount of movement calculated based on the output of the vehicle speed sensor causes an error steadily due to wheel slip or the like. On the other hand, by using a stereo camera, it is possible to calculate the amount of movement of the host vehicle based on the change in the three-dimensional information of the object, but the detection of the object according to the brightness of the surroundings and the presence of features for detecting the object The accuracy may change. Therefore, even when the movement amount of the own vehicle is calculated using either the vehicle speed sensor or the stereo camera, the accuracy of the movement amount calculated for specifying the vehicle position may be lowered.

  The present invention has been made in view of the above problems, and an object of the present invention is to provide a vehicle position specifying device and a vehicle position specifying method that can suppress a decrease in the vehicle position accuracy specified on a map. And

  In order to solve the above problem, in the present invention, an object recognition unit that recognizes three-dimensional information of an object based on an imaging result around the vehicle by the first imaging unit and the second imaging unit, and the recognized object A feature point extracting unit that extracts a feature point, a change in relative position of the feature point with respect to the own vehicle is tracked, and a first movement amount of the own vehicle is calculated based on the tracked change in the relative position Mounted on the host vehicle based on the first moving amount when the reliability is equal to or greater than a threshold value, and a reliability setting unit that sets the reliability of the first moving amount. A correction value calculation unit for calculating a vehicle speed correction value for correcting the output of the detected vehicle speed detection unit, and the vehicle position on the map based on the first movement amount when the reliability is equal to or greater than a threshold value A first position specifying unit for specifying the vehicle speed and the vehicle speed compensation when the reliability is less than a threshold value. A second position specifying unit that calculates a second movement amount based on the output of the vehicle speed detection unit after correction by a value, and specifies the vehicle position on the map based on the calculated second movement amount; .

  When the movement amount of the host vehicle is calculated based on the imaging results of the first imaging unit and the second imaging unit, the change in the relative position of the object may not be detected properly due to the decrease in the object detection accuracy. is there. On the other hand, when the amount of movement of the host vehicle is calculated based on the output from the vehicle speed detection unit instead of each imaging unit, the detection accuracy by each imaging unit is reduced, although errors constantly occur due to wheel slip or the like. There is an advantage that the amount of movement of the host vehicle can be calculated without being affected by conditions. In this regard, in the above configuration, the reliability of the first movement amount calculated by the change of the relative position of the feature points in time series is calculated. And when this reliability is more than a threshold value, while calculating the vehicle speed correction value for correct | amending the output of a vehicle speed detection part based on the 1st movement amount, the own vehicle position is calculated using the 1st movement amount. Identify. On the other hand, when the reliability is less than the threshold, the vehicle position is specified using the second movement amount calculated by correcting the output of the vehicle speed detection unit with the vehicle speed correction value. In this case, under conditions where the detection accuracy of each imaging unit is reduced, the vehicle position can be specified by the corrected output of the vehicle speed detection unit that is not easily affected by this condition, so that a decrease in accuracy of the vehicle position is suppressed. be able to.

1 is a configuration diagram of a vehicle control device 100. FIG. The figure explaining specification of the own vehicle position. The flowchart explaining the identification method of the own vehicle position CP on a map. The flowchart which shows the process of FIG.3 S16 in detail. The figure explaining tracking of the edge point P. FIG. The figure explaining tracking of the edge point P. FIG. The figure explaining the movement amount of the own vehicle CS. The figure explaining the setting of the reliability which concerns on 1st Embodiment. The figure explaining the change of the specified own vehicle position CP. The figure explaining the relationship between the similarity DS of the object which concerns on 2nd Embodiment, and the reliability of the 1st movement amount. The flowchart which shows the process implemented by FIG.3 S17 in detail. The figure explaining the calculation method of the reliability in 3rd Embodiment.

  Embodiments of a vehicle position specifying device and a vehicle position specifying method according to the present invention will be described with reference to the drawings. In the following embodiments, parts that are the same or equivalent to each other are denoted by the same reference numerals in the drawings, and the description of the same reference numerals is used.

(First embodiment)
FIG. 1 shows a vehicle control device 100 to which the vehicle position specifying device and the vehicle position specifying method are applied. The vehicle control device 100 is an example of a vehicle system mounted on a vehicle, for example, specifies the position of the host vehicle on a map, and supports driving of the host vehicle based on the specification result.

  As shown in FIG. 1, the vehicle control device 100 includes various sensors 30, an ECU 20 that functions as a vehicle position specifying device, and a driving support device 40.

  The various sensors 30 include a GPS receiver 31, a stereo camera 32, a vehicle speed sensor 35, and a yaw rate sensor 36.

  The GPS receiver 31 functions as part of a well-known satellite positioning system (GNSS), thereby receiving radio waves transmitted from the satellite as GPS information. The GPS information includes DOP (error information) that is information indicating an error in the position of the satellite, the time when the radio wave is transmitted, and the position of the satellite. The GPS receiver 31 calculates the distance from the satellite to the host vehicle CS based on the difference between the reception time when the GPS information is received and the transmission time included in the GPS information. Then, the calculated distance and the position of the satellite are output to the ECU 20.

  The stereo camera 32 is installed in the passenger compartment with the imaging axis facing the front of the host vehicle CS so that the front of the host vehicle CS can be imaged. In addition, the stereo camera 32 includes a first imaging unit 33 and a second imaging unit 34 that have different positions in the lateral direction of the vehicle. Each of the first imaging unit 33 and the second imaging unit 34 is constituted by, for example, a CCD image sensor or a CMOS image sensor, and is arranged so as to be shifted to the left and right in the lateral direction (X-axis direction) from the vehicle center. . For this reason, the right image and the left image captured by each of the image capturing units 33 and 34 are different in the angle at which the object is seen, and cause binocular parallax to the object in the image. In addition, images captured by the image capturing units 33 and 34 are output to the ECU 20 at predetermined intervals, respectively.

  The vehicle speed sensor 35 is provided on a rotating shaft that transmits power to the wheels of the host vehicle CS, and detects the speed of the host vehicle CS based on the number of rotations of the rotating shaft. The yaw rate sensor 36 detects a change in the direction of the host vehicle CS per unit time based on the yaw rate actually generated in the host vehicle CS, that is, the angular velocity around the center of gravity of the vehicle. The vehicle speed sensor 35 functions as a vehicle speed detection unit, and the yaw rate sensor 36 functions as a direction detection unit.

  ECU20 is comprised as a computer provided with CPU, ROM, and RAM. The CPU can function as each unit shown in FIG. 1 by executing a program stored in the internal memory. The ECU 20 is connected to an external memory 45 in which a map is recorded, and can read information recorded on the map from the external memory 45.

  Further, the ECU 20 specifies the own vehicle position CP on the current map. FIG. 2A shows the relative position Yr1 of the object Ob with reference to the own vehicle position CP (t1) at time t1, the relative position Yr2 of the object with reference to the own vehicle position CP (t2) at time t2, Is shown. If the position of the object Ob does not move, the movement amount TA of the host vehicle CS from the time t1 to the time t2 has a relationship equal to the difference between the relative position Yr1 at the time t1 and the relative position Yr2 at the time t2. Therefore, as shown in FIGS. 2B and 2C, the ECU 20 calculates the relative position of the object Ob with reference to the host vehicle CS based on the different time-series imaging results by the stereo camera 32. Then, the movement amount of the host vehicle CS is calculated based on the calculated change in the relative position. In the example of FIG. 2A, the vehicle position CP (t2) at time t2 can be specified by adding the movement amount TA to the vehicle position CP at time t1.

  The driving support device 40 controls the traveling of the host vehicle CS based on the host vehicle position CP specified by the ECU 20. For example, the driving support device 40 predicts the future position of the host vehicle CS based on the host vehicle position CP, the vehicle speed, and the yaw rate, and uses the predicted future position and the road recognition result to determine whether the host vehicle CS is a white line, etc. It is determined whether there is a risk of deviating from. For example, if the driving support device 40 has an alarm function, if it is determined that the host vehicle may deviate from the white line, a warning is displayed on the display included in the host vehicle CS, or the host vehicle CS includes A warning sound is generated by a speaker. Further, if the driving support device 40 has a driving assist function, when it is determined that the host vehicle CS may deviate from the white line, a steering force is applied to the steering device.

  In the above-described specification of the vehicle position CP using the imaging result of the stereo camera 32, the detection accuracy of the stereo camera 32 may be lowered depending on the surrounding brightness and the presence or absence of features for detecting an object. As the detection accuracy of the object Ob decreases, an error occurs in the relative position of the object Ob with respect to the host vehicle CS, and the accuracy of the calculated movement amount of the host vehicle CS is decreased. On the other hand, the movement amount calculated based on the output of the vehicle speed sensor 35 causes an error steadily due to wheel slip or the like. Therefore, the ECU 20 includes the components shown in FIG. 1 in order to suppress a decrease in the accuracy of the vehicle position CP accompanying the deterioration of the detection accuracy of the stereo camera 32.

  Returning to FIG. 1, the object recognizing unit 21 recognizes the three-dimensional information of the object based on the imaging result around the own vehicle by the stereo camera 32. The object recognition unit 21 calculates parallax information for each pixel block from the right image and the left image captured by the first imaging unit 33 and the second imaging unit 34, and three-dimensionally for each pixel based on the parallax information. A distance image including information (X, Y, Z) is generated. Then, each pixel is grouped based on the three-dimensional information, and each grouped pixel is recognized as an object. At this time, in addition to the three-dimensional information of each pixel, each pixel may be grouped as an object using the luminance distribution of each pixel and the direction of the moving direction vector calculated for each pixel.

  The feature point extraction unit 22 extracts the edge point P as the feature point of the object whose three-dimensional information is recognized by the object recognition unit 21. For example, the feature point extraction unit 22 extracts, as an edge point P, a pixel having a density gradient equal to or greater than a predetermined value using a known edge point extraction filter for an object for which three-dimensional information in the distance image is recognized.

  The movement amount calculation unit 23 tracks the change in the relative position of the edge point P in the object for which the three-dimensional information is recognized with reference to the own vehicle CS, and based on the tracked change in the relative position, the first amount of the own vehicle CS is tracked. Is calculated.

  The reliability setting unit 24 sets the reliability of the first movement amount. Here, the reliability is an index value indicating the calculation accuracy of the first movement amount. The higher the reliability, the higher the calculation accuracy of the first movement amount. Conversely, the lower the reliability, the lower the calculation accuracy of the first movement amount. As a method of calculating the reliability, for example, the reliability is calculated based on the number of edge points P used when calculating the first movement amount and the similarity of the object from which the edge points P are extracted.

  The correction value calculation unit 25 calculates a correction value for correcting the output of the vehicle speed sensor 35 mounted on the host vehicle CS based on the first movement amount when the reliability is equal to or greater than the threshold Th1. In this embodiment, a correction coefficient for changing the gain in the output of the vehicle speed sensor 35 is used as the correction value. The threshold value Th1 is a value calculated experimentally, for example, based on a value required as the calculation accuracy of the first movement amount.

  The first position specifying unit 26 specifies the vehicle position CP on the map based on the first movement amount when the reliability is equal to or greater than the threshold Th1. In addition, when the reliability is less than the threshold value Th1, the second position specifying unit 27 calculates the second movement amount based on the output of the vehicle speed sensor 35 after correction by the vehicle speed correction value α, and the calculated second position is calculated. The vehicle position CP on the map is specified based on the amount of movement.

  Next, a method for specifying the vehicle position CP on the map performed by the ECU 20 will be described with reference to FIG. The flowchart shown in FIG. 3 is a process performed by the ECU 20 at a predetermined cycle. In the following, the process of FIG. 3 that the ECU 20 is currently implementing is described as the current process, and is distinguished from the previous process.

  In step S11, GPS information is acquired. The ECU 20 acquires GPS information received by the GPS receiver 31. Step S11 functions as a satellite information acquisition unit.

  In step S12, the error of the GPS information acquired in step S11 is determined. For example, the ECU 20 determines the accuracy of the GPS information based on the DOP included in the GPS information. In addition to this, with reference to the map recorded in the external memory 45, when the current position of the host vehicle CS is a position where the reception accuracy of GPS information such as a tunnel is lowered, it is determined that the accuracy of the GPS information is low. May be.

  When the accuracy of the GPS information is high (step S12: YES), the process proceeds to step S13, and the position of the host vehicle CS on the map is corrected based on the GPS information. That is, when the accuracy of the GPS information is high, the vehicle position CP on the map is specified based on the GPS information. Then, the process of FIG. 3 is temporarily terminated.

  On the other hand, when the accuracy of the GPS information is low (step S12: NO), the vehicle position CP on the map is specified based on the output of the stereo camera 32 or the vehicle speed sensor 35. First, in step S14, three-dimensional information of objects around the host vehicle CS is recognized. The ECU 20 generates a distance image based on a pair of images imaged by the first imaging unit 33 and the second imaging unit 34 of the stereo camera 32, and the three-dimensional information (X, Y) of the object in the distance image. , Z). Step S14 functions as an object recognition process.

  In step S15, the edge point P in the object for which the three-dimensional information is recognized is extracted from the distance image. Step S15 functions as a feature point extraction step.

  In step S16, a change in the relative position of the edge point P extracted in step S15 with respect to the own vehicle CS is tracked, and a first movement amount of the own vehicle CS is calculated based on the tracked change in the relative position. . Step S16 functions as a movement amount calculation step.

  FIG. 4 is a flowchart showing in detail the process of step S16 of FIG. In step S31, it is determined whether or not the object from which the edge point P is extracted in step S15 includes a predetermined stationary object. For example, the ECU 20 records a dictionary for identifying utility poles, guardrails, signs, curbs, and the like as stationary objects, and determines whether or not the object from which the edge point P is extracted based on this dictionary is a stationary object. judge. In addition to this, it may be determined whether or not the object is a stationary object using the movement vector of the pixels constituting each object. Step S31 functions as an object determination unit.

  The distance image shown as an example in FIG. 5A includes objects Ob1 to Ob3 determined as stationary objects in the process of step S31 among the objects from which the edge points P are extracted. FIG. 5B is an enlarged view of a part of the object Ob2 shown in FIG. 5A, and the edge point P is extracted according to the density gradient of the pixels indicating the outer shape and texture of the object Ob2. .

  When the stationary object is not included in the object from which the edge point P is extracted (step S31: NO), the process of FIG. 4 is once ended. In this case, since the search for the change in the relative position at the edge point P cannot be performed, the processing of FIGS.

  On the other hand, if a stationary object is included (step S31: YES), in step S32, a search area for the edge point P in the distance image generated this time is set. For example, the ECU 20 records the host vehicle position CP on the previously specified map and the first movement amount calculated last time, and predicts the current moving direction of the host vehicle CS based on each information. . And the search area | region of the edge point P in the distance image produced | generated this time based on the predicted moving direction of the own vehicle CS is set. In FIG. 5C, EA1 to EA3 are set as search areas. Step S32 functions as a search area setting unit.

  In steps S33 to S35, edge point P tracking processing is performed. In this tracking process, as shown in FIG. 6A, edge points P are associated between distance images having different time series. Then, a change in relative position of the associated edge point P is calculated. In this embodiment, as an example, texture similarity between objects included in different distance images is determined, and edge points P in different distance images are associated with each other according to the similarity.

  First, in step S33, in order to determine the similarity of the texture of the object, a histogram indicating the luminance distribution around the edge point P included in each search area EA is calculated in the previously generated distance image. FIG. 6B is a histogram showing the luminance distribution around the edge point P included in the search area EA1 in the previously generated distance image. In this histogram, the luminance distribution around the edge point P is shown by setting the horizontal axis as the luminance B (for example, the number of gradations of 0 to 255) and the vertical axis as the pixel number PN of each luminance.

  In step S34, an area having a luminance distribution close to the luminance distribution of the edge point P calculated in step S33 is searched for each search area EA of the distance image generated this time. For example, the ECU 20 calculates a correlation coefficient with the luminance distribution calculated in step S33 for each search region in the distance image generated this time. Then, the pixel group having the highest correlation coefficient is determined as a region having a high texture similarity, and the edge point P included in the determined pixel group is associated. For example, in FIG. 6C, in the distance image generated this time, the luminance distribution around the edge point P included in the search area EA11 has the highest correlation coefficient with the luminance distribution shown in FIG. The edge point P of each pixel group is associated.

  In step S35, a change in the relative position of the edge point P is calculated. ECU20 calculates the difference of the three-dimensional information of each edge point P in the distance image from which the time series which matched in step S34 differed, and makes this difference the change value of a relative position.

  When tracking of all edge points P in each search area EA has not been performed (step S36: NO), the process returns to step S33, and the tracking of the edge points P is continued. On the other hand, when all the edge points P are tracked in each search area EA (step S36: YES), the process proceeds to step S37.

  In step S37, the first movement amount of the host vehicle CS is calculated based on the change in the relative position calculated in step S35. As shown in FIG. 7, the ECU 20 calculates the traveling direction component Vfy, the lateral direction component Vfx, and the rotational direction component Vfφ (angular velocity) of the host vehicle CS as the first movement amount. Here, the rotation direction component Vfφ is calculated by, for example, recognizing the center line of the road on which the host vehicle CS travels based on the imaging result of the stereo camera 32 and determining the traveling direction of the host vehicle CS with reference to the center line. Calculate by slope.

  If tracking of changes in relative positions at a plurality of edge points P is successful, changes in relative positions of the respective edge points P may be averaged, and the first movement amount may be calculated based on the average value. .

  Returning to FIG. 3, in step S17, the reliability of the movement amount calculated in step S16 is set. In the first embodiment, the ECU 20 calculates the number EN of the edge points P tracked in step S16, and sets the reliability lower as the calculated number EN of the edge points P is smaller. For example, the ECU 20 records a table that defines the correspondence between the number EN of edge points P used for tracking and the reliability. In the table shown in FIG. 8, the value is determined such that the reliability R1 increases as the number EN of the edge points P used for tracking increases. Step S17 functions as a reliability setting step.

In step S18, the reliability calculated in step S17 is compared using a threshold value Th1. If the reliability is equal to or higher than the threshold Th1 (step S18: YES), a vehicle speed correction value α for correcting the output of the vehicle speed sensor 35 is calculated in step S19. For example, the vehicle speed correction value α is calculated using the following formula (1).
α = Vfy / Vpy (1)
Here, Vfy is the vehicle speed of the traveling direction component in the first movement amount, and Vpy is the measurement value (output) of the vehicle speed sensor.

In step S20, a direction correction value β for correcting the output of the yaw rate sensor 36 is calculated. For example, the direction correction value β is calculated using the following formula (2).
β = Vfφ / Vpφ (2)
Here, Vfφ is a rotational direction component (angular velocity) in the first movement amount, and Vpφ is a measurement value (output) of the yaw rate sensor. Steps S19 and S20 function as a correction value calculation step.

  In step S21, the host vehicle position CP is specified based on the first movement amount calculated in step S16. The ECU 20 specifies a position changed on the map by the first movement amount calculated in step S16 from the vehicle position CP specified last time as the current vehicle position CP. Step S21 functions as a first position specifying step.

  On the other hand, if the reliability of the first movement amount is less than the threshold Th1 (step S18: NO), the process proceeds to step S22. In this case, since the reliability is less than the threshold Th1, if the first movement amount is used for specifying the own vehicle position CP, the accuracy of the own vehicle position CP on the map may be lowered. Therefore, in step S22 and subsequent steps, the ECU 20 specifies the host vehicle position CP using the output of the vehicle speed sensor 35.

  In step S22, the output of the vehicle speed sensor 35 corrected by each correction value is calculated as a second movement amount, and the host vehicle position CP on the map is specified based on the second movement amount. Specifically, the ECU 20 specifies the direction of the host vehicle CS based on a value obtained by integrating the output of the yaw rate sensor 36 corrected by the direction correction value β. Next, based on a value obtained by integrating the main force of the vehicle speed sensor 35 corrected by the vehicle speed correction value α, the amount of movement of the host vehicle CS in the traveling direction is calculated. For example, the corrected output of the yaw rate sensor 36 and the corrected output of the vehicle speed sensor 35 are calculated using the following equations (3) and (4).

AVpφ = βVpφ (3)
AVpy = αVpy (4)
Here, Apφ represents a value obtained by correcting the output of the yaw rate sensor 36 with the direction correction value β, and AVpy represents a value obtained by correcting the output of the vehicle speed sensor with the vehicle speed correction value α.

  In step S23, it is determined whether or not the vehicle position specified in step S22 is included in the error range calculated based on the error of the GPS information. For example, the ECU 20 sets an error range of the own vehicle position CP that is specified based only on the GPS information, based on the DOP included in the GPS information received in step S11. In FIG. 9, the error range ER is set concentrically with the position measured based on the GPS information as the center M.

  If the own vehicle position CP specified in step S22 is within the error range (step S23: YES), the process of FIG. 3 is once ended. In this case, the own vehicle position CP on the map identified in step S22 falls within the error range based on the GPS information, and thus the calculated position is set as the current own vehicle position CP.

  On the other hand, when the own vehicle position CP specified in step S22 exceeds the error range (step S23: NO), in step S24, the own vehicle position CP is changed to be within the error range. In this case, the ECU 20 changes the own vehicle position CP specified in step S22 to an error range value based on the GPS information. In FIG. 9, the position on the map is changed so that the vehicle position CP that is out of the error range falls within the error range.

  Steps S22 to S24 function as a second position specifying step. When the process of step S23 or the process of step S24 ends, the process of FIG. 3 is temporarily ended.

  As described above, in the first embodiment, the ECU 20 calculates the reliability of the first movement amount calculated by the change in the relative position of the edge point P in time series. When the reliability is equal to or greater than the threshold Th1, a vehicle speed correction value α for correcting the output of the vehicle speed sensor 35 is calculated based on the first movement amount, and the map is displayed using the first movement amount. The vehicle position CP at is identified. On the other hand, when the reliability is less than the threshold value, the output of the vehicle speed sensor 35 after correction by the vehicle speed correction value α is calculated as the second movement amount, and the own vehicle position is specified based on the calculated second movement amount. It was decided. In this case, under the condition that the detection accuracy of each imaging unit is lowered, the vehicle position CP can be specified by the corrected output of the vehicle speed sensor 35 that is not easily affected by this condition. Can be suppressed.

  The ECU 20 determines whether or not the object from which the edge point P is extracted is a predetermined stationary object that does not move, and based on the change in the relative position that tracks the edge point P of the object determined as a stationary object, A first movement amount of the host vehicle CS is calculated. Since the position of the edge point P extracted from the stationary object on the map does not change, tracking this edge point P reduces errors caused by changes in the relative position between the vehicle CS and the object. can do. As a result, it is possible to improve the reliability determination accuracy in the first movement amount.

  When the number of edge points P used to calculate the first movement amount is compared, the more the number of tracked points, the higher the robustness of the calculated first movement amount and the higher the calculation accuracy. . In this regard, in the above configuration, the reliability for the first movement amount is set according to the number of edge points P tracked by the ECU 20. In this case, the first movement amount with lower calculation accuracy can be easily replaced with the second movement amount using the output of the vehicle speed sensor 35, so that a decrease in the accuracy of the vehicle position can be suppressed.

  The second movement calculated by the corrected output of the vehicle speed sensor 35 by aligning the direction of the amount of movement used when calculating the correction of the measurement value with respect to the measurement value acquired by the vehicle speed sensor 35. The amount accuracy can be increased. In this regard, in the above configuration, the ECU 20 calculates the vehicle speed correction value α for correcting the output of the vehicle speed sensor 35 based on the traveling direction component of the host vehicle CS calculated as the first movement amount. In this case, it is possible to suppress a decrease in position accuracy in the traveling direction of the host vehicle CS on the map under the condition that the detection accuracy of each imaging unit is decreased.

  In addition to the amount of movement in the traveling direction by the vehicle speed sensor 35, the direction of the host vehicle CS is taken into account by the output of the yaw rate sensor, whereby the host vehicle position CP on the map can be specified with high accuracy. In this regard, in the above configuration, the ECU 20 calculates the direction correction value β for correcting the output of the yaw rate sensor 36 based on the direction calculated as the first movement amount, and the vehicle speed sensor corrected by the vehicle speed correction value α. In addition to the output of 35, the second movement amount is calculated based on the output of the direction detection unit after correction by the direction correction value. In this case, since the position of the host vehicle CS can be specified in consideration of the direction when the host vehicle CS is moving, the position accuracy of the host vehicle on the map is reduced even when the direction of the host vehicle CS is changed. Can be suppressed appropriately.

  When the position of the host vehicle CS on the map is estimated based on the satellite information transmitted from the satellite and the host vehicle position is specified using this position and the second movement amount, the specified host vehicle position is There is a possibility of being identified beyond the error range required by the satellite information. In this regard, in the above configuration, the ECU 20 compares the vehicle position CP specified based on the second movement amount with an error range, and if the specified vehicle position CP exceeds the error range, It was decided to change it to be included in the error range. For this reason, it is possible to prevent the specified vehicle position CP from deteriorating more than an error in information transmitted from the satellite.

  There is a high possibility that the region where the edge point P is extracted from the previously generated distance image and the region where the edge point P is extracted from the current generated distance image are close to each other. In this regard, in the above configuration, the ECU 20 makes a prediction based on the previously specified position on the map and the previously calculated first movement amount, and is generated this time based on the predicted movement direction of the host vehicle CS. The search area of the edge point P for tracking the change of the relative position in the obtained distance image is set. In this case, since the search area of the edge point P used for tracking can be narrowed down, the time required for calculating the first movement amount can be shortened.

(Second Embodiment)
In the second embodiment, the reliability is set according to the texture similarity of the object Ob to which the edge point P tracked in calculating the first movement amount belongs.

  As the texture similarity DS in the object used for the calculation of the first movement amount is higher, the mistake in associating the edge point P by the ECU 20 is reduced. Therefore, in the second embodiment, the higher the texture similarity DS in the object Ob from which the edge point P is extracted, the higher the reliability of the first movement amount is set.

  In this case, in step S17 of FIG. 3, the texture similarity DS between the objects Ob between the objects having different time series in which the three-dimensional information is recognized is calculated, and the first movement amount is determined according to the calculated similarity DS. Set the confidence level. For example, the texture similarity DS of the object Ob is calculated by calculating the correlation coefficient between the luminance distributions used in step S34 in FIG. Then, the smaller the correlation coefficient, the lower the similarity DS is calculated, and the higher the correlation coefficient, the higher the similarity DS is calculated.

  The ECU 20 records a table that defines the correspondence between the similarity DS and the reliability when calculating the reliability of the first movement amount from the texture similarity DS. In the table shown in FIG. 10, the value is determined so that the reliability R2 decreases as the object similarity DS between different distance images decreases.

  As described above, in the second embodiment, when the similarity of textures of objects used for tracking is compared, the higher the similarity is, the fewer correspondence errors between the edge points are, and the first movement amount. Is likely to be calculated appropriately. In this regard, in the above configuration, the ECU 20 sets the reliability with respect to the first movement amount in accordance with the texture similarity between objects with different imaging results by the first imaging unit 33 and the second imaging unit 34. did. In this case, since the first movement amount having a higher association error between the edge points P is more easily replaced with the second movement amount, the vehicle position accuracy is reduced due to the association error of the edge point P. Can be suppressed.

(Third embodiment)
In the third embodiment, the position of the same edge point P is calculated from different own vehicle positions CP, and the reliability of the first movement amount is calculated based on the shift amount of the position.

  FIG. 11 is a flowchart showing in detail the process performed in step S17 of FIG. FIG. 12 is a diagram illustrating a reliability calculation method according to the third embodiment.

  In step S41 in FIG. 11, the current vehicle position CP is specified based on the first movement amount calculated in step S16 in FIG.

  In step S42, a position AP (n-1) on the current relative coordinates of the edge point P in the previously generated distance image is calculated. The ECU 20 projects the relative position of the edge point P recognized by the three-dimensional information in the previous distance image onto the relative coordinates based on the vehicle position CP (n) specified this time, thereby The position AP (n-1) is calculated. For example, by changing each position of the edge point P in the distance image generated last time based on the first movement amount calculated this time, each position is changed to the position AP (n−1) in the relative coordinates. To do. Steps S41 and S42 function as a position calculation unit.

  In step S43, the position AP (n-1) on the relative coordinates of the edge point P calculated in step S42, and the position AP (n) of the same edge point P extracted in the distance image generated this time. The amount of deviation is calculated. FIG. 12A shows the position on the relative coordinate of each edge point P at each position. The ECU 20 sums the amount of deviation of the position of each edge point P on the calculated relative coordinates at all edge points.

  Here, when the edge point P is extracted from a stationary object, the position of each edge point P projected on the current relative coordinate should be the same position. Therefore, when the position of each edge point P on the relative coordinates is shifted, there is a high possibility that an error has occurred in the first movement amount used for calculating the vehicle position CP in step S41. Further, it is considered that the sum of the shift amounts of the positions of the edge points P on the map increases as the error of the first movement amount increases.

  Therefore, in step S44, the reliability is set according to the sum of the deviation amounts of the edge points P calculated in step S43. As the total sum of the shift amounts of the edge points P on the current relative coordinates increases, the first movement amount error is considered to be larger and the reliability is set to a lower value. For example, the ECU 20 records a table that defines the correspondence relationship between the sum PE of the deviation amounts calculated in step S43 and the reliability R3 of the first movement amount. In the table shown in FIG. 12B, the value is determined such that the reliability R3 decreases as the total amount PE of deviations increases.

  Returning to step S18 in FIG. 3, the vehicle position CP using the first movement amount or the vehicle position CP using the second movement amount is specified according to the set reliability R3. .

  In step S42, the ECU 20 calculates the relative position of the edge point P before the previous time as the position in the current relative coordinates of the edge point P based on the first movement amount calculated this time, and the calculated position and You may calculate the deviation | shift amount with the relative position of the same edge point P extracted by the imaging result this time.

  As described above, in the third embodiment, the position of the edge point P extracted from the stationary object does not change. Therefore, when the position of each edge point P in the time series different distance images is different, There is a possibility that an error has occurred in the first movement amount used for specifying the host vehicle position CP. In this regard, in the above configuration, the ECU 20 calculates the relative position of the edge point P at the previous or previous time as the position at the current relative coordinate of the edge point P based on the first movement amount calculated this time. At the same time, a deviation amount between the calculated position and the relative position of the same edge point P extracted from the current imaging result is calculated. The reliability is set to be lower as the calculated deviation amount is larger. In this case, by using the edge point P extracted from the stationary object on the map as a reference, the determination accuracy of the reliability in the first movement amount can be increased.

(Other embodiments)
The ECU 20 is configured to include the GPS receiver 31, but is not limited thereto, and may not include the GPS receiver 31. In this case, the ECU 20 does not have a function of specifying the vehicle position based on the GPS information shown in step S13 of FIG.

  The vehicle position CP on the map may be specified based only on the amount of movement of the host vehicle CS in the traveling direction. In this case, in step S22 of FIG. 3, the direction of the host vehicle CS is not specified, and based on the value obtained by integrating the main force of the vehicle speed sensor 35 corrected by the vehicle speed correction value α, the traveling direction of the host vehicle CS is increased. Only the amount of movement is calculated.

  In the first embodiment and the second embodiment described above, the object from which the edge point P is extracted is not limited to a predetermined stationary object, and may be another vehicle or a pedestrian traveling at a low speed on an adjacent traveling lane. . For example, when the host vehicle position CP on the map is specified based on the edge point P extracted from the object traveling at low speed, the ECU 20 changes the relative position of the object calculated by the edge point P extracted from the object. The first movement amount of the host vehicle is calculated by adding the output of the vehicle speed sensor 35. Then, the vehicle position on the map is specified based on the calculated first movement amount.

  The map referred to by the ECU 20 may be downloaded from a server (not shown) via a network in addition to being recorded in the external memory 45 connected to the ECU 20. Moreover, when ECU20 can access terminals, such as a smart phone, via I / F, you may acquire a map via this terminal.

  DESCRIPTION OF SYMBOLS 21 ... Object recognition part, 22 ... Feature point extraction part, 23 ... Movement amount calculation part, 24 ... Reliability setting part, 25 ... Correction value calculation part, 26 ... 1st position specific part, 27 ... 2nd position specific part, 33 ... 1st imaging part, 34 ... 2nd imaging part, CP ... Own vehicle position, CS ... Own vehicle.

Claims (10)

An object recognizing unit for recognizing three-dimensional information of an object based on a result of imaging around the vehicle by the first imaging unit (33) and the second imaging unit (32);
A feature point extraction unit for extracting feature points of the recognized object;
A movement amount calculation unit that tracks a change in the relative position of the feature point with respect to the own vehicle and calculates a first movement amount of the own vehicle based on the tracked change in the relative position;
A reliability setting unit for setting the reliability of the first movement amount;
When the reliability is greater than or equal to a threshold value, a correction value calculation unit that calculates a vehicle speed correction value for correcting the output of the vehicle speed detection unit (35) mounted on the host vehicle based on the first movement amount. When,
A first position specifying unit for specifying a vehicle position on a map based on the first movement amount when the reliability is equal to or greater than a threshold;
When the reliability is less than a threshold value, a second movement amount is calculated based on the output of the vehicle speed detection unit after correction by the vehicle speed correction value, and the map is based on the calculated second movement amount. A host vehicle position specifying device comprising: a second position specifying unit that specifies the host vehicle position. An object determination unit that determines whether the object from which the feature points are extracted is a predetermined stationary object that does not move;
The said movement amount calculation part calculates the 1st movement amount of the said own vehicle based on the change of the said relative position of the said feature point in the said object determined as the said stationary object. Car location device.   3. The reliability setting unit calculates the number of feature points tracked by the movement amount calculation unit, and sets the reliability lower as the calculated number of feature points is smaller. Vehicle positioning device as described in 1. The movement amount calculation unit compares textures between the objects having different time series of the imaging results of the first imaging unit and the second imaging unit, and is extracted from each object having a similar texture By tracking the change in the relative position of the feature point, the first movement amount is calculated,
3. The own vehicle position specifying device according to claim 1, wherein the reliability setting unit sets the reliability lower as the texture similarity between the objects from which the feature points are extracted is lower. . Based on the first movement amount calculated this time, the relative position of the feature point before or before the previous time is calculated as a position in the current relative coordinates of the feature point, and the calculated position and the current position A position calculation unit that calculates a deviation amount from the relative position of the same feature point extracted by the imaging result;
The vehicle position specifying device according to claim 2, wherein the reliability setting unit sets the reliability low as the calculated deviation amount increases. The movement amount calculation unit calculates a traveling direction component of the host vehicle as the first movement amount,
The vehicle position according to any one of claims 1 to 5, wherein the correction value calculation unit calculates the vehicle speed correction value based on the traveling direction component calculated by the movement amount calculation unit. Specific device. A direction calculation unit for calculating the direction of the host vehicle;
The correction value calculation unit corrects the output of the direction detection unit (36) mounted on the host vehicle based on the direction calculated by the movement amount calculation unit when the reliability is equal to or greater than a threshold value. Calculate the orientation correction value for
The second position specifying unit calculates the second movement amount based on the output of the vehicle speed detection unit after correction by the vehicle speed correction value and the output of the direction detection unit after correction by the direction correction value. The host vehicle position specifying device according to claim 6, wherein the host vehicle position on the map is specified based on the calculated second movement amount. A satellite information acquisition unit that acquires satellite information including the position of the satellite transmitted from the satellite and error information of the position of the satellite;
The second position specifying unit sets the specified vehicle position as the error range when the vehicle position specified based on the second movement amount exceeds an error range calculated based on the error information. The own vehicle position specifying device according to any one of claims 1 to 7, which is changed to be included. The object recognizing unit generates a distance image to which information indicating three-dimensional information of the object is added based on each image captured by the first imaging unit and the second imaging unit, and the distance image Recognizing the three-dimensional information of the object based on
The movement direction of the host vehicle based on the host vehicle position on the map specified last time based on the first movement amount or the second movement amount and the first movement amount calculated last time. And, based on the predicted movement direction of the host vehicle, the movement amount calculation unit tracks the feature point that is tracked to calculate the first movement amount with respect to the distance image generated this time. The own vehicle position specifying device according to any one of claims 1 to 8, further comprising a search area setting unit that sets a search area. An object recognition step of recognizing three-dimensional information of an object based on an imaging result around the vehicle by the first imaging unit and the second imaging unit;
A feature point extraction step of extracting feature points of the recognized object;
A movement amount calculating step of tracking a change in relative position of the feature point with respect to the own vehicle and calculating a first movement amount of the own vehicle based on the tracked change in the relative position;
A reliability setting step of setting the reliability of the first movement amount;
A correction value calculating step of calculating a vehicle speed correction value for correcting an output of a vehicle speed detection unit mounted on the host vehicle based on the first movement amount when the reliability is equal to or greater than a threshold;
A first position specifying step of specifying a vehicle position on a map based on the first movement amount when the reliability is equal to or greater than a threshold;
When the reliability is less than a threshold value, a second movement amount is calculated based on the output of the vehicle speed detection unit after correction by the vehicle speed correction value, and the map is based on the calculated second movement amount. A vehicle position specifying method comprising: a second position specifying step of specifying the vehicle position.

Images