JP4817019B2 - Own vehicle position recognition device and own vehicle position recognition program - Google Patents

Description

  The present invention relates to a host vehicle position recognition device and a host vehicle position recognition program for correcting host vehicle position information using a detection result of a course change operation of the host vehicle.

  In recent years, for example, in the field of navigation devices, in order to enable more appropriate route guidance and the like, it is required to recognize the position of the vehicle with high accuracy. As a technique related to recognition of such a vehicle position, for example, the following in-vehicle navigation device is disclosed in Patent Document 1 below. This device is mounted on a vehicle, detects a crossing by identifying a crossing symbol from a camera for recognizing a landscape along a traveling road ahead of the vehicle, and image information of a landscape promoted by the camera. An intersection detection unit for obtaining a distance to an intersection (intersection symbol) and vehicle position correcting means for correcting the vehicle position to a point on the traveling road that is separated from the intersection position obtained from map data by the distance.

  This device detects intersections by identifying intersection symbols (traffic signals, pedestrian crossings, white lines for median strips, etc.) using an in-vehicle camera, and then obtains the distance from the vehicle position to the intersection. Then, the vehicle position correcting means corrects the vehicle position to a position on the traveling road that is separated from the intersection position obtained from the map data by a distance from the vehicle position to the intersection. At this time, the distance from the vehicle position to the intersection is obtained by referring to the pixel / distance correspondence table using the camera pixel where the identified intersection symbol is located.

JP-A-9-243389

  In the technique described in Patent Document 1, it is necessary to perform image recognition processing for identifying intersection symbols such as traffic lights and pedestrian crossings from image information acquired by a camera. Furthermore, it is necessary to calculate and obtain the distance from the position of the host vehicle to the intersection position based on the arrangement of the intersection symbol in the identified image information. However, in order to accurately recognize the type and position of the intersection symbol included in the image information by the image recognition process, it is necessary to perform a very advanced image recognition process. Therefore, when the state of the image information is bad, recognition failure tends to occur, and in this case, the vehicle position information cannot be corrected.

  On the other hand, the course changing operation of the host vehicle at the intersection can be detected at a very high rate by comparing the detection result of the traveling direction of the host vehicle with a gyroscope or the like and the road information. is there. However, a technique for correcting the own vehicle position information by effectively using the detection result of the course change operation of the own vehicle at such an intersection has not been known so far.

  The present invention has been made in view of the above problems, and an object of the present invention is to recognize the position of the host vehicle that can appropriately correct the position information of the host vehicle using the detection result of the course change operation of the host vehicle. It is to provide an apparatus and a vehicle position recognition program.

  The characteristic configuration of the vehicle position recognition device according to the present invention for achieving the above object includes vehicle position information acquisition means for acquiring vehicle position information representing the current position of the vehicle, and road information for acquiring road information. Learning an acquisition means, an image information acquisition means for acquiring image information around the own vehicle, an image recognition means for performing an image recognition process of a target feature included in the image information, and a past course change operation of the own vehicle The route change information that can identify the route change intersection that is the intersection where the route change operation of the host vehicle has been performed and the target feature that has been image-recognized before the route change operation is generated. The target feature is recognized by the course change learning information storage means storing the course change learning information having the distance information from the recognition position to the position where the course change operation of the host vehicle is performed, and the image recognition means. If Based on the course change learning information for the target feature, a position that has traveled a distance indicated by the distance information from the recognized position of the target feature based on the vehicle position information is identified by the course change information. Vehicle position correcting means for correcting the vehicle position information so as to coincide with the position in the road information of the route change intersection to be performed.

  According to this characteristic configuration, the same course change learning information obtained by relating the past course change operation of the host vehicle to the information on the distance to the target feature that has been image-recognized before the course change operation is the same. When the target feature is image-recognized, the vehicle position information can be corrected according to the position in the road information of the course change intersection based on the course change learning information. That is, the vehicle position information can be appropriately corrected using information obtained by learning the detection result of the past course change operation of the host vehicle and the image recognition result of the target feature before the detection. At this time, since the course change learning information generated by learning the past course change operation of the host vehicle is used, the course change of the current host vehicle at the course change intersection specified by the course change learning information is performed. The vehicle position information can be corrected regardless of whether or not there is an operation.

  Furthermore, since the own vehicle position information is corrected according to the position in the road information of the course change intersection specified by the course change learning information, the own vehicle position information advances ahead of the actual position of the own vehicle due to an error. Even when the vehicle is recognized as such, the vehicle position indicated in the vehicle position information can be prevented from proceeding beyond the route change intersection. Therefore, for example, when the vehicle position is displayed on a map when the vehicle position recognition device is applied to a navigation device, the vehicle is in a state where the vehicle position information is ahead of the actual position of the vehicle. Even when the course is changed, it is possible to prevent the position of the vehicle on the map from jumping from the position straight ahead beyond the course change intersection onto the road after the course change. Similarly, when the vehicle changes its course from a state in which the vehicle position information is later than the actual position of the vehicle, after the course is changed from the road where the vehicle position on the map was going straight Can be prevented from jumping on the road.

  Here, the own vehicle position correcting means indicates that the own vehicle position information is indicated by the own vehicle position information when the own vehicle position information is recognized as being ahead of the actual position of the own vehicle due to an error. The host vehicle position indicated in the host vehicle position information until the host vehicle actually travels the distance indicated in the distance information from the recognized position of the target feature after the position has reached the route change intersection. It is preferable that the vehicle position information is corrected by stopping the vehicle on the route change intersection.

  According to this configuration, when the vehicle position information is recognized to be ahead of the actual position of the vehicle due to an error, the vehicle position indicated in the vehicle position information is determined by the course change learning information. It is possible to prevent the vehicle from proceeding beyond the specified route change intersection, and to appropriately correct the vehicle position information in accordance with the position in the road information of the route change intersection.

  In addition, when the vehicle position information is recognized so that the vehicle position information is delayed behind the actual position of the vehicle due to an error, the vehicle position correction unit is configured to detect the vehicle from the recognition position of the target feature. When the vehicle travels the distance indicated by the distance information, it is preferable to correct the vehicle position information by moving the vehicle position indicated by the vehicle position information onto the route change intersection.

  According to this configuration, when the vehicle position information is recognized as being behind the actual position of the vehicle due to an error, the position in the road information of the route change intersection specified by the route change learning information Accordingly, the vehicle position information can be corrected appropriately.

  Further, the road information is information representing a road by a connection relationship between a plurality of links and nodes, and the route change information is information on a route of a link traveled by the own vehicle during the route change operation, and the route A configuration including one or both of the node information corresponding to the intersection where the changing operation is performed is preferable.

  According to this configuration, the route change information can be information that can appropriately specify a route change intersection in relation to road information that represents a road by a connection relationship between a plurality of links and nodes.

  Further, it is preferable that the road information acquisition unit acquires the road information around the host vehicle from a predetermined map information storage unit based on the host vehicle position information.

  According to this configuration, road information necessary for correcting the vehicle position information can be appropriately acquired.

  Further, a route change detection means for detecting a course change operation of the host vehicle, and a target feature recognized by the image recognition means prior to the detection of the course change operation of the host vehicle by the route change detection means, Distance information acquisition means for acquiring distance information indicating the distance from the recognition position of the object to the position where the course of the host vehicle has been changed, and a path that can identify the course change intersection that is the intersection where the course change operation has been performed Detection operation information storage means for storing detection operation information including change information and the distance information acquired by the distance information acquisition means for the course change operation, and the same course change operation is detected a plurality of times. Course change learning that generates the course change learning information based on a plurality of the detected action information for the same course change operation stored in the detected action information storage means Comprising a multi-address generating means, wherein the course change learning information generated by the course change learning information generating means, it is preferable that a configuration stored in the course change learning information storing section.

  According to this configuration, the course change operation of the host vehicle is appropriately learned in relation to the distance information with the target feature that has been image-recognized before the course change operation, and the learning result is the course change learning information. Can be stored in the course change learning information storage means. At this time, even when the detection operation information as a detection result of each of the course change operations of the own vehicle includes an error, when generating the course change learning information, a plurality of detection operation information is used. Errors can be averaged. Therefore, it is possible to improve the accuracy of the learning result of the course changing operation of the host vehicle.

  Further, a feature information storage means storing feature information including position information and attribute information about a plurality of target features, and the surroundings of the vehicle from the feature information storage means based on the vehicle position information. And a feature information obtaining unit that obtains the feature information, and the image recognition unit performs a recognition process of the target feature included in the image information based on the feature information. Is preferred.

  According to this configuration, since the image recognition process can be performed based on the feature information around the own vehicle acquired from the feature information storage unit, the recognition rate of the target feature is increased and erroneous recognition is suppressed. Is possible. Therefore, it is possible to improve the accuracy of the process of correcting the vehicle position information and generating the course change learning information.

  The target feature is preferably a road marking provided on a road surface.

  According to this configuration, since a road sign that is relatively easy to recognize images as a target feature compared to a three-dimensional feature such as a road sign, a guide sign, and a traffic light is used as a target feature, the recognition rate of the target feature Can be increased. Therefore, it is possible to increase the frequency at which the processing for correcting the vehicle position information and the generation of the course change learning information can be performed.

  The characteristic configuration of the navigation device according to the present invention includes a vehicle position recognition device having the above-described configurations, map information storage means storing map information including the road information, the map information, and the vehicle position. An application program that operates by referring to the vehicle position information recognized by a recognition device, and a guide information output unit that operates according to the application program and outputs guide information.

  According to this feature configuration, guidance information can be output based on the corrected vehicle position information corrected using the course change learning information. Therefore, for example, the display of the vehicle position and route guidance can be performed more appropriately.

  The characteristic configuration of the vehicle position recognition program according to the present invention includes: a vehicle position information acquisition step that acquires vehicle position information that represents a current position of the vehicle; a road information acquisition step that acquires road information; An image information acquisition step for acquiring peripheral image information, an image recognition step for performing image recognition processing of a target feature included in the image information, and when the target feature is recognized by the image recognition step, It was generated by learning the past course change operation of the vehicle, and the course change information that can identify the course change intersection that is the intersection where the course change operation of the host vehicle was performed, and the image was recognized before the course change operation. A route change learning information storage device storing route change learning information having distance information from the recognition position of the target feature to a position where the route change operation of the host vehicle is performed for the target feature From the course change learning information acquisition step for acquiring the course change learning information for the recognized target feature, and the vehicle position based on the course change learning information acquired by the course change learning information acquisition step The own vehicle so that a position traveled by a distance indicated by the distance information from a recognized position of the target feature based on the information coincides with a position in the road information of the course change intersection specified by the course change information. The vehicle position correction step for correcting the position information is executed by a computer.

  According to this characteristic configuration, the same course change learning information obtained by relating the past course change operation of the host vehicle to the information on the distance to the target feature that has been image-recognized before the course change operation is the same. When the target feature is image-recognized, the vehicle position information can be corrected according to the position in the road information of the course change intersection based on the course change learning information. That is, the vehicle position information can be appropriately corrected using information obtained by learning the detection result of the past course change operation of the host vehicle and the image recognition result of the target feature before the detection. At this time, since the course change learning information generated by learning the past course change operation of the host vehicle is used, the course change of the current host vehicle at the course change intersection specified by the course change learning information is performed. The vehicle position information can be corrected regardless of whether or not there is an operation.

  Furthermore, since the own vehicle position information is corrected according to the position in the road information of the course change intersection specified by the course change learning information, the own vehicle position information advances ahead of the actual position of the own vehicle due to an error. Even when the vehicle is recognized as such, the vehicle position indicated in the vehicle position information can be prevented from proceeding beyond the route change intersection. Therefore, for example, when the vehicle position is displayed on the map when this vehicle position recognition program is applied to the navigation program of the navigation device, the vehicle position information is advanced from the actual position of the vehicle. Even when the host vehicle changes the course, it is possible to prevent the position of the host vehicle on the map from jumping from the position where the host vehicle goes straight beyond the course change intersection onto the road after the course change.

  Next, embodiments of the present invention will be described with reference to the drawings. FIG. 1 is a block diagram showing a schematic configuration of a navigation device 1 according to the present embodiment. The navigation device 1 includes a host vehicle position recognition device 2 as an embodiment of the present invention. The navigation device 1 learns the course change operation of the host vehicle C (see FIG. 4) in relation to the information on the distance from the target feature that has been image-recognized before the course change operation. When the same target feature is image-recognized using the information S, the vehicle position information P is matched with the position in the road information Ra of the course change intersection N (see FIG. 7) based on the course change learning information S. To correct. In addition, in order to acquire such course change learning information S, the navigation device 1 detects the course change operation of the host vehicle C, and the target feature that has been image-recognized prior to the detection of the course change operation. Learning in relation to the distance information. Furthermore, in this embodiment, the navigation apparatus 1 also learns the feature information F that also represents the position of the target feature, the feature attribute, and the like based on the image recognition result of the target feature.

  The functional units of the navigation device 1 shown in FIG. 1 are hardware units that perform various processes on input data using a common arithmetic processing unit such as a CPU as a core member. Alternatively, it is implemented by software (program) or both. Each of these functional units is configured to exchange information with each other. Further, each of the databases DB1 to DB5 of the navigation device 1 is a recording medium that can store information, such as a hard disk drive, a DVD drive equipped with a DVD-ROM, a CD drive equipped with a CD-ROM, and its drive. And a device having a means as a hardware configuration. Hereinafter, the configuration of each part of the navigation device 1 according to the present embodiment will be described in detail.

1. Map database The map database DB1 is a database in which map information M divided into predetermined sections is stored. FIG. 2 is a diagram illustrating an example of the configuration of the map information M stored in the map database DB1. As shown in this figure, the map information M has road information Ra representing a road network by connection relations between a large number of nodes n corresponding to intersections and links k corresponding to roads connecting the intersections. . Each node n has information on the position (coordinates) on the map expressed by latitude and longitude. Each link k is connected via a node n. Each link k has information such as a road type, a link length, a road width, and a shape interpolation point for expressing a link shape as attribute information. Here, the road type information is information on the road type when the road is divided into a plurality of types, such as an automobile-only road, a city road, a narrow street, a mountain road, and the like. The attribute information of these links k corresponds to the road attribute information Rb (see FIG. 1). And in this embodiment, this map database DB1 functions as a map information storage means. In FIG. 2, only the road information Ra of one section is illustrated, and the road information Ra of other sections is omitted.

2. Feature database The feature database DB2 is a database in which information on various features provided on or around the road, that is, feature information F is stored. As shown in FIG. 1, in this embodiment, two types of information, initial feature information Fa and learning feature information Fb, are stored in the feature database DB2. Here, the initial feature information Fa is feature information F about a plurality of features that are prepared and stored in advance in the feature database DB2. Such initial feature information Fa is maintained only in some areas such as the periphery of a large city and a main road among all areas where the map information M including the road information Ra is maintained. On the other hand, the learning feature information Fb is generated by the learning feature information generation unit 44 as a result of learning using the image recognition result of the target feature by the image recognition unit 24, as will be described later, and is stored in the feature database DB2. This is feature information F to be stored. In the following description, when “feature information F” is simply referred to, the initial feature information Fa and the learned feature information Fb are collectively referred to. In this embodiment, this feature database DB2 functions as a feature information storage unit.

  Features for which feature information F is stored in the feature database DB2 include road markings (paint markings) provided on the road surface. FIG. 3 is a diagram illustrating an example of the feature information F of the road marking stored in the feature database DB2. Features such as road markings include, for example, pedestrian crossings, stop lines, speed markings representing maximum speed, zebra zones, lane markings that divide lanes along the road (solid lines, broken lines, double lines, etc.) Including various lane markings), traffic direction markings by direction of travel that specify the direction of travel of each lane (including arrow markings such as straight arrows, right turn arrows, etc.), and the like. In addition to such road markings, the features in which the feature information F is stored can include various features such as traffic lights, signs, overpasses, and tunnels.

  The feature information F includes position information of each feature and feature attribute information associated therewith as its contents. Here, the position information includes information on the position (coordinates) on the map of the representative point of each feature associated with the link k or the node n constituting the road information Ra, and the direction of each feature. In this example, the representative point is set near the center in the length direction and width direction of each feature. The feature attribute information includes identification information (feature ID) for identifying each feature from other features, type information indicating the feature type of each feature, or the shape, size, and color of the feature. The feature form information such as is included. Here, the feature type specifically indicates the type of feature having basically the same form such as “pedestrian crossing”, “stop line”, “speed indication (30 km / hour)”, and the like. Information. In addition, the feature information F may include feature-related information that represents a relationship with other nearby features and distance information between features that represents the distance between the features with the other features. Is preferred. Here, the feature-related information can predict other features existing ahead by recognizing one feature image while the host vehicle C (see FIG. 4) is traveling along the road. It is information to do. Further, the distance information between features is information for accurately predicting the distance from the host vehicle C to such a feature existing ahead.

3. Feature Learning Database The feature learning database DB3 is a database that stores recognized feature information A generated by a recognized feature information generating unit 42 described later. In this feature learning database DB3, recognized feature information A for each of a plurality of target features that have been successfully recognized by the image recognition unit 24 is stored. Specific contents of the recognized feature information A stored in the feature learning database DB3 will be described in detail later. In the present embodiment, this feature learning database DB3 functions as a recognized feature storage means.

4). Course Change Learning Database The course change learning database DB4 is a database that stores detection operation information B generated by a detection operation information generation unit 48 described later. In this course change learning database DB4, detection operation information B for each of a plurality of course change operations detected by the course change detection unit 17 is stored. Specific contents of the detection operation information B stored in the course change learning database DB4 will be described in detail later. In this embodiment, this course change learning database DB4 functions as a detection operation information storage unit.

5). Course Change Database The course change database DB5 is a database that stores the course change learning information S generated by the course change learning information generation unit 50 described later. In the course change database DB5, course change learning information S for each of the plurality of course change operations detected by the course change detection unit 17 is stored. Specific contents of the course change learning information S stored in the course change database DB5 will be described in detail later. In this embodiment, this course change database DB5 functions as a course change learning information storage unit.

6). Image Information Acquisition Unit The image information acquisition unit 12 functions as an image information acquisition unit that acquires image information G around the host vehicle captured by the imaging device 11. Here, the imaging device 11 is an in-vehicle camera or the like provided with an imaging element, and is provided at a position where at least a road surface of the road around the host vehicle C can be imaged. As such an imaging device 11, for example, it is preferable to use a back camera that images the road surface behind the host vehicle C as shown in FIG. The image information acquisition unit 12 captures imaging information captured by the imaging device 11 at predetermined time intervals via a frame memory (not shown). The time interval for capturing the image information G at this time can be set to, for example, about 10 to 50 ms. Thereby, the image information acquisition unit 12 can continuously acquire the image information G of a plurality of frames captured by the imaging device 11. The acquired image information G is output to the image recognition unit 24.

7). Own vehicle position information acquisition unit The own vehicle position information acquisition unit 16 functions as own vehicle position information acquisition means for acquiring own vehicle position information P indicating the current position of the own vehicle C. Here, the vehicle position information acquisition unit 16 is connected to the GPS receiver 13, the direction sensor 14, and the distance sensor 15. Here, the GPS receiver 13 is a device that receives GPS signals from GPS (Global Positioning System) satellites. This GPS signal is normally received every second and is output to the vehicle position information acquisition unit 16. The own vehicle position information acquisition unit 16 can analyze a signal from a GPS satellite received by the GPS receiver 13 and acquire information such as a current position (coordinates), a traveling direction, and a moving speed of the own vehicle C. . The direction sensor 14 is a sensor that detects the traveling direction of the host vehicle C or a change in the traveling direction. The azimuth sensor 14 is constituted by, for example, a gyroscope, a geomagnetic sensor, or the like. Then, the direction sensor 14 outputs the detection result to the vehicle position information acquisition unit 16. The distance sensor 15 is a sensor that detects the vehicle speed and the moving distance of the host vehicle C. The distance sensor 15 is, for example, a vehicle speed pulse sensor that outputs a pulse signal every time a drive shaft or wheel of the vehicle rotates by a certain amount, a yaw / G sensor that detects the acceleration of the host vehicle C, and an integration of the detected acceleration. Configured by a circuit or the like. Then, the distance sensor 15 outputs information on the vehicle speed and movement distance as the detection result to the vehicle position information acquisition unit 16. In the present embodiment, the azimuth sensor 14 and the distance sensor 15 also output the detection results to the course change detection unit 17.

  Based on the outputs from the GPS receiver 13, the azimuth sensor 14, and the distance sensor 15, the own vehicle position information acquisition unit 16 performs a calculation for specifying the own vehicle position by a known method. In addition, the vehicle position information acquisition unit 16 acquires road information Ra around the vehicle position extracted from the map database DB1 by a road information acquisition unit 23, which will be described later, and performs known map matching based on the acquired road information Ra. Correction for matching the vehicle position to the road indicated by the road information Ra is also performed. Further, in the present embodiment, the vehicle position information P acquired by the vehicle position information acquisition unit 16 is corrected by the vehicle position information correction unit 51 in order to further improve the position accuracy. The configuration of the vehicle position information correction unit 51, which is the most characteristic configuration in this embodiment, will be described in detail later. Thus, the own vehicle position information acquisition unit 16 obtains the own vehicle position information P including the current position information of the own vehicle C represented by the coordinates (latitude and longitude) and the traveling direction information of the own vehicle C. get.

8). Course Change Detection Unit The course change detection unit 17 functions as a course change operation detection unit that detects a course change operation of the host vehicle C. As shown in FIG. 1, the course change detection unit 17 detects a course change operation of the host vehicle C in response to input of signals from various sensors provided in each part of the host vehicle C. Various sensors that output signals to the course change detection unit 17 include, for example, the steering sensor 19 and the acceleration sensor 20 in addition to the direction sensor 14 and the distance sensor 15 described above. Here, the steering sensor 19 is a sensor that detects a steering operation by the driver of the host vehicle C, such as an optical rotation sensor attached to the rotating part of the steering wheel, a rotary resistance volume, an angle sensor attached to the wheel part, and the like. Consists of. The acceleration sensor 20 is a sensor that detects acceleration in the traveling direction and lateral direction of the host vehicle C.

  In the present embodiment, the course change operation of the host vehicle C detected by the course change detection unit 17 is an auto turn such as a right turn or a left turn other than a case where the host vehicle C travels straight on a road on the road at the intersection. This is the operation of the vehicle C. Then, the course change detection unit 17 detects the travel locus of the host vehicle C detected by the direction sensor 14 and the distance sensor 15, the steering operation amount and operation time detected by the steering sensor 19, and the self-detection detected by the acceleration sensor 20. Based on information such as acceleration acting on the vehicle C, a route change operation such as a right turn or a left turn of the host vehicle C is detected.

  The course change detection unit 17 includes a course change information generation unit 18. When the course change operation of the host vehicle C is detected by the course change detection unit 17, for example, as shown in FIG. 7, the course change information generation unit 18 generates the course change information Ba representing the content of the course change operation. It functions as a route change information generating means for generating. The route change information Ba constitutes a part of the detection operation information B and the route change learning information S as described later. In the present embodiment, the route change information Ba includes the mode information Bd indicating the route change mode and the intersection specification information Be for specifying the route change intersection N that is the intersection where the route change operation is performed. It is configured.

  Here, the mode information Bd is information representing a classification for identifying the type of the course changing operation detected by the course changing detector 17 with respect to other course changing operations. Therefore, the mode information Bd is information such as “left turn”, “right turn”, “turn (U-turn)”, and the like. The classification of the mode information Bd is merely an example, and the level of subdivision of these classifications can be arbitrarily determined.

  The intersection specifying information Be is information for specifying the course changing intersection N where the course changing operation of the host vehicle C is performed. In the present embodiment, the information on the route of the link k (see FIG. 7A) through which the host vehicle C traveled during the course changing operation is the intersection specifying information Be. Specifically, as shown in FIG. 7B, the intersection identification information Be includes a plurality of links k that the host vehicle C has traveled during a course change operation at the course change intersection N, and the plurality of links k. In order to specify the order in which the host vehicle C has passed, link identification information (here, link IDs) is arranged in the order in which the host vehicle C has passed. In the example of FIG. 7B, the intersection specifying information Be is “ID = ◯◯◯◯ → ID = ΔΔΔΔ”. Therefore, according to the intersection specifying information Be, the node n (ID = ID) of the connection point of the two links k “ID = XXX” and “ID = ΔΔΔΔ” shown in FIG. It can be specified based on the road information Ra that the intersection corresponding to (XX) is the course change intersection N.

  By the way, with respect to the intersection specifying information Be, for example, as shown in FIG. 8A, if the vehicle position information P is recognized to be ahead of the actual position of the vehicle C due to an error, In a state after the vehicle position indicated in the vehicle position information P has advanced to the link k of ID = □□□□ in the straight direction beyond the node n (ID = ○ × ○ ×) of the route change intersection N, The actual route change operation of the left turn by the own vehicle C is performed. Therefore, a position jump of the own vehicle position information P occurs, and the own vehicle position indicated in the own vehicle position information P moves to the link k of ID = ΔΔΔΔ after the course change. In this case, the link ID array of the link k through which the vehicle position indicated by the vehicle position information P passes during the course changing operation is “ID = XXXXXX → ID = □□□□ → ID = △△△△ 」. However, based on the road information Ra, it can be determined that the actual host vehicle C does not pass the link k of ID = □□□□. Therefore, the intersection specifying information Be in this case is “ID = ○○○ ○ → ID = ΔΔΔΔ ”. On the other hand, for example, as shown in FIG. 8B, when the vehicle position information P is recognized as being delayed behind the actual position of the vehicle C due to an error, the vehicle position information P is indicated. Turn left by the actual own vehicle C in the state where the own vehicle position is in progress of the link k of ID = ○○○○ and before reaching the node n (ID = ○ × ○ ×) of the route change intersection N The course changing operation is performed. Therefore, a position jump of the own vehicle position information P occurs, and the own vehicle position indicated in the own vehicle position information P moves to the link k of ID = ΔΔΔΔ after the course change. In this case, the arrangement of the link IDs of the link k through which the vehicle position indicated by the vehicle position information P passes during the course changing operation is “ID = XXXXX → ID = ΔΔΔΔ”. The intersection specifying information Be is assumed to be “ID = ◯◯◯◯ → ID = ΔΔΔΔ”.

9. Road Information Acquisition Unit The road information acquisition unit 23 extracts and acquires road information Ra around the own vehicle C from the map database DB1 based on the own vehicle position information P acquired by the own vehicle position information acquisition unit 16. It functions as a road information acquisition means. That is, the road information acquisition unit 23 is required in the own vehicle position information acquisition unit 16, the course change information generation unit 18 of the course change behavior detection unit 17, the host vehicle position information correction unit 51, the navigation calculation unit 27, and the like. The road information Ra around the vehicle C in the range is extracted from the map database DB1 and output to the respective parts.

10. Image Recognizing Unit The image recognizing unit 24 functions as an image recognizing unit that performs an image recognizing process on a target feature included in the image information G acquired by the image information acquiring unit 12. In the present embodiment, the image recognition unit 24 performs image recognition processing using a road marking provided on the road surface as a target feature. Specifically, the image recognition unit 24 performs binarization processing, edge detection processing, and the like on the image information G when performing image recognition of the target feature, and the target feature included in the image information G The contour information of (road marking) is extracted. After that, the image recognition unit 24 performs pattern matching between the extracted contour information of the feature and the feature quantities of various feature forms that can be the target feature, so that the target feature included in the image information G is detected. Extract images. In addition, the image recognition unit 24 recognizes the feature type of the feature whose feature information matches the contour information of the extracted feature as the feature type of the target feature included in the image information G. Then, the image recognition unit 24 determines that the image recognition of the target feature has succeeded when the pattern matching is successful. On the other hand, as a result of performing the image recognition process on the image information G, the image recognition unit 24 determines that the image recognition of the target feature has failed when the pattern matching fails.

  In the present embodiment, the image recognition unit 24 includes a feature attribute information generation unit 25. The feature attribute information generation unit 25 functions as a feature attribute information generation unit that generates, for the target feature recognized by the image recognition unit 24, feature attribute information representing the attribute of the target feature. This feature attribute information constitutes a part of the recognized feature information A and the learned feature information Fb as described later. Here, the attribute of the target feature represented by the feature attribute information may be anything that can identify the one target feature from other target features. Therefore, for example, the feature attribute information includes the feature type of the target feature, the specific shape and size of the target feature, the link ID of the link k where the target feature exists, and the target feature exists. It is configured to have one or more pieces of information selected from the approximate positions. Information constituting such feature attribute information is, for example, the image recognition result of the target feature by the image recognition unit 24, or the vehicle position information P at the time of acquiring the image information G in which the target feature is recognized. Based on.

11. Navigation Calculation Unit The navigation calculation unit 27 operates in accordance with the application program 28 to execute navigation functions such as display of the vehicle position, route search from the departure place to the destination, route guidance to the destination, and destination search. Arithmetic processing means. Here, the application program 28 refers to the vehicle position information P, the map information M, the course change learning information S, the feature information F, and the like, and causes the navigation calculation unit 27 to execute various navigation functions. For example, the navigation calculation unit 27 acquires map information M around the host vehicle C from the map database DB1 based on the host vehicle position information P, displays a map image on the display screen of the display input device 29, and Based on the vehicle position information P, the vehicle position mark is superimposed and displayed on the map image. Further, the navigation calculation unit 27 performs a route search from a predetermined departure place to a destination based on the map information M stored in the map database DB1. Furthermore, the navigation calculation unit 27 uses one or both of the display input device 29 and the audio output device 30 based on the searched route from the departure point to the destination and the vehicle position information P, and uses the driver or the driver. Route guidance for. The display input device 29 is a combination of a display device such as a liquid crystal display device and an input device such as a touch panel and operation switches. The audio output device 30 includes a speaker and the like. In the present embodiment, the navigation calculation unit 27, the display input device 29, and the voice output device 30 function as the guidance information output means 31 in the present invention.

12 Recognition position information acquisition unit The recognition position information acquisition unit 41 functions as a recognition position information acquisition unit that acquires recognition position information representing the recognition position of the target feature for the target feature that has been successfully recognized by the image recognition unit 24. To do. In the present embodiment, the recognition position information acquisition unit 41 first monitors whether or not the image recognition of the target feature has succeeded in the image recognition processing by the image recognition unit 24. When the image recognition unit 24 succeeds in image recognition of the target feature, the recognition position information acquisition unit 41 acquires the image recognition result and the vehicle position information acquired by the vehicle position information acquisition unit 16. Based on P, the recognition position of the target feature is derived. Here, the recognition position information acquisition part 41 acquires the own vehicle position information P when acquiring the image information G including the image of the target feature that has been successfully recognized as the recognition position information of the target feature. Since the recognition position information of the target feature acquired in this way is derived based on the own vehicle position information P, it becomes position information reflecting an error of the own vehicle position information P.

13. Recognized feature information generating unit The recognized feature information generating unit 42 is the feature attribute information of the target feature generated by the feature attribute information generating unit 25 for the target feature that has been successfully recognized by the image recognizing unit 24. And a recognized feature information generating unit that generates recognized feature information A including the recognized position information of the target feature acquired by the recognized position information acquiring unit 41. Then, the recognized feature information generation unit 42 stores the generated recognized feature information A in the feature learning database DB3. Hereinafter, details of the processing performed by the recognized feature information generation unit 42 will be described with reference to FIGS. 5 and 6. FIG. 5 is an explanatory diagram for explaining an outline of the learning process of the feature information F based on the image recognition result of the target feature. FIG. 5A is an example of a road marking (target feature) provided on an actual road on which the host vehicle C travels. In this example, the image recognizing unit 24 recognizes the image of the character “30” of the speed indicator as the target feature f1. FIG. 5B is an example of the recognized feature information A stored in the feature learning database DB3. FIG. 5C is an example of the feature database DB2 in which the learning result stored in the feature learning database DB3 is reflected.

  In the present embodiment, the recognized feature information generation unit 42 recognizes the recognized feature information A for each target feature, as shown in FIG. Is generated as a learning value for a predetermined position range to which the recognition position of the target feature indicated by the recognition position information belongs. The recognized feature information generation unit 42 adds and stores the learning value for each position range every time the target feature is recognized. In this example, the predetermined position range is a range that is divided and set at a certain distance in the direction along the link k representing the road, for example, every 0.5 [m] in the direction along the link k. The range is divided. The learning value is a value added to the position range to which the recognition position of the target feature belongs in the feature learning database DB3 every time image recognition of one target feature is successful. Each time a feature image is successfully recognized, one point is given. That is, in this example, the recognized feature information A includes information indicating a position range including the recognized position of the target feature and information of the learning value “1” as the recognized position information of the target feature. Configured.

  FIG. 6 is an enlarged view of a portion related to the target feature f1 of the learning value stored in the feature learning database DB3 shown in FIG. For example, in the example of FIG. 5A, when the image recognition of the target feature f1 is successful, the recognition position of the target feature f1 acquired by the recognition position information acquisition unit 41 is “ In the case of the position range indicated as “a4”, 1 is added to the learning value of the position range a4 as indicated by a broken line in FIG. And if the same target feature f1 is image-recognized a plurality of times when the own vehicle C passes the same road a plurality of times, the feature learning database DB3 stores the feature target database as shown in FIG. 5B and FIG. The learning values as the plurality of recognition feature information A generated each time the target feature is recognized are accumulated and accumulated for each position range representing the recognition position of the target feature. Then, as will be described later, when the learning value becomes equal to or greater than a predetermined learning threshold value T1, learning feature information Fb for the target feature is generated by the learning feature information generation unit 44 and stored in the feature database DB2. Is done. In the example of FIG. 5, as shown in FIG. 5C, learning feature information Fb1 corresponding to the target feature f1 is stored in the feature database DB2.

  In addition, the recognized feature information generation unit 42 is generated by the feature attribute information generation unit 25 so that the target feature indicated by the recognized feature information A can be distinguished from other target features. It has the feature attribute information of the target feature. That is, the recognized feature information A stored in the feature learning database DB3 includes information indicating the position range as the recognition position information of the target feature and information of the learning value “1”, and the target feature. The feature attribute information representing the feature attributes of the two is associated with each other. As described above, for example, the feature attribute information includes the feature type of the target feature, the specific shape and size of the target feature, and the link k where the target feature exists. It is configured with one or more information selected from a link ID, an approximate position where the target feature exists, and the like.

14 Estimated position determination unit The estimated position determination unit 43 acquires estimated position information obtained by statistically processing a plurality of the recognized position information for each target feature stored in the feature learning database DB3. It functions as an acquisition means. Therefore, the estimated position determination unit 43 is based on a plurality of recognized feature information A for the same target feature stored in the feature learning database DB3 by image recognition of the same target feature multiple times. 5, the estimated recognition position pa for the target feature is determined, and the estimated position pg of the target feature is determined by converting the estimated recognition position pa into a position on the road of the target feature. Perform the process. In the present embodiment, the estimated position determination unit 43 first determines a representative value of the distribution based on the distribution of the plurality of recognized feature information A for the same target feature, and the estimated recognition position pa for the target feature. Judge as. Here, the mode value is used as the representative value of the distribution. That is, the estimated position determination unit 43 determines the position representing the position range in which the learning value as the recognized feature information A for each target feature first becomes equal to or greater than the predetermined learning threshold T1. The estimated recognition position pa for the object is determined. As an example, a determination method in the case of determining the estimated recognition position pa of the target feature f1 in the example of FIG. 5 will be described. As shown in FIG. 6, the learning value as the recognized feature information A for the target feature f1 is initially equal to or greater than the learning threshold value T1 in the position range a4. Therefore, the estimated position determination unit 43 determines a position representing the position range a4, for example, the center position pa4 of the position range a4, as the estimated recognition position pa of the target feature f1.

  Next, the estimated position determination unit 43 converts the estimated recognition position pa of the target feature determined as described above into a position on the road of the target feature, and calculates the estimated position pg of the target feature. judge. Such conversion can be performed based on the positional relationship between the subject vehicle C and the target feature in the image information G, which is theoretically determined from the mounting position, mounting angle, and angle of view of the imaging device 11. Information representing the estimated position pg of the target feature obtained by the estimated position determination unit 43 in this way is acquired as estimated position information of the target feature.

15. Learning feature information generation unit The learning feature information generation unit 44 recognizes a plurality of pieces of recognized feature information A about the same target feature stored in the feature learning database DB3 by image recognition of the same target feature multiple times. The learning feature information generating means for generating the learning feature information Fb representing the learning result of the target feature. Here, the learned feature information Fb includes feature attribute information about the same target feature indicated in the plurality of recognized feature information A, and a plurality of recognized position information A about the target feature by the estimated position determination unit 43. And estimated position information that represents the estimated position pg of the target feature. That is, the learning feature information generation unit 44 includes the estimated position information representing the estimated position pg of each target feature acquired by the estimated position determination unit 43 and the feature included in the recognized feature information A about the target feature. Learning feature information Fb is generated in association with the object attribute information. At this time, the learned feature information generation unit 44 assigns identification information (feature ID) for distinguishing each feature from other features as the feature attribute information of each learned feature information Fb. Thereby, the learning feature information Fb is generated as information including the position information and the feature attribute information associated therewith like the initial feature information Fa. The learned feature information Fb generated by the learned feature information generation unit 44 is stored in the feature database DB2. In the present embodiment, as shown in FIG. 5C, learning feature information Fb1 is generated by the learning feature information generation unit 44 and stored in the feature database DB2. The black square “■” shown in this figure represents the estimated position pg of the target feature f1 indicated by the position information of the learned feature information Fb1.

16. Relationship Information Generation Unit When the route change operation of the host vehicle C is detected by the route change detection unit 17, the relationship information generation unit 45 detects the detected route change operation of the host vehicle C and the route change operation. It functions as a relationship information acquisition unit that acquires relationship information Br (see FIG. 7) that represents a relationship with the target feature recognized by the image recognition unit 24 earlier. Here, the relationship information generation unit 45 acquires distance information 15 as a movement distance detection unit that detects the movement distance of the own vehicle C, and own vehicle position information that acquires own vehicle position information P that represents the current position of the own vehicle C. The relationship information Br is acquired based on information from at least one of the acquisition units 16. Therefore, in this embodiment, the relationship information generation part 45 is comprised including the distance information generation part 46 and the feature specific information generation part 47 as shown in FIG. Therefore, the details of the processing performed by the distance information generation unit 46 and the feature identification information generation unit 47 will be described below.

  In the following description, FIG. 7 showing the contents of the detection operation information B and the route change learning information S regarding a specific example of the route change operation of the host vehicle C will be referred to as appropriate. Here, Fig.7 (a) has shown the specific example of the condition of the road where the own vehicle C actually drive | works. In the example of FIG. 7 (a), the host vehicle C passes over the target feature f1 indicated by the character “30” in the speed indicator, and after making a distance of 100 m, turns left at the intersection N. Yes. Accordingly, in this example, after the image recognition unit 24 recognizes the target feature f1, the route change detection unit 17 detects the route change operation of the “left turn” of the host vehicle C at the intersection N that has traveled a distance of 100 [m]. Will do. FIG. 7B shows an example of the detection operation information B for the route change operation, and FIG. 7C shows an example of the route change learning information S for the route change operation.

16-1. Distance Information Generating Unit The distance information generating unit 46 detects the target feature recognized by the image recognizing unit 24 before the course change detecting unit 17 detects the course changing operation of the host vehicle C from the recognition position of the target feature. It functions as distance information acquisition means for acquiring distance information Bc (see FIG. 7B) that represents the distance to the position where the course of the host vehicle C has been changed. That is, as shown in the example of FIG. 7A, the distance information generation unit 46 performs the course change operation of the host vehicle C by the course change detection unit 17 after the target feature f1 is recognized by the image recognition unit 24. When detected, the distance L between the feature and the course change, which is the distance from the recognition position of the target feature to the detection position of the course change operation, is detected to generate distance information Bc. Therefore, in the present embodiment, the distance information generation unit 46 detects the feature-course change distance L using information on the movement distance of the host vehicle C detected by the distance sensor 15.

  That is, the distance information generation unit 46 starts from the position of the host vehicle C when the target feature is recognized by the image recognition unit 24, and until the course change operation of the host vehicle C is detected by the course change detection unit 17. The moving distance of the host vehicle C is detected as the feature-path change distance L. In this way, by using the output from the distance sensor 15, the feature-path change distance L can be detected regardless of the vehicle position information P. Then, the distance information generation unit 46 sets the information indicating the detected feature-path change distance L as distance information Bc. In the example shown in FIG. 7A, the actual feature-route change distance L is 100 [m]. Therefore, as a detection result of the distance L between the feature and the course change by the distance sensor 15, as shown in FIG. 7B, a value including the detection error of the distance sensor 15 is converted into a distance by the distance information generation unit 46. Generated as information Bc.

16-2. The feature specifying information generating unit The feature specifying information generating unit 47 functions as a feature specifying information generating unit that generates the feature specifying information Bb for specifying the target feature recognized by the image recognition unit 24. That is, as shown in the example of FIG. 7A, the feature specifying information generation unit 47 changes the course of the host vehicle C by the course change detection unit 17 after the target feature f1 is recognized by the image recognition unit 24. When the motion is detected, the feature specifying information Bb for specifying the recognized target feature f1 is generated. In the present embodiment, the feature specifying information Bb is identification information for specifying the recognized feature information A stored in the feature learning database DB3 (the feature ID (ID = × in the example of FIG. 7B)). Xxx)). That is, the feature specifying information generating unit 47 recognizes the recognized feature information A about the recognized target feature f1 by the recognized feature information generating unit 42 after the image recognition unit 24 recognizes the target feature f1. When it is generated and stored in the feature learning database DB3, identification information for specifying the recognized feature information A is acquired. Then, the feature specifying information generation unit 47 generates the identification information as the feature specifying information Bb. Accordingly, the detection operation information is obtained by referring to the recognized feature information A stored in the feature learning database DB3 on the basis of the feature specifying information Bb included in the detection operation information B or the course change learning information S described later. It becomes possible to specify the feature attribute and recognition position of the target feature indicated in B or the course change learning information S.

  The identification information for specifying the recognized feature information A stored in the feature learning database DB3 is not limited to the feature ID given to each recognized feature information A. For example, the feature learning Other information such as the storage location of each recognized feature information A in the database DB3 may be used.

  7B, in the present embodiment, the distance information Bc generated by the distance information generating unit 46 and the feature specifying information Bb generated by the feature specifying information generating unit 47 are used. , Relationship information Br is configured.

17. Detection Operation Information Generation Unit The detection operation information generation unit 48 includes route change information Ba that can identify the route change intersection N where the route change operation has been performed, and distance information acquired by the distance information generation unit 46 regarding the route change operation. It functions as detection course change operation generation means for generating detection operation information B including Bc. In the present embodiment, the detection operation information B further includes feature specifying information Bb. As described above, the route change information Ba is generated by the route change information generation unit 18 when the route change detection unit 17 detects the route change operation of the host vehicle C. Further, the distance information Bc and the feature specifying information Bb constituting the relationship information Br are generated by the distance information generating unit 46 and the feature specifying information generating unit 47 of the relationship information generating unit 45. Therefore, the detection operation information generation unit 48 generates the detection operation information B by associating the route change information Ba and the relationship information Br including the distance information Bc. Then, the detection operation information generation unit 48 stores the generated detection operation information B in the course change learning database DB4.

  Here, in the plurality of detection operation information B for a certain course change operation, the course change mode of the course change operation, the course change intersection N, and the target feature that is image-recognized before the detection of the course change operation is performed. Are the same. Therefore, in the present embodiment, the detection operation information generation unit 48 uses the route change information Ba and the feature specifying information Bb as a set of information about the plurality of detection operation information B for the same route change operation. It memorize | stores in change learning database DB4. FIG. 7B shows an example in which a plurality of detection operation information B for one course changing operation (“left turn” at the intersection N) shown in the example of FIG. ing. In this example, the detection operation information B includes a course changing operation in which the mode information Bd is “left turn” and the intersection identification information Be is “ID = XXXXXX → ID = ΔΔΔΔ”. It is composed of a plurality of distance information Bc associated with one course change information Ba and feature specifying information Bb containing identification information of “ID = xxxx”. And since these several distance information Bc includes the detection error by the distance sensor 15, although it is a value close | similar to 100 [m] which is the actual feature-distance change distance L, each becomes a little different value. Yes.

18. Average distance determination unit The average distance determination unit 49 determines an average value of a plurality of distance information Bc for the same course changing operation indicated in the plurality of detection operation information B for the same course changing operation, and generates average distance information Sc. To do. In the present embodiment, the average distance determination unit 49 includes a plurality of pieces of distance information of the detection operation information B that are stored together as a common route change information Ba and feature specifying information Bb for a certain route change operation. An average value is determined for Bc. And the average value of several distance information Bc about the determined same course change operation | movement is produced | generated as average distance information Sc. In the example of the course change learning information S shown in FIG. 7C, the average distance information Sc is the detection operation information B for one course change action (“left turn”) shown in the example of FIG. It is an average value (“100.1 [m]”) of all values of the plurality of distance information Bc. The statistical relationship information in the present invention is obtained from the average distance information Sc generated by the average distance determination unit 49 and the feature specifying information Bb common to the plurality of detection operation information B for the same course changing operation. Sr is configured. Therefore, in the present embodiment, the average distance determination unit 49 uses the statistical relationship information Sr obtained by statistically processing the plurality of relationship information Br for the same course changing operation indicated by the plurality of detection operation information B. It functions as a statistical relationship information generation means for generating.

19. Course Change Learning Information Generation Unit Course change learning information generation unit 50 is based on the detected action information B stored in the course change learning database DB4, and the past course change operation of the host vehicle C associated with the target feature. It functions as a course change learning information generating means for generating course change learning information S representing the learning result. In the present embodiment, the course change learning information generation unit 50 basically detects a plurality of detections for the same course change operation stored in the course change learning database DB4 by detecting the same course change operation a plurality of times. Based on the motion information B, the course change learning information S is generated. The course change learning information S includes the course change information Ba that can identify the course change intersection N where the course change operation of the host vehicle C has been performed, and the target features that are image-recognized before the course change operation. And statistical relationship information Sr including average distance information Sc as distance information from the recognized position of the target feature to the position where the course change operation of the host vehicle C is performed. Therefore, this course change learning information S includes common course change information Ba and feature specifying information Bb for a plurality of detection operation information B for a certain course change operation, and a plurality of detection operation information for the course change operation. B is configured as information associated with average distance information Sc that is an average value of a plurality of distance information Bc for the same course changing operation shown in B. In the example shown in FIG. 7C, the course change learning information S includes the mode information Bd of “left turn” and the intersection specification information Be of “ID = ◯◯◯◯ → ID = ΔΔΔΔ”. One course change information Ba containing a certain course change operation, feature specifying information Bb containing identification information “ID = xxx”, and “100.1 [m]” associated therewith ”And the average distance information Sc. The course change learning information generation unit 50 stores the generated course change learning information S in the course change database DB5.

  When there is only one detection operation information B in the route change learning database DB4 for a certain route change operation, the route change learning information generation unit 50 determines the route change learning information based on the one detection operation information B. S is generated. In this case, since the average distance information Sc included in the statistical relationship information Sr of the course change learning information S is the same as the distance information Bc of the detection operation information B, the content of the course change learning information S is substantially detected. The same as the operation information B. Below, the case where the course change learning information S is generated based on a plurality of detection operation information B will be described as an example.

20. Self-vehicle position information correction unit The self-vehicle position recognition device 2 included in the navigation device 1 uses the course change learning information S generated as described above to cause the self-vehicle position information correction unit 51 to The vehicle position information P acquired by the information acquisition unit 16 is corrected. That is, when the target feature is recognized by the image recognizing unit 24, the own vehicle position information correcting unit 51 is based on the route change learning information S about the target feature stored in the route change database DB5. It functions as own vehicle position correcting means for correcting the vehicle position information P. Therefore, in the present embodiment, the own vehicle position information correction unit 51 extracts the course change learning information S about the target feature recognized by the image recognition unit 24 from the course change database DB5, A first correction processing unit 53 that performs processing for correcting the vehicle position information P based on the extracted course change learning information S and road information Ra. Further, in the present embodiment, the vehicle position information correction unit 51 uses the image recognition result of the target feature when the course change learning information S about the target feature recognized by the image recognition unit 24 does not exist. A second correction processing unit 54 that performs processing for correcting the vehicle position information P based on the vehicle is also provided. Below, the structure of these each part of the own vehicle position information correction | amendment part 51 is each demonstrated.

20-1. Learning Information Extraction Unit The learning information extraction unit 52 extracts the course change learning information S about the recognized target feature from the course change database DB5 when the target feature is image-recognized by the image recognition unit 24. It functions as a route change learning information extraction means to obtain. Therefore, the learning information extraction unit 52 monitors whether or not the image recognition unit 24 succeeds in image recognition of the target feature while the host vehicle C is traveling. Then, when the image recognition of the target feature is successful, the route change database DB5 is searched, and the route change learning information S about the target feature is extracted. The search for the course change learning information S for the target feature in the course change database DB5 is performed based on the feature specifying information Bb included in each course change learning information S. That is, if the image recognition unit 24 succeeds in image recognition of the target feature, the feature attribute information generation unit 25 generates the feature attribute information. Therefore, the learning information extraction unit 52 extracts the feature information F having the same feature attribute from the feature database DB2 based on the generated feature attribute information. Then, the learning information extraction unit 52 searches the course change database DB5 for course change learning information S having the feature identification information Bb of the identification information that matches the identification information (feature ID) of the feature information F. To extract. The course change learning information S extracted by the learning information extraction unit 52 is used for correcting the vehicle position information P in the first correction processing unit 53.

20-2. First Correction Processing Unit The first correction processing unit 53 is a correction processing unit that performs processing for correcting the vehicle position information P based on the course change learning information S and the road information Ra extracted by the learning information extraction unit 52. Function. That is, the first correction processing unit 53 recognizes the target feature based on the course change information Ba and the average distance information Sc included in the course change learning information S about the target feature recognized by the image recognition unit 24. In the road information Ra of the route change intersection N specified by the route change information Ba, the position that has traveled by the distance indicated by the average distance information Sc from the recognized position of the target feature based on the own vehicle position information P at the time The vehicle position information P is corrected so as to match the position of the vehicle. A specific example of the correction processing of the vehicle position information P by the first correction processing unit 53 will be described below with reference to FIGS. 8 and 9. 8 shows the trajectory of the host vehicle C and the host vehicle position information P when the host vehicle position information P is not corrected by the first correction processor 53. FIG. The trajectory of the own vehicle C and the own vehicle position information P when the own vehicle position information P is corrected is shown.

  As shown in FIG. 8, when the vehicle position information P is not corrected by the first correction processing unit 53, the vehicle position information P in the middle of the link k due to an error in the traveling direction of the vehicle position information P. There is a case where a position jump to a different link k occurs due to an attempt to change the travel direction. For example, as shown in FIG. 8A, the course change operation of the host vehicle C is performed in a state where the host vehicle position information P is recognized to be ahead of the actual position of the host vehicle C due to an error. In this case, after the vehicle position indicated in the vehicle position information P has traveled beyond the node n (ID = ○ × ○ ×) of the intersection N to the link k in the straight direction ID = □□□□ In the state, the route change operation of the left turn by the actual own vehicle C is performed. Therefore, a position jump of the own vehicle position information P occurs in the middle of the link k of ID = □□□□, and the own vehicle position indicated by the own vehicle position information P is a link of ID = Δ △△△ after the course is changed. Move to k. On the other hand, when the course change operation of the host vehicle C is not performed, the host vehicle position information P ′ is more than the actual position of the host vehicle C ′ due to an error, as indicated by a two-dot chain line in FIG. The state recognized as proceeding ahead is not corrected, and the vehicle position information P further travels with an error included.

  Further, for example, as shown in FIG. 8B, the course change operation of the host vehicle C is performed in a state where the host vehicle position information P is recognized as being delayed behind the actual position of the host vehicle C due to an error. If it is performed, the vehicle position indicated in the vehicle position information P is in progress of the link k with ID = XXX, and before reaching the node n (ID = XXXXX) at the intersection N In this state, the route change operation of the left turn by the actual own vehicle C is performed. Therefore, the position jump of the own vehicle position information P occurs in the middle of the link k of ID = ○○○○, and the own vehicle position indicated by the own vehicle position information P is the link of ID = △△△△ after the course is changed. Move to k. On the other hand, when the course change operation of the host vehicle C is not performed, the host vehicle position information P ′ is more erroneous than the actual position of the host vehicle C ′ due to an error, as indicated by a two-dot chain line in FIG. The state recognized as being delayed later is not corrected, and the vehicle position information P further travels with an error included.

  In order to suppress the position jump of the own vehicle position information P as shown in the example of FIG. 8 and to correct the own vehicle position information P even when the course changing operation at the intersection N is not performed. The first correction processing unit 53 corrects the vehicle position information P as follows. Here, the correction processing of the own vehicle position information P by the first correction processing unit 53 is advanced ahead of the actual position of the own vehicle C due to an error as shown in FIG. The vehicle position information P is divided into a case where the vehicle position information P is recognized as being delayed later than the actual position of the vehicle C, as shown in FIG. Each will be explained. In any case, in the correction in any case, the position changed by the distance indicated by the average distance information Sc from the recognized position of the target feature based on the own vehicle position information P when the target feature is recognized is the course change information. The own vehicle position information P is corrected so as to coincide with the position in the road information Ra of the route change intersection N specified by Ba.

  As shown in FIG. 9A, when the vehicle position information P is recognized as being advanced ahead of the actual position of the host vehicle C due to an error, the host vehicle C becomes a course change intersection N. Before the vehicle arrives, in other words, before the vehicle C actually moves the distance from the image recognition of the target feature f1 to the route change intersection N, the vehicle shown in the vehicle position information P The position reaches the position of the node n (ID = OxOx), which is the location in the road information Ra of the route change intersection N. Here, the distance from the time when the host vehicle C recognizes the target feature f1 until it reaches the route change intersection N is the target location related to the previous route change operation of the host vehicle C at the same route change intersection N. It is equal to the feature-route change distance L, which is the distance from the recognition position of the object f1 to the position where the course change of the host vehicle C is performed. As described above, the average distance information Sc included in the course change learning information S almost accurately represents the actual feature-path change distance L. Therefore, in this case, the first correction processing unit 53 targets the host vehicle C after the host vehicle position indicated by the host vehicle position information P reaches the course change intersection N specified by the course change information Ba. The host vehicle position indicated by the host vehicle position information P is stopped on the course change intersection N until the distance indicated by the average distance information Sc actually travels from the recognized position of the feature f1. At this time, information on the actual travel distance of the host vehicle C can be acquired by the distance sensor 15. Thereby, since the actual position of the own vehicle C catches up and coincides with the own vehicle position indicated in the own vehicle position information P, the own vehicle position information P can be corrected. As a result, the host vehicle position information acquisition unit 16 acquires the corrected host vehicle position information P with high accuracy.

  In the example shown in FIG. 9A, it is recognized that the vehicle position information P is 40 m ahead of the actual position of the vehicle C due to an error, and the vehicle C is the target feature f1. The distance from the image recognition to the route change intersection N is 100 [m], whereas the road from the recognition position of the target feature f1 based on the vehicle position information P to the route change intersection N The distance to the position of the node n (ID = OxOx) which is the position in the information Ra is 60 [m]. Therefore, the first correction processing unit 53 determines that the vehicle position indicated by the vehicle position information P is 60 [m] and the node n (ID = XXXXX) in the road information Ra corresponding to the route change intersection N. X) After reaching the top, the vehicle position indicated in the vehicle position information P is stopped on the node n (ID = OxOx). At this time, the stop period of the host vehicle position indicated by the host vehicle position information P is the distance 100.1 [m] indicated by the average distance information Sc from the recognition position of the target feature f1 of the host vehicle C (FIG. )) Is the period until the actual progress. Thereby, the actual position of the own vehicle C is made to follow the own vehicle position indicated by the own vehicle position information P so that the positions of both coincide with each other, and the correction of the own vehicle position information P is completed. After such correction of the vehicle position information P, the acquisition of the vehicle position information P is performed as usual. Thereby, the own vehicle position indicated in the corrected own vehicle position information P moves according to the movement of the actual position of the own vehicle C. The correction of the host vehicle position information P by the first correction processing unit 53 can be performed regardless of whether the host vehicle C has changed its course or traveled straight at the course change intersection N. Accordingly, as shown by the two-dot chain line own vehicle C ′ and own vehicle position information P ′ in FIG. 9A, the course change was also made when the course change operation of the own vehicle C was not performed. As in the case, the vehicle position information P is corrected.

  Further, as shown in FIG. 9B, when the vehicle position information P is recognized as being delayed behind the actual position of the vehicle C due to an error, the vehicle position information P is indicated. Before the vehicle position reaches the position of the node n (ID = OxOx), which is the position in the road information Ra of the route change intersection N, the host vehicle C reaches the route change intersection N, in other words. The vehicle C actually moves the distance from the image recognition of the target feature f1 to the route change intersection N. As described above, the distance from when the host vehicle C recognizes the target feature f1 to the route change intersection N is the feature related to the past route change operation of the host vehicle C at the same route change intersection N. The average distance information Sc which is equal to the course change distance L and is included in the course change learning information S almost accurately represents the actual feature-course change distance L. Therefore, in this case, the first correction processing unit 53 indicates the own vehicle position information P when the own vehicle C actually travels the distance indicated by the average distance information Sc from the recognition position of the target feature f1. The vehicle position to be moved is instantaneously moved (jumped) on the route change intersection N specified by the route change information Ba. At this time, information on the actual travel distance of the host vehicle C can be acquired by the distance sensor 15. As a result, the own vehicle position indicated by the own vehicle position information P catches up with the actual position of the own vehicle C, so that the own vehicle position information P can be corrected. As a result, the host vehicle position information acquisition unit 16 acquires the corrected host vehicle position information P with high accuracy.

  In the example shown in FIG. 9B, it is recognized that the own vehicle position information P is delayed 40 [m] from the actual position of the own vehicle C due to an error, and the own vehicle C detects the target feature f1. While the distance from the image recognition to the route change intersection N is 100 [m], the road information of the route change intersection N from the recognized position of the target feature f1 based on the own vehicle position information P. The distance to the position of node n (ID = OxOx), which is the position in Ra, is 140 [m]. Therefore, the first correction processing unit 53 actually travels the distance 100.1 [m] (see FIG. 7C) indicated by the average distance information Sc from the recognition position of the target feature f1. Next, the vehicle position indicated by the vehicle position information P is moved onto the node n (ID = OxOx) in the road information Ra corresponding to the route change intersection N. As a result, the vehicle position indicated in the vehicle position information P is made to follow the actual position of the vehicle C, the positions of the two vehicles are matched, and the correction of the vehicle position information P is completed. After such correction of the vehicle position information P, the acquisition of the vehicle position information P is performed as usual. Thereby, the own vehicle position indicated in the corrected own vehicle position information P moves according to the movement of the actual position of the own vehicle C. The correction of the host vehicle position information P by the first correction processing unit 53 can be performed regardless of whether the host vehicle C has changed its course or traveled straight at the course change intersection N. Accordingly, as shown by the two-dot chain line of the own vehicle C ′ and the own vehicle position information P ′ in FIG. 9B, the course is changed even when the course change operation of the own vehicle C is not performed. As in the case, the vehicle position information P is corrected.

20-3. Second correction processing unit The second correction processing unit 54 is based on the image recognition result of the target feature when the course change learning information S about the target feature recognized by the image recognition unit 24 does not exist. It functions as a correction processing means for performing processing for correcting the position information P. That is, the second correction processing unit 54 is based on the result of the image recognition processing of the target feature by the image recognition unit 24 and the feature information F stored in the feature database DB2 for the target feature. The position information P is corrected. Specifically, the second correction processing unit 54 first determines the target feature based on the result of the image recognition processing by the image recognition unit 24 and the mounting position, mounting angle, and angle of view of the imaging device 11. The positional relationship between the host vehicle C and the target feature at the time of acquiring the image information G including the image is calculated. Next, the second correction processing unit 54 extracts the feature information F about the target feature recognized by the image recognition unit 24 from the feature database DB2. Then, the second correction processing unit 54 is based on the calculation result of the positional relationship between the host vehicle C and the target feature and the position information of the target feature included in the feature information F about the target feature. Thus, the position information of the target vehicle C in the traveling direction of the host vehicle C (feature information F) is calculated and acquired with high accuracy. And the 2nd correction process part 54 correct | amends the own vehicle position information P acquired by the own vehicle position information acquisition part 16 based on the highly accurate position information of the own vehicle C acquired in this way. As a result, the host vehicle position information acquisition unit 16 acquires the corrected host vehicle position information P with high accuracy.

21. Course Change Operation Learning Process Next, a course change operation learning process procedure of the host vehicle C executed by the host vehicle position recognition device 2 included in the navigation device 1 according to the present embodiment (a course change operation learning program). Will be described. FIG. 10 is a flowchart showing the overall procedure of the course change operation learning process according to the present embodiment. FIG. 11 is a flowchart showing the procedure of the feature learning process included as part of the course change operation learning process shown in FIG. 10 in the present embodiment. The processing procedure described below is executed by hardware and / or software (program) or both of the above-described functional units. When each of the functional units is configured by a program, the arithmetic processing device included in the navigation device 1 operates as a computer that executes a course change operation learning program that configures the functional units. Below, first, the whole procedure of the learning process of course change operation | movement is demonstrated.

21-1. Overall Procedure of Course Change Operation Learning Process As shown in FIG. 10, in the vehicle position recognition device 2, first, the vehicle position information acquisition unit 16 acquires the vehicle position information P (step # 01). Next, the image information acquisition unit 12 acquires image information G around the host vehicle C imaged by the imaging device 11 (step # 02). Thereafter, the image recognition unit 24 performs image recognition processing of the target feature included in the image information G (step # 03). If the target feature is not recognized in step # 03 (step # 04: No), the process returns to step # 01, and the vehicle position information P and the image information G are acquired again. When the target feature included in the image information G is recognized in step # 03 (step # 04: Yes), the feature learning process is executed (step # 05). The procedure of the feature learning process will be described later in detail based on the flowchart shown in FIG.

  In parallel with the feature learning process in step # 05, the distance information generation unit 46 of the relation information generation unit 45 starts measuring the distance L between the feature and the course change (step # 06). Here, the distance L between the feature and the course change is, as described above, the detection of the course change operation of the host vehicle C detected in step # 07 described later from the recognition position of the target feature recognized in step # 03. The distance to the position. In step # 06, the distance measurement is started from the position of the host vehicle C when the target feature is recognized in step # 03. Next, the course change detection unit 17 performs a course change operation detection process for the host vehicle C (step # 07). In the course change operation detection process of step # 07, the course change detection unit 17 stands by in a state where the course change operation of the host vehicle C can be detected. Then, when the route change operation of the host vehicle C is detected (step # 08: No), when the feature-route change distance L being measured is equal to or greater than a predetermined stop threshold value (step #). 16: Yes), the measurement of the distance L between the feature and the course change is stopped (step # 17), and the process ends. On the other hand, when the course change operation of the host vehicle C is detected in a state where the distance L between the feature being measured and the course change is less than the predetermined stop threshold (step # 16: No) (step # 08: Yes), the measurement of the distance L between the feature and the course change by the distance information generating unit 46 is finished (step # 09). Thereafter, relation information Br including distance information Bc representing the measured feature-distance change distance L and feature specifying information Bb generated by the feature specifying information generating unit 47 is acquired (step #). 10).

  Next, the route change information Ba that can identify the route change intersection N where the route change operation detected in step # 07 is performed is acquired by the detection operation information generation unit 48 and the route change operation is acquired in step # 10. The detection operation information B including the related information Br (including the distance information Bc) is generated (step # 11). Then, the generated detection operation information B is stored in the course change learning database DB4 (step # 12). Next, average distance information Sc is generated by the average distance determination unit 49 based on a plurality of detection operation information B for the same course change operation stored in the course change learning database DB4 by the above processing (step # 13). ). Here, as described above, the average distance information Sc is information on the average value of the plurality of distance information Bc for the same course changing operation shown in the plurality of detection operation information B for the same course changing operation. Then, based on a plurality of detection operation information B about the same route change operation stored in the route change learning database DB4 by detecting the same route change operation a plurality of times, the route change learning information generation unit 50 causes the target location The course change learning information S representing the learning result of the past course change operation of the host vehicle C associated with the object is generated (step # 14). Then, the generated course change learning information S is stored in the course change database DB5 (step # 15). Thus, the entire procedure of the course change operation learning process in the vehicle position recognition device 2 is completed.

21-2. Next, the procedure of the feature learning process included as a part of the learning process of the course changing operation according to the present embodiment will be described. In the feature learning process in step # 05 of FIG. 10, as shown in FIG. 11, first, the recognized position information of the target feature recognized in step # 03 of FIG. Obtained based on the information P (step # 21). Next, the recognized feature information A having the learning value information is generated by the recognized feature information generation unit 42 (step # 22). That is, as described above, the recognized feature information A includes the feature attribute information of the target feature generated by the feature attribute information generating unit 25 when the target feature is recognized, and the target acquired in step # 21. The learning value information for each position range based on the recognized position information of the feature is associated with each other. Then, for example, as shown in FIG. 5B, the recognized feature information A having such learning value information is stored in the feature learning database DB3 (step # 23). Here, when the learning value of the recognized feature information A for the target feature stored in the feature learning database DB3 is less than the predetermined learning threshold T1 (step # 24: No), processing is performed. Ends as is.

  On the other hand, when the learning value of the recognized feature information A for the target feature stored in the feature learning database DB3 is equal to or greater than a predetermined learning threshold T1 (step # 24: Yes), the estimated position The determination unit 43 determines the estimated position pg of the target feature (step # 25). Thereafter, the learned feature information generation unit 44 associates the estimated position pg determined in step # 25 for the target feature with the feature attribute information included in the recognized feature information A for the target feature. Learning feature information Fb is generated (step # 26). Then, the generated learned feature information Fb is stored in the feature database DB2 (step # 27). This is the end of the feature learning processing procedure.

22. Next, the procedure of the vehicle position recognition process (vehicle position recognition program) executed in the vehicle position recognition device 2 included in the navigation device 1 according to the present embodiment will be described. FIG. 12 is a flowchart showing an overall procedure of the vehicle position recognition process according to the present embodiment. FIG. 13 is a flowchart showing the procedure of the vehicle position information correction process in step # 37 of FIG. 12 in the present embodiment. The processing procedure described below is executed by hardware and / or software (program) or both of the above-described functional units. When each of the above function units is configured by a program, the arithmetic processing device included in the navigation device 1 operates as a computer that executes the vehicle position recognition program that configures each of the above function units. Below, first, the whole procedure of the own vehicle position recognition process will be described.

22-1. Overall Procedure of Own Vehicle Position Recognition Process As shown in FIG. 12, in the own vehicle position recognition process in the own vehicle position recognition device 2, first, the own vehicle position information acquisition unit 16 acquires the own vehicle position information P (step # 31). Next, the road information acquisition unit 23 extracts and acquires road information Ra around the host vehicle C from the map database DB1 based on the host vehicle position information P acquired in step # 31 (step # 32). . Thereafter, the image information acquisition unit 12 acquires image information G around the host vehicle C imaged by the imaging device 11 (step # 33). Then, the image recognition unit 24 performs image recognition processing of the target feature included in the image information G (step # 34). If the target feature is not recognized in step # 34 (step # 35: No), the process returns to step # 31, and the vehicle position information P, the image information G, and the like are acquired again. When the target feature included in the image information G is recognized in step # 34 (step # 35: Yes), the learned target extraction unit 52 of the vehicle position information correction unit 51 recognizes the recognized target feature. The course change learning information S about the object is extracted and acquired from the course change database DB5 (step # 36). Then, based on the course change learning information S acquired at step # 36, the first correction processing unit 53 of the own vehicle position information correction unit 51 performs a process of correcting the own vehicle position information P (step # 37). The procedure of the correction processing of the vehicle position information P by the first correction processing unit 53 will be described in detail later based on the flowchart shown in FIG. Above, the procedure of the own vehicle position recognition process in the own vehicle position recognition apparatus 2 is complete | finished.

22-2. Procedure of own vehicle position information correction processing Next, a procedure of own vehicle position information correction processing by the own vehicle position information correction unit 51 will be described. In the own vehicle position information correction process in step # 37 of FIG. 12, first, the course change information Ba and the average distance information Sc (FIG. 7) included in the course change learning information S acquired in step # 36, as shown in FIG. (See (c)) is acquired (step # 41). Next, information on the position in the road information Ra of the course change intersection N specified by the course change information Ba is acquired (step # 42). In the example of FIG. 9, the information on the position of the node n (ID = OxOx) corresponding to the course change intersection N in the road information Ra acquired in step # 32 is included in the course change information Ba. It is acquired based on the specific information Be and the road information Ra. Thereafter, it is determined whether or not the vehicle position indicated by the vehicle position information P has reached the course change intersection N (step # 43), and the movement distance of the vehicle C from the recognition position of the target feature is It is determined whether or not the distance indicated by the average distance information Sc has been reached (step # 44). Here, the judgment of step # 43 is based on the road information of the node n (ID = OxOx) corresponding to the course change intersection N acquired in step # 42. This is based on whether or not the position in Ra matches. Further, the determination in step # 44 is based on the movement distance of the host vehicle C detected by the distance sensor 15 starting from the recognition position of the target feature when the target feature has been successfully recognized in step # 35. This is performed based on whether or not the distance indicated by the average distance information Sc acquired in # 42 is greater than or equal to the distance.

  And before the moving distance of the own vehicle C from the recognition position of the target feature reaches the distance indicated by the average distance information Sc (step # 44: No), the own vehicle position indicated by the own vehicle position information P is When the vehicle has reached the route change intersection N (step # 43: Yes), it means that the vehicle position information P has been recognized as being ahead of the actual position of the vehicle C due to an error. . Accordingly, in this case, the first correction processing unit 53 stops the own vehicle position indicated by the own vehicle position information P on the route change intersection N (step # 45). Specifically, as shown in FIG. 9A, the vehicle position indicated by the vehicle position information P is set to the node n (ID = ○ × ○ ×) in the road information Ra corresponding to the route change intersection N. Stop on top. Next, it is determined again whether or not the moving distance of the host vehicle C from the recognition position of the target feature has reached the distance indicated by the average distance information Sc (step # 46). And until the moving distance of the own vehicle C from the recognition position of the target feature reaches the distance indicated by the average distance information Sc (step # 44: No), the own vehicle position indicated by the own vehicle position information P is determined. The state stopped on the route change intersection N is continued (step # 45). Thereafter, when the moving distance of the host vehicle C from the recognition position of the target feature has reached the distance indicated by the average distance information Sc (step # 44: Yes), the host vehicle position information P is acquired as usual. The normal operation state is restored (step # 47).

  On the other hand, before the own vehicle position indicated by the own vehicle position information P reaches the route change intersection N (step # 43: No), the moving distance of the own vehicle C from the recognition position of the target feature is the average distance information. When the distance indicated by Sc is reached (step # 44: Yes), it means that the vehicle position information P is recognized as being delayed behind the actual position of the vehicle C due to an error. Therefore, in this case, the first correction processing unit 53 instantaneously moves the own vehicle position indicated by the own vehicle position information P onto the route change intersection N specified by the route change information Ba (step # 48). ). Thereafter, the vehicle position information P is returned to the normal operation state where the vehicle position information P is acquired as usual (step # 47). This is the end of the procedure for correcting the vehicle position information.

23. Other embodiments

(1) In the above embodiment, an example has been described in which the intersection specifying information Be constituting the route change information Ba is information on the route of the link that the host vehicle C has traveled during the route change operation. However, the embodiment of the present invention is not limited to this. Therefore, for example, the intersection specifying information Be constituting the route change information Ba may be information on the node n corresponding to the intersection where the route change operation of the host vehicle C is performed. One. In this case, the intersection specifying information Be can be, for example, node identification information (node ID or the like) for specifying the node n corresponding to the intersection where the host vehicle C has performed the course changing operation.

(2) In the above-described embodiment, when the vehicle position information correction unit 51 corrects the vehicle position information P, the vehicle position information P matches the actual position of the vehicle C at the route change intersection N. An example in which the vehicle position indicated in the vehicle position information P is stopped on the route change intersection N or the vehicle position indicated in the vehicle position information P is instantaneously moved on the route change intersection N. explained. However, the specific mode of the correction processing of the vehicle position information P by the first correction processing unit 53 of the vehicle position information correction unit 51 is not limited to this. Therefore, for example, after the target feature is recognized by the image recognition unit 24, the host vehicle position indicated by the host vehicle position information P between the target feature recognition position and the course change intersection N is that of the host vehicle C. The moving speed of the host vehicle position information P toward the course change intersection N is adjusted so as to gradually approach the actual position, and the host vehicle position indicated by the host vehicle position information P at the course change intersection N is the actual position of the host vehicle C. It is one of the preferred embodiments of the present invention to correct the vehicle position information P so as to match. In this case, specifically, referring to FIGS. 9A and 9B, the position in the road information Ra from the recognized position of the target feature f1 based on the own vehicle position information P to the course change intersection N is shown. As for the distance to the node n, the time when the vehicle position indicated by the vehicle position information P moves and the time when the vehicle C travels the average distance information Sc included in the course change learning information S coincide with each other. It is preferable to adjust the moving speed of the vehicle position toward the route change intersection N indicated by the vehicle position information P. At this time, if the own vehicle position information P is recognized as being advanced ahead of the actual position of the own vehicle C due to an error, the moving speed toward the course change intersection N of the own vehicle position information P is automatically determined. If the vehicle position information P is adjusted so as to be slower than the actual moving speed of the vehicle C and the vehicle position information P is recognized as being delayed behind the actual position of the vehicle C due to an error, the vehicle position information The moving speed toward the route change intersection N of P is adjusted so as to be faster than the actual moving speed of the host vehicle C.

(3) In the above embodiment, the image recognition unit 24 always performs the image recognition process on the image information G acquired by the image information acquisition unit 12, so that the target location included in the image information G is obtained. The case where processing for recognizing an object is performed has been described as an example. However, since the vehicle position recognition device 2 includes the feature database DB2 as described above, it is also suitable as a configuration for performing image recognition processing using the feature information F stored in the feature database DB2. It is. In this case, the own vehicle position recognition device 2 includes the own vehicle in addition to the feature database DB2 as the feature information storage means in which the feature information F including the position information and attribute information about the plurality of target features is stored. On the basis of the position information P, a feature information acquisition unit that acquires the feature information F around the host vehicle C from the feature database DB2 is provided, and the image recognition unit 24 is based on the acquired feature information F. It is preferable that the target feature included in the image information G is recognized. At this time, the feature information F acquired by the feature information acquisition unit may be either the initial feature information Fa or the learned feature information Fb described above. With such a configuration, it is possible to increase the recognition rate of the target feature and suppress erroneous recognition. Therefore, it is possible to improve the accuracy of the process of correcting the vehicle position information P and generating the course change learning information S.

(4) In the above embodiment, as shown in FIG. 7B, for a plurality of detection operation information B for the same route change operation, a set of route change information Ba and feature specifying information Bb in common. An example of storing information in the course change learning database DB4 as information has been described. However, the format of the detection operation information B when stored in the course change learning database DB4 is not limited to this. Therefore, it is also a preferred embodiment of the present invention that a plurality of detection operation information B for the same course changing operation is stored separately in the course changing learning database DB4.

(5) In the above-described embodiment, the average distance information Sc is generated by determining the average value of the plurality of distance information Bc for the same course change operation indicated in the plurality of detection action information B for the same course change operation, The case where this is the distance information of the course change learning information S has been described as an example. However, the content of the distance information of the course change learning information S is not limited to this. Therefore, for example, based on the distribution of a plurality of distance information Bc for the same course changing operation, other representative values such as the mode value and median value of the distribution are determined as the statistical distance information, and the statistical distance information Is also the distance information of the course change learning information S is one of the preferred embodiments of the present invention.

(6) In the above embodiment, an example has been described in which the recognition position information acquired by the recognition position information acquisition unit 41 is information representing the position of the host vehicle C when image recognition is successful. However, the position information acquired as the recognition position information indicating the recognition position of the target feature is not limited to this. Therefore, for example, for a target feature that has been successfully image-recognized, based on the vehicle position information P and the image recognition result of the image information G, the position of the target feature on the road based on the vehicle position information P It is also one of preferred embodiments of the present invention that the recognition position information acquisition unit 41 acquires the position of the target feature on the road as the recognition position information.

(7) In the above embodiment, the estimated position determination unit 43 determines the mode of the distribution based on the distribution of the plurality of recognized feature information A for the same target feature. An example has been described in which it is determined as pa and the estimated recognition position pa is converted into a position on the road of the target feature to determine the estimated position pg of the target feature. However, the determination method of the estimated position pg by the estimated position determination unit 43 is not limited to this. Therefore, for example, based on the distribution of the recognized feature information A, another representative value such as an average value or a median value of the distribution may be determined as the estimated recognition position pa of the target feature. This is one of the preferred embodiments.

(8) In the above embodiment, the case where various road markings provided on the road surface are the target features has been described as an example. However, the embodiment of the present invention is not limited to this, and various features provided around the road can be the target features. Therefore, for example, various features such as road signs, guide signs, signboards, traffic lights, manholes, and the like can be used as target features.

(9) The way of dividing the databases DB1 to DB5 in the above embodiment is merely an example, and the hardware configuration of each of the databases DB1 to DB5 is not limited. Therefore, for example, the feature database DB2 and the feature learning database DB3 are one database, the course change learning database DB4 and the course change database DB5 are one database, or the map database DB1 and the feature database DB2 are Any configuration such as a single database can be employed.

(10) In the above embodiment, the case where all the configurations of the navigation device 1 including the vehicle position recognition device 2 according to the present invention are mounted on the vehicle C has been described as an example. However, the embodiment of the present invention is not limited to this. That is, a server device in which a part of the configuration of the vehicle position recognition device 2 including each of the databases DB1 to DB5 is communicably connected to a plurality of vehicles C via a wireless communication line or the like as shown in FIG. It is also one of the preferred embodiments of the present invention to have a configuration installed at 60. If the own vehicle position recognizing device 2 has such a configuration, the route change learning database DB4 or the feature learning in which the learning result of the route changing operation by the plurality of vehicles C and the learning result of the target feature are installed in the server device 60 are used. It can be accumulated in the database DB3. Therefore, it is possible to quickly generate the course change learning information S and the learning feature information Fb by using more detection operation information B and recognition feature information A. In addition, by installing the map database DB1 and the feature database DB2 in the server device 60 and updating them as needed, the latest road information Ra, feature information F, etc. are always provided to the driver of the vehicle C. Is possible. The configuration of the vehicle position recognition device 2 installed in the server device 60 is not limited to each of the databases DB1 to DB5, and needs to be mounted on the vehicle C such as the imaging device 11 and the vehicle position information acquisition unit 16. All configurations other than a certain configuration can be installed in the server device 60.

(11) In each of the above embodiments, an example in which the vehicle position recognition device 2 according to the present invention is used for the navigation device 1 has been described. However, the embodiment of the present invention is not limited to this. Therefore, for example, the present invention is different from the above-described embodiment, such as using the vehicle position recognition device 2 according to the present invention for a travel control device for a vehicle unrelated to the navigation device 1. Of course, it is also possible to apply to.

  INDUSTRIAL APPLICABILITY The present invention can be suitably used for a host vehicle position recognition device and a host vehicle position recognition program that corrects host vehicle position information using a detection result of a course change operation of the host vehicle.

The block diagram which shows schematic structure of the navigation apparatus containing the own vehicle position recognition apparatus which concerns on embodiment of this invention. The figure which shows the example of a structure of the map information memorize | stored in the map database The figure which shows the example of the feature information of the road marking memorize | stored in the feature database The figure which shows an example of the arrangement configuration of the imaging device to the own vehicle Explanatory drawing for demonstrating the outline of the learning process of the feature information based on the image recognition result of the target feature FIG. 5B is an enlarged view of the main part of the learning value stored in the learning database shown in FIG. The figure which shows the content of the detection operation information regarding the specific example of the course change operation | movement of the own vehicle, and course change learning information The figure which shows the locus | trajectory of the own vehicle when not correcting the own vehicle position information which concerns on this invention, and own vehicle position information The figure which shows the locus | trajectory of the own vehicle at the time of correcting the own vehicle position information which concerns on this invention, and own vehicle position information The flowchart which shows the whole procedure of the learning process of the course change operation | movement which concerns on embodiment of this invention. The flowchart which shows the procedure of the feature learning process which concerns on embodiment of this invention. The flowchart which shows the whole procedure of the own vehicle position recognition process which concerns on embodiment of this invention. The flowchart which shows the procedure of the own vehicle position information correction process which concerns on embodiment of this invention. The figure which shows the example by which the one part structure of the own vehicle position recognition apparatus which concerns on embodiment of this invention was installed in the server apparatus.

Explanation of symbols

1: Navigation device 2: Vehicle position recognition device 12: Image information acquisition unit (image information acquisition means)
16: Own vehicle position information acquisition unit (own vehicle position information acquisition means)
17: Course change detection unit (Course change operation detection means)
23: Road information acquisition unit (road information acquisition means)
24: Image recognition unit (image recognition means)
28: Application program 31: Guidance information output means 46: Distance information generation unit (distance information acquisition means)
50: Course change learning information generation unit (course change learning information generation means)
51: Own vehicle position information correction unit (own vehicle position information correction means)
DB1: Map database (map information storage means)
DB2: Feature database (feature information storage means)
DB4: Course change learning database (detection operation information storage means)
DB5: Course change database (Course change learning information storage means)
C: own vehicle G: image information P: own vehicle position information M: map information Ra: road information n: node k: link F: feature information f1: target feature B: detection operation information Ba: route change information Bc: Distance information (distance information included in detection operation information)
S: Course change learning information Sc: Average distance information (distance information possessed by course change learning information)

Claims (10)

Own vehicle position information acquisition means for acquiring own vehicle position information representing the current position of the own vehicle;
Road information acquisition means for acquiring road information;
Image information acquisition means for acquiring image information around the host vehicle;
Image recognition means for performing image recognition processing of a target feature included in the image information;
It is generated by learning the previous course change operation of the host vehicle, and the course change information that can identify the course change intersection that is the intersection where the course change operation of the host vehicle has been performed, and the image is recognized before the course change operation. Route change learning information storage means storing route change learning information having distance information from the recognized position of the target feature to the position where the course change operation of the host vehicle has been performed,
When the target feature is recognized by the image recognition means, based on the course change learning information about the target feature, the recognition position of the target feature based on the own vehicle position information is changed to the distance information. Vehicle position correction means for correcting the vehicle position information so that the position traveled by the indicated distance matches the position in the road information of the route change intersection specified by the route change information;
A vehicle position recognition device comprising:   When the vehicle position correction means recognizes that the vehicle position information is ahead of the actual position of the vehicle due to an error, the vehicle position indicated in the vehicle position information is After reaching the route change intersection, the vehicle position indicated in the vehicle position information is changed to the route until the vehicle actually travels the distance indicated in the distance information from the recognition position of the target feature. The own vehicle position recognition device according to claim 1, wherein the own vehicle position information is corrected by stopping on the changed intersection.   The own vehicle position correcting means, when the own vehicle position information is recognized as being delayed behind the actual position of the own vehicle due to an error, the own vehicle is the distance information from the recognition position of the target feature The vehicle position information is corrected by moving the vehicle position indicated by the vehicle position information onto the route change intersection when the vehicle actually travels the distance shown in FIG. Car position recognition device. The road information is information representing a road by a connection relationship of a plurality of links and nodes,
The route change information includes one or both of information on a route of a link traveled by the host vehicle during the route change operation and information on a node corresponding to an intersection where the route change operation is performed. The vehicle position recognition device according to any one of claims 3 to 4.   5. The vehicle position according to claim 1, wherein the road information acquisition unit acquires the road information around the host vehicle from a predetermined map information storage unit based on the host vehicle position information. Recognition device. A course change detection means for detecting a course change operation of the host vehicle;
With respect to the target feature recognized by the image recognition unit prior to the detection of the route change operation of the host vehicle by the route change detection unit, from the recognition position of the target feature to the position where the route change of the host vehicle has been performed. Distance information acquisition means for acquiring distance information representing a distance;
Detection that stores detection operation information including route change information that can identify a route change intersection that is an intersection on which the route change operation has been performed, and the distance information that is acquired by the distance information acquisition unit regarding the route change operation Motion information storage means;
Course change learning that generates the course change learning information based on a plurality of the detected action information for the same course change operation stored in the detected action information storage means by detecting the same course change operation a plurality of times An information generating means,
The host vehicle position recognition apparatus according to any one of claims 1 to 5, wherein the course change learning information generated by the course change learning information generation unit is stored in the course change learning information storage unit. Feature information storage means storing feature information including position information and attribute information about a plurality of target features, and based on the vehicle position information, the feature information storage means and the surroundings of the vehicle A feature information acquisition means for acquiring feature information;
The vehicle position recognition device according to any one of claims 1 to 6, wherein the image recognition unit performs a recognition process of a target feature included in the image information based on the feature information.   The vehicle position recognition device according to any one of claims 1 to 7, wherein the target feature is a road marking provided on a road surface. The vehicle position recognition device according to any one of claims 1 to 8,
Map information storage means storing map information including the road information;
An application program that operates with reference to the map information and the vehicle position information recognized by the vehicle position recognition device;
Guidance information output means that operates according to the application program and outputs guidance information;
A navigation device comprising: Own vehicle position information acquisition step for acquiring own vehicle position information indicating the current position of the own vehicle;
A road information acquisition step for acquiring road information;
An image information acquisition step of acquiring image information around the host vehicle;
An image recognition step for performing an image recognition process of the target feature included in the image information;
When the target feature is recognized by the image recognition step, it is possible to identify a course change intersection that is generated by learning a past course change operation of the own vehicle and is an intersection where the course change operation of the own vehicle has been performed. Route change information, and distance information from the recognition position of the target feature to the position where the route change operation of the host vehicle is performed for the target feature that has been image-recognized before the route change operation. A course change learning information acquisition step of acquiring the course change learning information about the recognized target feature from the course change learning information storage means in which learning information is stored;
Based on the course change learning information acquired in the course change learning information acquisition step, a position that has traveled a distance indicated by the distance information from a recognized position of the target feature based on the vehicle position information is the course change. A vehicle position correction step for correcting the vehicle position information so as to coincide with the position in the road information of the course change intersection specified by the information;
A vehicle position recognition program that runs a computer.

Images