JP6589410B2 - Map generating apparatus and program - Google Patents

Description

  The present invention relates to a map generation device and a program.

  2. Description of the Related Art Conventionally, there has been known a technique for capturing a road surface and acquiring position information of a photographing point while a vehicle travels on a road (Patent Document 1). In the technique described in Patent Document 1, a road surface image is generated by converting a road into an image viewed from directly above by image conversion, and connecting the converted images based on the coordinates of the shooting location. Further, the road surface map is generated by translating the road surface image according to the position of the road surface image with high positional accuracy so that the corresponding points of the plurality of road surface images having different positions coincide with each other.

  In addition, a technique for acquiring and storing image data, distance measurement data, and positioning data by traveling a plurality of vehicles is known (Patent Document 2). In the technique described in Patent Document 2, data having an image of the same spot is extracted from the accumulated data, data including a captured image that does not include a moving object is identified based on pixel value dispersion, and a three-dimensional map is obtained. Generate data.

  In addition, a technique is known in which a travelable area is detected by a distance measuring sensor mounted on a vehicle and transmitted to a server together with position information and road marking characteristics (Patent Document 3). In the technique described in Patent Document 3, the server integrates the travelable area and the map database received based on the position information and the road marking feature, and updates the travelable area map.

JP 2010-175756 A JP2012-155660A JP 2014-228637 A

  In order to align a plurality of different road surface images so that the corresponding points of the road surface image coincide with each other as in the technique described in Patent Document 1, the position information of the shooting point needs to be highly accurate, There is a problem that the accuracy of normal GPS for navigation is not sufficient.

  Further, in the technique described in Patent Document 1, data of the same position is detected based on the correlation between the position information and the image, and a three-dimensional map is registered. Further, in the technique described in Patent Document 3, the alignment with the map database is performed based on the position information and the characteristic object obtained by the distance measuring sensor. In the techniques of Patent Document 1 and Patent Document 3 described above, the position information registered in the map is one of vehicle travel data or a position registered in advance in the map. In order to generate a wide-area map without a large deviation, the accuracy of absolute position information registered in the map must be high. However, in the techniques of Patent Document 1 and Patent Document 3 described above, a highly accurate absolute position is determined. There is a problem that it is difficult or a high-precision measurement must be performed in advance to generate a map.

  The present invention has been made in view of the above circumstances, and provides a map generation device and a program that can accurately generate map data even when a measurement error occurs with respect to the absolute position of the vehicle. The purpose is to do.

  In order to achieve the above object, a map generation apparatus according to the present invention includes a plurality of travel route data registered along a travel locus of a mobile object, and a surrounding environment obtained from a sensor mounted on the mobile object. Based on a database in which a plurality of road data generated based on information is stored, for each of the corresponding road data, the relative positional relationship between the pairs with respect to each of the pairs of the road data and other road data And a relative positional relationship between the road data and the virtual road data serving as a reference based on the deformation amount representing the relative positional relationship calculated for each of the pairs. A relative relationship estimating means for estimating a deformation amount representing the above, and the virtual running as a reference, each of the plurality of running road data being estimated for the running road data by the relative relationship estimating means. Is deformed according to the deformation quantity representing the relative positional relationship between the data, by integrating the plurality of track data modification is configured to include an integrated means for registering in the map data.

  In addition, the program of the present invention is a plurality of track data registered along a travel locus of a mobile object, and a plurality of programs generated based on surrounding environment information obtained from sensors mounted on the mobile object. For each of the corresponding track data, a deformation amount representing a relative positional relationship between the pairs is calculated for each pair of the track data and other track data, based on a database that stores the track data of Based on the deformation amount representing the relative positional relationship calculated for each of the pairs, the deformation amount representing the relative positional relationship between the track data and the virtual track data serving as a reference is estimated. Relative relationship estimation means, and each of the plurality of track data is estimated with respect to the track data by the relative relationship estimation means, and the reference virtual track data Depending on the deformation amount representing the relative positional relationship between deform, by integrating the plurality of road data obtained by modifying a program for functioning as an integrated unit to be registered in the map data.

  According to the map generation device of the present invention, the relative relationship estimation means obtains from a sensor mounted on the moving body, which is a plurality of traveling road data representing the road surface registered along the traveling locus of the moving body. Relative positional relationship between pairs for each pair of track data and other track data for each corresponding track data based on a database that accumulates multiple track data generated based on surrounding environment information A deformation amount representing the relative positional relationship between the track data and the virtual track data serving as a reference is calculated based on the deformation amount representing the relative positional relationship calculated for each of the pairs. Estimate the amount.

  Then, the integration unit transforms each of the plurality of track data according to the deformation amount that is estimated for the track data by the relative relationship estimation unit and represents the relative positional relationship with the virtual track data serving as a reference. Integrate multiple deformed track data and register it in map data.

  In this way, a plurality of track data representing road surfaces registered along the travel locus of the moving body, and a plurality of track data generated based on the surrounding environment information obtained from the sensor mounted on the mobile body For each of the corresponding track data, the amount of deformation representing the relative positional relationship between the pair is calculated for each of the corresponding track data and each other track data, Based on the deformation amount representing the relative positional relationship calculated with respect to, the deformation amount representing the relative positional relationship between the track data and the virtual track data serving as a reference is estimated, and each of the plurality of track data Is transformed in accordance with the deformation amount representing the relative positional relationship between the estimated virtual track data and the reference virtual track data, and the transformed multiple track data are integrated into map data. By, even when the measurement error for the absolute position of the vehicle has occurred, it is possible to accurately generate the map data.

  Further, the relative relationship estimation means of the present invention averages the deformation amount representing the relative positional relationship calculated for each of the pairs, and the relative position between the reference virtual lane data The deformation amount representing the relationship can be estimated.

  Further, the relative relationship estimation means of the present invention represents the relative positional relationship of the partial images between the pairs for each partial image of the track data for each pair of the track data and other track data. The amount of deformation is calculated, and based on the amount of deformation representing the relative positional relationship of the partial images calculated for each of the pairs, the road data and the reference for each partial image of the road data A deformation amount representing a relative positional relationship between virtual runway data is estimated, and the integration unit is configured to obtain, for each partial image, the partial images of each of the plurality of runway data by the relative relationship estimation unit. The partial images of the plurality of modified roadway data are integrated by being deformed according to the deformation amount representing the relative positional relationship between the partial image of the roadway data and the virtual roadway data serving as the reference. , It is possible to be registered in the map data.

  The present invention further includes classification means and overall relationship estimation means, wherein the classification means classifies each of the corresponding track data according to the position of the road surface represented by the track data, and the relative relationship estimation. Means for each class of the track data classified by the classifying means, for each of the track data belonging to the classification, relative to each pair of the track data and other track data; A deformation amount representing a positional relationship is calculated, and based on the deformation amount representing the relative positional relationship calculated for each of the pairs, the relative between the track data and the virtual track data serving as a reference A deformation amount representing a positional relationship is estimated, the road data is deformed according to the estimated deformation amount, and the overall relationship estimation unit is configured to deform the road data for each classification. For each pair of runway data and the other runway data that is classified into the classification and deformed, a deformation amount representing a relative positional relationship between the pairs is calculated, and for each of the pairs Based on the calculated deformation amount representing the relative positional relationship, the deformation amount representing the relative positional relationship between the running road data and the virtual running road data serving as an overall reference is estimated, and the integration unit includes: Each of the plurality of runway data is estimated for the runway data by the relative relationship estimation unit, and a deformation amount representing a relative positional relationship between the reference virtual runway data and the overall relationship estimation unit In accordance with the deformation amount representing the relative positional relationship between the overall reference virtual road data estimated with respect to the road data, and integrating the plurality of the changed road data Te, it is possible to be registered in the map data.

  Further, the overall relationship estimation means of the present invention is configured such that the relative positional relationship between the overall reference virtual lane data according to the number of the lane data belonging to the classification classified by the classification means. Can be estimated.

  In addition, the present invention provides a travel locus that generates a travel locus of the mobile body based on the position information and motion information of the mobile body acquired by the vehicle state detection unit that acquires the position information and motion information of the mobile body. The database may further include a generation unit, and the database may store the plurality of travel path data registered along the travel locus of the moving body generated by the travel locus generation unit.

  Further, according to the present invention, the traveling road data is a road surface image representing a road surface registered along a traveling locus of the moving body, and is based on surrounding environment information obtained from a sensor mounted on the moving body. It can be made to be a generated road surface image.

  Further, according to the present invention, the traveling road data is three-dimensional point cloud data registered along a traveling locus of the moving body, and is based on surrounding environment information obtained from a sensor mounted on the moving body. The generated three-dimensional point cloud data can be used.

  The storage medium for storing the program of the present invention is not particularly limited, and may be a hard disk or a ROM. Further, it may be a CD-ROM, a DVD disk, a magneto-optical disk or an IC card. Furthermore, the program may be downloaded from a server or the like connected to the network.

  As described above, according to the map generation device and the program of the present invention, a plurality of pieces of road data representing road surfaces registered along the travel locus of the moving body and obtained from the sensor mounted on the mobile body. Between each pair of the corresponding track data and other track data for each corresponding track data based on a database storing a plurality of track data generated based on the surrounding environment information. Calculate the amount of deformation representing the relative positional relationship, and based on the amount of deformation representing the relative positional relationship calculated for each of the pairs, the relative positional relationship between the track data and the reference virtual track data Is estimated, and each of the plurality of track data is deformed according to the deformation amount that is estimated for the track data and that represents the relative positional relationship between the reference virtual track data. By integrating multiple deformed track data and registering it in map data, it is possible to generate map data with high accuracy even when a measurement error occurs in the absolute position of the vehicle Is obtained.

It is a block diagram which shows the map production | generation apparatus concerning the 1st Embodiment of this invention. It is a figure which shows an example of a road surface image. It is a figure which shows an example of the road surface image classified according to the travel route. It is a figure showing the correspondence between road surface images. It is a figure showing the correspondence between the partial images of a road surface image. It is a figure showing the correspondence between road surface images from which classification differs. It is a figure showing the integrated road surface image. It is a flowchart which shows the content of the map generation process routine in the 1st Embodiment of this invention. It is a flowchart which shows the content of the driving | running | working data classification | category processing routine in the 1st Embodiment of this invention. It is a flowchart which shows the content of the relative relationship estimation process routine in the 1st Embodiment of this invention. It is a flowchart which shows the content of the whole relationship estimation process routine in the 1st Embodiment of this invention. It is a block diagram which shows the map production | generation apparatus which concerns on the 2nd Embodiment of this invention. It is a flowchart which shows the content of the map production | generation process routine in the 2nd Embodiment of this invention. It is a flowchart which shows the content of the relative relationship estimation process routine in the 2nd Embodiment of this invention. It is a block diagram which shows the map production | generation apparatus which concerns on the 3rd Embodiment of this invention. It is a block diagram which shows the structural example of the driving | running | working data generation part of the map generation apparatus which concerns on the 3rd Embodiment of this invention. It is a figure for demonstrating an example of the method of determining a solid object. It is an example of the road surface image produced | generated in the 3rd Embodiment of this invention. It is a block diagram which shows the map production | generation apparatus which concerns on the 4th Embodiment of this invention. It is a block diagram which shows the structural example of the driving | running | working data generation part of the map generation apparatus which concerns on the 4th Embodiment of this invention. It is a figure which shows the example of the partial point cloud of three-dimensional point cloud data. It is a figure which shows the example of estimation of the deformation | transformation amount showing the relative positional relationship between partial point groups. It is a flowchart which shows the content of the map generation process routine in the 4th Embodiment of this invention. It is a flowchart which shows the content of the relative relationship estimation process routine in the 4th Embodiment of this invention. It is a flowchart which shows the content of the whole relationship estimation process routine in the 4th Embodiment of this invention. It is a block diagram which shows the map generation system which concerns on the 5th Embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described in detail with reference to the drawings. In the present embodiment, a case where the present invention is applied to a map generation device that is mounted on a vehicle and generates map data from a plurality of road surface images will be described as an example. In this embodiment, a case where the moving body is a vehicle will be described as an example.

[First Embodiment]
<Configuration of map generation apparatus 10 according to the first embodiment>
As shown in FIG. 1, a map generation apparatus 10 according to the first embodiment of the present invention includes an external field observation sensor 12 that measures a surrounding environment of a vehicle, and a position / movement that measures position information and movement information of the vehicle. A measurement sensor 14, and a computer 16 that generates map data based on the surrounding environment information acquired by the external field observation sensor 12 and the vehicle position information and motion information measured by the position / motion measurement sensor 14. ing. The external field observation sensor 12 is an example of a sensor. The position / motion measurement sensor 14 is an example of vehicle state detection means.

  The external field observation sensor 12 measures the surrounding environment of the vehicle at each time. As the external field observation sensor 12, for example, a camera that acquires an image around the vehicle, a laser radar that measures distance information around the vehicle, and the like can be used. In the present embodiment, a case where a camera is used as the external field observation sensor 12 will be described as an example. The external field observation sensor 12 acquires a vehicle surrounding image captured by a camera as vehicle surrounding environment information.

  The position / motion measurement sensor 14 measures vehicle position, orientation, and speed information at each time as vehicle position information and motion information. As the position / motion measurement sensor 14, for example, GPS and IMU (IMU: Internal Measurement Unit (Inertial Measurement Unit)) can be used. In the present embodiment, a case where GPS and IMU are used as the position / motion measurement sensor 14 will be described as an example.

  The computer 16 includes a CPU, a ROM that stores a program for realizing a map generation processing routine that will be described later, a RAM that temporarily stores data, and a storage device such as an HDD. It is configured.

  As shown in FIG. 1, the computer 16 includes a travel trajectory generation unit 160 that generates a travel trajectory of the vehicle based on the vehicle position, orientation, and speed information measured by the position / motion measurement sensor 14, and external field observation. Based on the surrounding image of the vehicle acquired by the sensor 12 and the traveling locus of the vehicle generated by the traveling locus generation unit 160, a road surface image representing the road surface along the traveling locus of the vehicle is generated, and the road surface image and the traveling are generated. A travel data generation unit 162 that generates travel data representing a combination with a trajectory, and travel data storage that stores each of the travel data generated by the travel data generation unit 162 in the travel database 166 in association with the position of the travel trajectory Unit 164, a travel database 166 in which each of the travel data generated by the travel data generation unit 162 is stored, On the other hand, the travel data classification unit 168 that classifies each of the travel data stored in the travel database 166 according to the position of the road surface represented by the road surface image, and for each point, for each classification, For each of the belonging traveling data, a relative relationship estimating unit 170 that estimates a deformation amount representing a relative positional relationship between the road surface image and the reference virtual road surface image, and for each point, the traveling data of the point For each, an overall relationship estimation unit 172 that estimates a deformation amount representing a relative positional relationship between a road surface image and a virtual road surface image that serves as an overall reference, and a road image of travel data stored in the travel database 166. The amount of deformation estimated for the road surface image by the relative relationship estimation unit 170 and the amount of deformation estimated for the road surface image by the overall relationship estimation unit 172, respectively. A travel data integration unit 174 that integrates a plurality of deformed road surface images and registers them in the map database 176, and a map database 176 that stores the map data registered by the travel data integration unit 174. ing. The travel data integration unit 174 is an example of an integration unit, and the travel data classification unit 168 is an example of a classification unit.

  The travel trajectory generation unit 160 generates a travel trajectory of the vehicle based on the vehicle position, direction, and speed information measured by the position / motion measurement sensor 14 at each time. The travel locus is information including vehicle position, direction, and speed information at each time.

The travel data generation unit 162 is based on the surrounding image of the vehicle at each time acquired by the external field observation sensor 12 and the vehicle travel trajectory generated by the travel trajectory generation unit 160 along the road surface along the vehicle travel trajectory. Is generated and combined with the travel locus, absolute coordinates are given to the road image. A road surface image is an example of runway data.
Specifically, the travel data generation unit 162 converts the surrounding image of the vehicle at each time into an orthographic image in a state where the road surface is viewed from directly above, and based on the position and orientation at each time of the travel locus, A road surface image is generated by connecting the orthogonal projection images along the traveling locus. Then, the travel data generation unit 162 generates travel data representing the road surface image to which the absolute coordinates are given by combining the road surface image and the travel locus. FIG. 2 shows an example of a road surface image. The travel data generation unit 162 generates a road surface image as shown in FIG.

  The travel data accumulation unit 164 stores each travel data generated by the travel data generation unit 162 in a travel database 166 described later.

  Each travel data generated by the travel data generation unit 162 is stored in the travel database 166.

  The travel data classification unit 168 classifies each travel data corresponding to the point stored in the travel database 166 according to the position of the road surface represented by the road surface image.

  Specifically, the travel data classification unit 168 first determines a road surface image of each travel data of the point stored in the travel database 166 for each point, as shown in FIG. The vehicle is divided for each absolute position section, and section travel data including a section road surface image divided for each section is generated.

  Next, the travel data classification unit 168 classifies the segment travel data of the same section according to the travel route of the vehicle obtained from the travel locus included in the travel data corresponding to the section travel data. In FIG. 3, the example of the division road surface image of the division traveling data classified according to the traveling route of the vehicle is shown. In FIG. 3, two travel routes are shown, and the travel data classification unit 168 classifies the segment travel data according to the travel route of the vehicle as shown in FIG.

  Specifically, the travel data classifying unit 168 collects the sections determined to be the same route for each of the segment travel data of the same section and sets them as one class. Whether or not the two section travel data are the same route can be determined based on the distance between the travel trajectories and the travel direction included in the travel data corresponding to the section travel data. For example, the travel data classification unit 168 determines whether or not the two section travel data are the same route based on the condition that the travel directions are the same and the distance between the travel trajectories is equal to or less than a predetermined threshold. Can be determined. Further, when the error in the absolute position of the travel locus is large and it is difficult to determine the same route, the similarity of the section road surface image may be added to the condition.

  For each section of each point, the relative relationship estimation unit 170 for each section, for each section road image of the section traveling data classified into the classification by the traveling data classification section 168, the section road surface image and other sections For each pair with the road surface image, a deformation amount representing a relative positional relationship between the pair is calculated.

  Since each travel data stored in the travel database 166 includes an error in absolute position, the absolute positions of the section road surface images do not match even between the section travel data classified as the same travel route. . Further, since the absolute position error varies with time even within the same segment travel data, the shapes of the segment road surface images of different segment travel data do not completely match. For this reason, in the present embodiment, the relative positional relationship between the section traveling data is represented not by a single translation and rotation amount but by a deformation amount of the entire section road surface image, or a parallel movement and rotation amount at a plurality of points.

  FIG. 4 shows an example of obtaining the corresponding positions at a plurality of feature points and estimating the relative relationship between the divided road surface images for each of the divided road surface images belonging to the same classification.

  FIG. 5 shows an example of a method for obtaining correspondence at a plurality of feature points for each of the divided road surface images belonging to the same classification. For example, as a method for obtaining correspondence at a plurality of feature points, as shown in FIG. 5, partial images are extracted from a section road surface image with a predetermined interval and a predetermined size along a travel locus. For each partial image, it is possible to use a method of matching with another section road surface image to obtain a correspondence relationship.

  When estimating the correspondence using the partial images, the relative relationship estimation unit 170 for each partial image of the divided road surface image, for each pair of the divided road surface image and the other divided road surface images, the partial image between the pairs. The amount of deformation representing the relative positional relationship is calculated. Then, for each of the divided road surface images, the relative relationship estimation unit 170 performs, for each partial image of the divided road surface image, based on the deformation amount representing the relative positional relationship of the partial images calculated for each of the pairs. A deformation amount representing a relative positional relationship between the partial image and the partial image of the virtual section road surface image serving as a reference is estimated.

When matching is performed for each partial image, some partial images do not include a characteristic pattern, such as the partial image 3 illustrated in FIG. 5. When there is a partial image that does not include a characteristic pattern as in the partial image 3 of FIG. 5 described above, it may be difficult to accurately detect the corresponding position in the other section road surface image. Therefore, a method that considers the continuity of the correspondence between the partial images can be used. The deformation quantity representing the correspondence between the partial image i and _{x i,} partial images i = 1, ..., the amount of deformation of the _{m x = {x 1, ...} , x m} represented by. At this time, the deformation amount x that minimizes E (x) is estimated according to the following equation (1).

Here, C _i (x _i ) in the above expression represents the matching cost when the correspondence of the partial image i is the deformation amount x _i . Matching cost represents the difference between the partial image i, position the partial image i in deformation amount x _i, and the other partial image of one compartment road image when deformed the rotation angle (shape). For the calculation of the matching cost in the above formula, the sum of absolute values of pixel value differences can be used. Further, when calculating the matching cost, calculation such as obtaining a difference after performing edge extraction may be performed.

Further, S (x _j , x _{j + 1} ) in the above expression represents the difference between the deformation amounts x _j and x _{j + 1} of the adjacent partial images. When the deformation amount is a translation amount, the distance between two translation vectors can be used. In the case of deformation including rotation, a weighted sum with a difference in rotation amount can be used.

  By obtaining all deformation amounts x so as to minimize the above expression E (x), a deformation amount with a low matching cost is selected for a partial image that includes a characteristic pattern and is easy to match, and a correspondence is obtained. For a partial image that is difficult to correct, the amount of deformation that is smoothly connected with little change is estimated from the amount of deformation of an adjacent partial image.

Then, the relative relationship estimation unit 170 calculates the relative positional relationship calculated for each of the pairs with respect to each section at each point, for each section, for each section road image of the section traveling data classified into the classification. Is estimated based on the deformation amount representing the relative positional relationship between the section road image and the partial image of the virtual section road image serving as a reference.
Specifically, when the relative positional relationship between the section road surface images of all the section traveling data of the same classification is estimated, the relative relationship estimation unit 170 each of the section road surface images of the section traveling data classified into the classification. For each partial image of the partitioned road surface image, a deformation amount representing the relative positional relationship between the partial image and the reference virtual partitioned road surface image is estimated based on the deformation amount representing the relative positional relationship. . For example, the deformation amount from the section road image of the section travel data 1 to the section road surface image of the section travel data 2 is expressed as x ^1,2 = {x ₁ ^1,2 , x ₂ ^1,2 ,..., X _m ^{1, 2} } and assuming that there are k section traveling data of the same classification, the relative relationship estimation unit 170 calculates the deformation amount of the section traveling data 1 from the section road surface image to the reference position according to the following equation (2). To do. Note that the deformation amount from the section road surface image to the reference position corresponds to a deformation amount representing a relative positional relationship between the section road surface image and the reference virtual section road surface image.

Note that x ^{i, i} = {0,..., 0}. The deformation amount x ^ ¹ calculated by the above equation (2) is based on the average of the section road surface images of all section travel data as a reference position, and is obtained from the section road surface images of other section travel data in the same manner. When the deformation amount to the reference position is calculated and the section road surface image of each section traveling data is deformed according to the calculated deformation amount to the reference position, the section road surface images of all the section traveling data coincide at the reference position. .

  Note that the amount of deformation to the reference position is not an average, but may be an average excluding data that is far away, a weighted average based on the positioning accuracy of the absolute position, or the like.

  Further, the relative relationship estimation unit 170 represents a relative positional relationship between the partial image of the divided road surface image and the reference virtual divided road surface image estimated for the partial image for each of the divided road surface images. Deform according to the amount of deformation.

  The overall relationship estimation unit 172 is transformed by the relative relationship estimation unit 170 for each segment of each point for each segment road surface image of the segment traveling data classified into the classification by the traveling data classification unit 168 for each classification. For each pair of the divided road surface image and the other divided road surface image classified into each category and deformed, a deformation amount representing a relative positional relationship between the pairs is calculated.

  FIG. 6 shows an example of a method in which the overall relationship estimation unit 172 estimates the relative positional relationship. Specifically, as shown in FIG. 6, the overall relationship estimation unit 172 estimates the relative positional relationship between the section road surface images of the section traveling data of different classifications by the same method as the relative relationship estimation unit 170. At this time, the section road surface image of each section travel data is deformed based on the deformation amount to the reference position estimated by the relative relationship estimation unit 170. Between different classifications, the travel route is different, so there are few common parts of the section road surface image, and it is difficult to obtain the correspondence relationship, but by applying the transformation to the reference position in advance by the relative relationship estimation unit 170, Since the position error of the travel data is reduced, the range of the deformation amount can be reduced, and it is easy to obtain the correspondence only with the common part. In the present embodiment, overall relationship estimating section 172 calculates the deformation amount using each of the partial images set by relative relationship estimating section 170 as a common portion.

Then, the overall relationship estimation unit 172 calculates the relative positional relationship calculated for each of the pairs for each section of each point, for each section of the section road surface image of the section traveling data classified into the section. Is estimated based on the deformation amount indicating the relative positional relationship between the partitioned road surface image and the virtual partitioned road surface image serving as the overall reference.
For example, ^assuming that the deformation amount from the segment travel data i of the category p to any one of the segment travel data j of the class q is y ^{pi, qj} and the number of classifications is r, the segment travel data i of the category p The deformation amount to the entire reference position is calculated by the following formula. The amount of deformation from the section road surface image to the entire reference position corresponds to the amount of deformation representing the relative positional relationship between the section road surface image and the virtual section road surface image serving as the entire reference.

However, y ^{pi, pi} = {0, ..., 0}. W ^q is the weight of the classification q. For example, the ratio of the number of section traveling data belonging to the classification q to the number of all section traveling data can be used as w ^q . As a result, the amount of deformation weighted according to the number of section traveling data belonging to each classification becomes the amount of deformation representing the relative positional relationship with the virtual section road surface image serving as the entire reference.

  The travel data integration unit 174 estimates the section road surface image for each section of each point by the relative relationship estimation unit 170 for each section road surface image of the section travel data classified into the classification. A deformation amount representing a relative positional relationship with the reference virtual section road image, and an overall reference virtual section road image estimated by the overall relationship estimation unit 172 for the section road image. Are deformed according to the deformation amount representing the relative positional relationship between them, and the deformed plural section road surface images are integrated and registered in the map database 176.

Specifically, the travel data integration unit 174 calculates the segment travel data based on the relative positional relationship within the classification estimated by the relative relationship estimation unit 170 and the relative positional relationship between the categories estimated by the overall relationship estimation unit 172. Correct the absolute position of. The travel data integration unit 174 corrects the position of the section travel data i of the classification p by applying the following deformation amount z ^pi to each partial image of the section road surface image of the section travel data i.

  Then, the travel data integration unit 174 integrates all of the corrected section travel data, and registers each integrated section travel data as map data in the map database 176. The travel data integration unit 174 generates a road surface image obtained by integrating all the section road surface images as shown in FIG. The integrated road surface image can be an image obtained by superimposing all the road surface images and taking the average of the pixel values. The integrated road surface image and all travel locus data whose positions are corrected are registered in the database as map information.

  The map database 176 stores map data registered by the travel data integration unit 174.

<Operation of Map Generation Device 10 according to First Embodiment>
Next, the operation of the map generation device 10 according to the first embodiment will be described. When a vehicle on which the map generation device 10 is mounted travels, a surrounding image of the vehicle is acquired by the external observation sensor 12, and the position, direction, and speed information of the vehicle at each time is measured by the position / motion measurement sensor 14. The traveling locus generation unit 160 generates a traveling locus of the vehicle based on the position, direction, and speed information of the vehicle at each time measured by the position / motion measurement sensor 14. In addition, the travel data generation unit 162 follows the travel locus of the vehicle based on the surrounding image of the vehicle at each time acquired by the external field observation sensor 12 and the travel locus of the vehicle generated by the travel locus generation unit 160. A road surface image representing the road surface is generated, and absolute coordinates are given to the road surface image in combination with the travel locus. Then, the travel data generation unit 162 generates travel data representing the road surface image to which the absolute coordinates are given by combining the road surface image and the travel locus. Next, the travel data accumulation unit 164 stores each of the travel data generated by the travel data generation unit 162 in the travel database 166.

  When a plurality of travel data is stored in the travel database 166, the map generation device 10 executes a map generation processing routine shown in FIG. 8 for each point at a predetermined timing.

  In step S100, the travel data classification unit 168 classifies each of the travel data corresponding to the target points stored in the travel database 166 according to the position of the road surface represented by the road surface image. Step S100 is realized by a travel data classification process routine shown in FIG.

<Running data classification processing routine>
In step S200, the road image of each travel data of the target point stored in the travel database 166 is divided into sections of predetermined absolute positions, and the section travel data including the section road surface image divided for each section. Is generated.

  In step S202, the classification is set as undecided for each of the segment traveling data generated in step S200.

  In step S204, 1 is stored in the variable n.

  In step S206, it is determined whether or not there is segment traveling data whose classification is not yet determined. If there is segment travel data whose classification has not been determined, the process proceeds to step S208. On the other hand, if there is no section traveling data whose classification is not yet determined, the process proceeds to step S216.

  In step S208, one piece of section traveling data whose classification is not determined is selected, and the selected section traveling data is set as reference data.

  In step S210, the classification number of the reference data set in step S208 is set to n updated last time in step S204 or step S214 described later.

  In step S212, based on the travel trajectory included in the reference data set in step S208 and each travel trajectory included in other segment travel data, the segment travel data having the same travel route as the reference data. Each classification number is set to the classification number n set in step S210, n is incremented in step S214, and the process returns to step S206.

  In step S216, the classification result set in step S212 is output as a result, and the running data classification processing routine is terminated.

  Next, returning to the map generation processing routine, in step S102, the relative relationship estimation unit 170 for each section of the section travel data for each section of the target point, for each classification output in step S100. For each pair of the section road surface image and the other section road surface image, a deformation amount representing a relative positional relationship between the pairs is calculated. And the relative relationship estimation part 170 is a deformation | transformation showing the relative positional relationship calculated with respect to each of a pair about each division road surface image of the division traveling data classified into the said classification | category for every classification with respect to each division. Based on the amount, a deformation amount representing a relative positional relationship between the road surface image and a reference virtual road surface image is estimated. Step S102 is realized by the relative relationship estimation processing routine shown in FIG.

<Relative relationship estimation processing routine>
In step S300, one section is set.

  In step S302, one classification is set.

  In step S304, one section road surface image of the section traveling data belonging to the section set in step S300 and belonging to the classification set in step S302 is set.

  In step S306, a partial image group of the section road surface image set in step S304 is set.

  In step S308, a pair with a corresponding partial image of another section road surface image different from the section road surface image set for each of the partial images included in the partial image group set in step S306 is set. And the deformation | transformation amount showing the relative positional relationship between a pair is calculated according to said Formula (1) about each of the partial image contained in the partial image group of the set division road surface image.

  In step S310, it is determined whether or not the processing in step S308 has been executed with all other divided road surface images. If the process of step S308 is executed with all other section road surface images, the process proceeds to step S312. On the other hand, when the process of step S308 is not executed with all other divided road surface images, the process returns to step S308.

  In step S312, for each of all partial images included in the partial image group set in step S306, the partial image is calculated according to the equation (2) based on the deformation amount calculated in step S308. And a deformation amount representing a relative positional relationship between the partial image of the reference virtual road surface image is estimated.

  In step S314, it is determined whether or not the processing in steps S304 to S312 has been executed for the section road surface images of all section traveling data belonging to the classification set in step S302. When the processing of step S304 to step S312 is executed for the section road surface images of all section traveling data belonging to the classification, the process proceeds to step S316. On the other hand, if there is a section road surface image of section traveling data that has not been subjected to the processing of step S304 to step S312 above, the process returns to step S304.

  In step S316, it is determined whether or not the processing in steps S302 to S314 has been executed for all the classifications belonging to the section set in step S300. If the processes in steps S302 to S314 have been executed for all the classifications belonging to the section set in step S300, the process proceeds to step S318. On the other hand, if there is a classification for which the processes in steps S302 to S314 are not executed, the process returns to step S302.

  In step S318, it is determined whether or not the processing in steps S300 to S316 has been executed for all sections of the travel data. When the processes in steps S300 to S316 are executed for all the sections, the process proceeds to step S320. On the other hand, if there is a partition for which the processes in steps S300 to S316 are not performed, the process returns to step S300.

  In step S320, each deformation amount estimated in step S312 is output as a result, and the relative relationship estimation processing routine is terminated.

  Next, returning to the map generation processing routine, in step S104, the relative relationship estimation unit 170 determines the partial image of the section road surface image for each of the section road surface images, and the reference image estimated for the partial image in step S102. Is deformed in accordance with a deformation amount representing a relative positional relationship with a partial image of the virtual section road surface image.

  In step S106, the overall relationship estimation unit 172 for each segment of the target point, for each classification output in step S100, for each segment road surface image of the segment travel data, the segment road surface deformed in step S104. For each pair of an image and another segmented road surface image that is classified into a plurality of classifications and deformed, a deformation amount representing a relative positional relationship between the pairs is calculated. The overall relationship estimation unit 172 then calculates the relative positional relationship calculated for each of the pairs for each division of the target point, for each division, for each division road surface image of the division traveling data classified into the classification. Is estimated based on the deformation amount indicating the relative positional relationship between the partitioned road surface image and the virtual partitioned road surface image serving as the overall reference. Step S106 is realized by the overall relationship estimation processing routine shown in FIG.

<Overall relationship estimation processing routine>
In step S400, one section is set.

  In step S402, one classification is set.

  In step S404, one section road surface image of the section traveling data belonging to the section set in step S400 and belonging to the classification set in step S402 is set.

  In step S406, a partial image group of the section road surface image set in step S404 is set.

  In step S408, for each partial image included in the partial image group set in step S406, a pair with a corresponding partial image of a section road surface image of section traveling data belonging to one classification is set. And the deformation | transformation amount showing the relative positional relationship between a pair is calculated according to said Formula (1) about each of the partial image contained in the partial image group of the set division road surface image.

  In step S410, it is determined whether or not the processing in step S408 has been executed with all classifications. If the process of step S408 has been executed for all classifications, the process proceeds to step S412. On the other hand, if the process of step S408 has not been executed with all classifications, the process returns to step S408.

  In step S412, for each of all the partial images included in the partial image group set in step S406, the partial image is calculated according to the equation (2) based on the deformation amount calculated in step S408. And the amount of deformation representing the relative positional relationship between the virtual road surface image serving as the overall reference is estimated.

  In step S414, it is determined whether or not the processes in steps S404 to S412 have been executed for the section road surface images of all the section traveling data belonging to the classification set in step S402. If the processing of step S404 to step S412 is executed for the section road surface images of all section traveling data belonging to the classification, the process proceeds to step S416. On the other hand, when there is a section road surface image of the section travel data for which the processes of steps S404 to S412 are not executed, the process returns to step S404.

  In step S416, it is determined whether or not the processing in steps S402 to S414 has been executed for all the classifications belonging to the section set in step S400. When the processes of steps S402 to S414 have been executed for all the classifications belonging to the section set in step S400, the process proceeds to step S418. On the other hand, if there is a classification for which the processing of step S402 to step S414 is not performed, the process returns to step S402.

  In step S418, it is determined whether or not the processing in steps S400 to S416 has been executed for all sections of the travel data. When the processes in steps S400 to S416 are executed for all the sections, the process proceeds to step S420. On the other hand, if there is a partition for which the processes in steps S400 to S416 are not performed, the process returns to step S400.

  In step S420, each deformation amount estimated in step S412 is output as a result, and the overall relationship estimation processing routine is terminated.

  Next, returning to the map generation processing routine, in step S108, the travel data integration unit 174 for each section of the target point, for each classification, for each of the section road surface image of the section travel data classified into the classification. Deformation is performed according to the deformation amount estimated in step S102 and the deformation amount estimated in step S106, and a plurality of deformed divided road surface images are integrated.

  In step S110, the travel data integration unit 174 registers the data integrated in step S108 in the map database 176, and ends the map generation processing routine.

  As described above, according to the map generation device according to the present embodiment, a plurality of road surface images representing road surfaces registered along the traveling locus of the vehicle, and obtained from sensors mounted on the vehicle. Based on a database that stores a plurality of road surface images generated based on the surrounding environment information, for each of the road surface images, a deformation representing a relative positional relationship between the road surface image and a reference virtual road surface image Estimating the amount, deforming the road surface image according to the estimated deformation amount, estimating a deformation amount representing a relative positional relationship between the road surface image and the virtual road surface image serving as an overall reference, and a plurality of Each of the road surface images is transformed into a deformation amount representing a relative positional relationship between the reference virtual road surface image and a deformation amount representing a relative positional relationship between the entire virtual road surface image serving as a reference. Deformed in response, deformed plural By integrating road surface images and registering them in map data, map data can be generated with high accuracy in consideration of the travel route of the vehicle even when a measurement error occurs in the absolute position of the vehicle. it can.

  Further, even if the position / motion measurement sensor is inexpensive and includes errors in absolute position measurement, highly accurate map data can be generated from data including errors in absolute position measurement.

  Since the absolute position measurement includes an error for each run, if a single map is generated by combining multiple trip data, a shift occurs and the map cannot be generated correctly. By estimating the relative positional relationship and considering the relative positional relationship between the traveling data, and integrating the whole, it is possible to reduce the positional error that occurs in each traveling and to generate highly accurate map data without any deviation it can.

[Second Embodiment]
Next, a second embodiment of the present invention will be described. In addition, about the part which becomes the structure similar to 1st Embodiment, the same code | symbol is attached | subjected and description is abbreviate | omitted.

  The second embodiment is different from the first embodiment in that the segment traveling data is not classified and the relative positional relationship between the classifications is not estimated.

<Configuration of map generation apparatus 210 according to the second embodiment>
As shown in FIG. 12, the computer 216 of the map generation device 210 according to the second embodiment travels the vehicle based on the vehicle position, direction, and speed information measured by the position / motion measurement sensor 14. Based on the travel trajectory generating unit 160 that generates the trajectory, the surrounding image of the vehicle acquired by the external field observation sensor 12, and the travel trajectory of the vehicle generated by the travel trajectory generating unit 160, the travel trajectory along the vehicle A road data representing a road surface is generated, a travel data generating unit 162 that generates travel data representing a combination of the road surface image and the travel trajectory, and each of the travel data generated by the travel data generating unit 162 is represented by the position of the travel trajectory. Of the travel data generated by the travel data storage unit 164 stored in the travel database 166 and the travel data generation unit 162 And a travel database 166 in which each is stored, and for each point, a relative amount that estimates a deformation amount representing a relative positional relationship between a road surface image and a reference virtual road surface image for each of the travel data of the point. Each of the road surface images of the travel data stored in the relationship estimation unit 270 and the travel database 166 is deformed according to the deformation amount estimated for the road surface image by the relative relationship estimation unit 170, and a plurality of deformed road surface images are obtained. A travel data integration unit 274 that is integrated and registered in the map database 176 and a map database 176 that stores the map data registered by the travel data integration unit 274 are provided. The travel data integration unit 274 is an example of an integration unit.

  The relative relationship estimation unit 270, for each segment at each point, for each segment road surface image of the segment travel data, for each pair of the segment road surface image and the other segment road surface image, Similarly, a deformation amount representing the relative positional relationship between the pairs is calculated. Then, the relative relationship estimation unit 270 determines the relative relationship based on the deformation amount representing the relative positional relationship calculated for each of the pairs for each of the divided road surface images of the divided traveling data for each divided portion of each point. Similar to the estimation unit 170, a deformation amount representing a relative positional relationship between the section road image and the reference virtual road image is estimated.

  The travel data integration unit 274, for each section at each point, for each section road image of the section travel data, a virtual section road image serving as a reference estimated by the relative relationship estimation unit 170 for the section road surface image. Are deformed according to the deformation amount representing the relative positional relationship between the two and the plurality of deformed section road surface images are integrated and registered in the map database 176.

<Operation of Map Generating Device 210 according to Second Embodiment>
Next, the operation of the map generation device 210 according to the second embodiment will be described. When a vehicle equipped with the map generation device 210 travels, a surrounding image of the vehicle is acquired by the external observation sensor 12, and the position, direction, and speed information of the vehicle at each time is measured by the position / motion measurement sensor 14. The traveling locus generation unit 160 generates a traveling locus of the vehicle based on the position, direction, and speed information of the vehicle at each time measured by the position / motion measurement sensor 14. In addition, the travel data generation unit 162 follows the travel locus of the vehicle based on the surrounding image of the vehicle at each time acquired by the external field observation sensor 12 and the travel locus of the vehicle generated by the travel locus generation unit 160. A road surface image representing the road surface is generated, and absolute coordinates are given to the road surface image in combination with the travel locus. Then, the travel data generation unit 162 generates travel data representing the road surface image to which the absolute coordinates are given by combining the road surface image and the travel locus. Next, the travel data accumulation unit 164 stores each of the travel data generated by the travel data generation unit 162 in the travel database 166.

  When a plurality of travel data is stored in the travel database 166, the map generation apparatus 10 executes a map generation processing routine shown in FIG. 13 for each point at a predetermined timing.

  In step S502, the relative relationship estimation unit 270 performs, for each segment of the target point, for each segment road surface image of the segment traveling data, for each pair of the segment road surface image and the other segment road surface image. Similar to step S308 described in the first embodiment, a deformation amount representing the relative positional relationship between the pairs is calculated. Then, the relative relationship estimation unit 170 performs the above-described calculation based on the deformation amount representing the relative positional relationship calculated for each of the pairs for each of the section road surface images of the section travel data for each section of the target point. Similar to step S312 described in the first embodiment, a deformation amount representing a relative positional relationship between the section road surface image and a reference virtual road surface image is estimated. Step S502 is realized by the relative relationship estimation processing routine shown in FIG.

  In step S508, the travel data integration unit 274, for each section of the target point, for each section road surface image of the section travel data, the section road surface image estimated in step S502 and a virtual reference as a reference. Deformation is performed according to the amount of deformation representing the relative positional relationship with the road surface image, and a plurality of deformed section road surface images are integrated.

  In addition, about the other structure and effect | action of the map production | generation apparatus 210 which concern on 2nd Embodiment, since it is the same as that of 1st Embodiment, description is abbreviate | omitted.

  As described above, according to the map generation device according to the present embodiment, a plurality of road surface images representing road surfaces registered along the traveling locus of the vehicle, and obtained from sensors mounted on the vehicle. Based on a database that stores a plurality of road surface images generated based on the surrounding environment information, for each corresponding road surface image, relative to each pair of the road surface image and another road surface image A deformation amount representing a positional relationship is calculated, and a relative positional relationship between the road surface image and a reference virtual road surface image is calculated based on the deformation amount representing the relative positional relationship calculated for each pair. The deformation amount to be expressed is estimated, and each of the plurality of road surface images is deformed in accordance with the deformation amount representing the relative positional relationship between the reference virtual road surface image and the estimated road surface image. The road image of To, by registering in the map data, even when the measurement error for the absolute position of the vehicle has occurred, it is possible to accurately generate the map data.

[Third Embodiment]
Next, a third embodiment of the present invention will be described. In addition, about the part which becomes the same structure as 1st or 2nd embodiment, the same code | symbol is attached | subjected and description is abbreviate | omitted.

  The map generating apparatus according to the third embodiment generates a road surface image based on three-dimensional point cloud data measured by a laser radar, and the point at which map data is generated is the first or second embodiment. Is different.

<Configuration of map generation apparatus 310 according to the third embodiment>
As shown in FIG. 15, a map generation apparatus 310 according to the third embodiment of the present invention includes an external field observation sensor 312 that measures a surrounding environment of a vehicle, and a position / motion that measures position information and motion information of the vehicle. A measurement sensor 14, a computer 316 that generates map data based on the surrounding environment information acquired by the external field observation sensor 312 and the vehicle position information and movement information measured by the position / movement measurement sensor 14. ing.

  In the third embodiment, a case where a laser radar is used as the external field observation sensor 312 will be described as an example. The external field observation sensor 312 acquires three-dimensional point cloud data around the vehicle measured by the laser radar as vehicle surrounding environment information.

  The computer 316 includes a CPU, a ROM that stores a program for realizing a map generation processing routine, which will be described later, a RAM that temporarily stores data, and a storage device such as an HDD. It is configured.

  As shown in FIG. 15, the computer 316 includes the vehicle trajectory generation unit 160, vehicle position, orientation, and speed information measured by the position / motion measurement sensor 14, and the vehicle periphery acquired by the external field observation sensor 312. A road surface image representing the road surface along the vehicle travel path is generated based on the three-dimensional point cloud data of the vehicle and the vehicle travel path generated by the travel path generation unit 160, and a combination of the road surface image and the travel path Data generation unit 362 for generating travel data representing the travel data, travel data storage unit 164, travel database 166, travel data classification unit 168, relative relationship estimation unit 170, overall relationship estimation unit 172, and travel data integration A unit 174 and a map database 176 are provided.

  As illustrated in FIG. 16, the travel data generation unit 362 includes a three-dimensional object determination unit 364, a grouping unit 366, a moving object determination unit 368, and a two-dimensional image generation unit 370.

  The three-dimensional object determination unit 364 performs three-dimensional object determination on the three-dimensional point cloud data based on the three-dimensional point cloud data around the vehicle acquired by the external field observation sensor 312, and converts the three-dimensional point cloud data to the road surface. Are classified into three-dimensional point group data representing a three-dimensional object and three-dimensional point group data representing a three-dimensional object. A local height distribution is used to determine whether or not each point of the three-dimensional point cloud data is a solid object.

  FIG. 17 shows an example of a method for estimating the local height. Specifically, the three-dimensional object determination unit 364 performs three-dimensional analysis on a two-dimensional grid partitioned by a predetermined size, as shown in FIG. 17, based on the three-dimensional point cloud data acquired by the external field observation sensor 312. Vote for each point in the point cloud data. Then, the three-dimensional object determination unit 364 calculates the distribution of the height of the voted three-dimensional points for each of the two-dimensional grids, and if the calculated height is equal to or greater than a predetermined threshold value. For example, the three-dimensional point voted on the grid is determined to be a three-dimensional object. On the other hand, if the calculated height is smaller than a predetermined threshold value, the three-dimensional object determination unit 364 determines that the three-dimensional point voted on the grid is a road surface point.

  The grouping unit 366 performs grouping on the three-dimensional point group data determined to be a point of the three-dimensional object by the three-dimensional object determination unit 364, and classifies the three-dimensional point group data into a plurality of groups. Specifically, the grouping unit 366 has predetermined distances in the three-dimensional space for each of the three-dimensional points of the three-dimensional point cloud data determined by the three-dimensional object determination unit 364 as the points of the three-dimensional object. The points less than the value are grouped together as belonging to the same group, and the three-dimensional point cloud data is classified into a plurality of groups.

  The moving object determination unit 368 is based on the vehicle position, direction, and speed information measured by the position / motion measurement sensor 14 and the plurality of groups of the three-dimensional point cloud data classified by the grouping unit 366. Track groups across time. Then, the moving object determination unit 368 compares the moving speed and moving direction of the group with the vehicle orientation and speed information measured by the position / motion measurement sensor 14, and moves the group different from the movement of the own vehicle. Judge as an object.

  For example, the moving object determination unit 368 calculates, for each of the plurality of groups of the three-dimensional point cloud data, the difference between the speed information of the host vehicle measured by the position / motion measurement sensor 14 and the moving speed of the group, A movement that moves in the same direction as the traveling direction of the host vehicle when the calculated difference is equal to or smaller than a predetermined first threshold and the traveling direction of the group is the same as the traveling direction of the host vehicle. It is determined that the object is a three-dimensional object.

  Further, the moving object determination unit 368 calculates, for each of the plurality of groups of the three-dimensional point cloud data, a difference between the speed information of the host vehicle measured by the position / motion measurement sensor 14 and the moving speed of the group. Movement that moves in a direction opposite to the traveling direction of the host vehicle when the difference is equal to or greater than a predetermined second threshold and the moving direction of the group is opposite to the traveling direction of the host vehicle. It is determined that the object is a three-dimensional object.

  Further, the moving object determination unit 368 calculates, for each of the plurality of groups of the three-dimensional point cloud data, a difference between the speed information of the host vehicle measured by the position / motion measurement sensor 14 and the moving speed of the group. If the calculated difference is less than a predetermined second threshold and the moving direction of the group is opposite to the traveling direction of the host vehicle, it is determined that the object is a stationary solid object.

  The two-dimensional image generation unit 370 generates a road surface image based on the three-dimensional point cloud data of the group determined to be a stationary solid object by the moving object determination unit 368. Specifically, the two-dimensional image generation unit 370 votes each point of the three-dimensional point cloud data of the group determined to be a stationary solid object by the moving object determination unit 368 to the two-dimensional grid, thereby locating the road surface image. Is generated.

  FIG. 18 shows an example of a road surface image generated by the two-dimensional image generation unit 370. For example, as shown in FIG. 18A, the two-dimensional image generation unit 370 can generate an edge point image in which only the points of the stationary solid object of the three-dimensional point group data are plotted as a road surface image. Further, as illustrated in FIG. 18B, the two-dimensional image generation unit 370 can generate, as a road surface image, an image in which the height or reflection intensity of a stationary three-dimensional object of the three-dimensional point cloud data is a pixel value. .

  In addition, when estimating the deformation | transformation amount showing a relative positional relationship from the road surface image which the two-dimensional image generation part 370 produced | generated, similarly to the said 1st or 2nd embodiment using the camera for the external field observation sensor, Cost calculation based on the matching cost of the above equation (1) can be used.

  In addition, about the other structure and effect | action of the map production | generation apparatus 310 which concern on 3rd Embodiment, since it is the same as that of 1st Embodiment, description is abbreviate | omitted.

  As described above, according to the map generation device according to the present embodiment, a plurality of road surface images representing road surfaces registered along the traveling locus of the vehicle, and obtained from sensors mounted on the vehicle. Between each pair of the road surface image and another road surface image for each corresponding road surface image based on a database in which a plurality of road surface images generated based on the three-dimensional point cloud data are accumulated. The relative position between the road surface image and the reference virtual road surface image is calculated based on the deformation amount representing the relative positional relationship calculated for each of the pairs. The amount of deformation representing the relationship is estimated, and each of the plurality of road surface images is deformed according to the amount of deformation representing the relative positional relationship between the reference virtual road surface image estimated for the road surface image, and deformed. Multiple road images Integrated, by registering in the map data, even when the measurement error for the absolute position of the vehicle has occurred, it is possible to accurately generate the map data.

  Note that the third embodiment may be applied to the second embodiment.

[Fourth Embodiment]
Next, a fourth embodiment of the present invention will be described. In addition, about the part which becomes the structure similar to the 1st-3rd embodiment, the same code | symbol is attached | subjected and description is abbreviate | omitted.

  The map generation apparatus according to the fourth embodiment is different from the first to third embodiments in that map data is generated from three-dimensional point cloud data measured by a laser radar.

<Configuration of Map Generating Device 410 According to Fourth Embodiment>
As shown in FIG. 19, a map generation apparatus 410 according to the fourth embodiment of the present invention includes an external field observation sensor 312 that measures a surrounding environment of a vehicle, and a position / motion that measures vehicle position information and motion information. A measurement sensor 14, a computer 416 that generates map data based on the surrounding environment information acquired by the external field observation sensor 312 and the vehicle position information and motion information measured by the position / motion measurement sensor 14. ing.

  In the fourth embodiment, as in the third embodiment, a case where a laser radar is used as the external field observation sensor 312 will be described as an example. The external field observation sensor 312 acquires three-dimensional point cloud data around the vehicle measured by the laser radar as vehicle surrounding environment information.

  The computer 416 includes a CPU, a ROM that stores a program for realizing a map generation processing routine described later, a RAM that temporarily stores data, and a storage device such as an HDD. It is configured.

  As shown in FIG. 19, the computer 416 includes the vehicle trajectory generation unit 160, vehicle position, direction, and speed information measured by the position / motion measurement sensor 14, and the vehicle periphery acquired by the external field observation sensor 312. Based on the three-dimensional point cloud data of the vehicle and the traveling locus of the vehicle generated by the traveling locus generation unit 160, the three-dimensional point cloud data and the traveling locus are generated along the traveling locus of the vehicle. Data generating unit 462 that generates driving data representing a combination of the driving data, and a driving data accumulating unit that stores each of the driving data generated by the driving data generating unit 462 in the driving database 466 in association with the position of the driving locus. 164, a travel database 466 in which each of the travel data generated by the travel data generation unit 462 is stored, The travel data classifying unit 468 classifies each of the travel data stored in the travel database 466 according to the position of the road surface represented by the three-dimensional point cloud data, and for each point, the classification A relative relationship estimating unit 470 that estimates a deformation amount representing a relative positional relationship between the three-dimensional point cloud data and the reference virtual three-dimensional point cloud data, and An overall relationship estimation unit 472 that estimates a deformation amount representing a relative positional relationship between the 3D point cloud data and the virtual 3D point cloud data serving as the overall reference for each of the travel data of the point; For each of the three-dimensional point cloud data of the traveling data stored in the traveling database 466, the deformation amount estimated for the three-dimensional point group data by the relative relationship estimating unit 470 and the overall relationship estimating unit 4 A travel data integration unit 474 that integrates a plurality of deformed three-dimensional point cloud data and registers them in the map database 476; A map database 476 in which the map data registered by the unit 474 is stored. The three-dimensional point cloud data is an example of runway data.

  As illustrated in FIG. 20, the travel data generation unit 462 includes a three-dimensional object determination unit 364, a grouping unit 366, a moving object determination unit 368, and a data generation unit 464.

  The data generation unit 464 includes three-dimensional point cloud data of a group determined to be a stationary solid object by the moving object determination unit 368, three-dimensional point group data determined to represent a road surface by the three-dimensional object determination unit 364, Based on the travel locus of the vehicle generated by the travel locus generation unit 160, 3D point cloud data is generated along the travel locus of the vehicle, and travel data representing a combination of the 3D point cloud data and the travel locus is generated. Generate.

  The travel data classifying unit 468 classifies each of the travel data corresponding to the point stored in the travel database 166 according to the position of the road surface represented by the three-dimensional point group data. Specifically, the travel data classifying unit 468 divides, for each point, the three-dimensional point cloud data of each travel data stored in the travel database 166 for each predetermined absolute position section. Then, section traveling data including section three-dimensional point cloud data divided for each section is generated. Next, the travel data classification unit 468 classifies the segment travel data of the same section according to the travel route of the vehicle obtained from the travel locus included in the travel data corresponding to the segment travel data.

  The relative relationship estimation unit 470 performs, for each classification of each section at each point, for each of the section three-dimensional point cloud data of the section traveling data classified into the classification by the traveling data classification unit 468. For each pair of group data and other three-dimensional point group data, a deformation amount representing the relative positional relationship between the pairs is calculated.

  FIG. 21 and FIG. 22 show an example of a method for obtaining correspondence at a plurality of three-dimensional points for each of the divided three-dimensional point group data belonging to the same classification. For example, the relative relationship estimation unit 470 obtains a partial point as shown in FIG. 21A from the partitioned 3D point cloud data for each of the partitioned 3D point cloud data, as a method for obtaining correspondence at a plurality of 3D points. A group is extracted, and for each partial point group, matching is performed with a partial point group of other divided three-dimensional point group data as shown in FIG.

Specifically, as shown in FIG. 22, the relative relationship estimation unit 470 performs, for each of the two partial point groups shown in FIG. The closest point is obtained from the partial point group of the other other three-dimensional point group data.
Then, the relative relationship estimation unit 470 calculates, as a matching cost, the sum of the distances between the point and the point closest to the point for each point of the partial point group of one piece of the three-dimensional point group data. The matching cost is calculated by the following equation, where A and B are the point groups of the partial areas of the two section travel data and the deformation amount is x.

d _near (v, B) represents the distance between the point v of the point group A and the point (wεB) of the point group B closest to the point v, and is calculated by the above formula. Note that | · | ₂ represents the L2 norm. In order to deal with outliers, an upper limit value of d _near (v, B) may be set.

  Then, the relative relationship estimation unit 470 estimates the deformation amount x that minimizes the matching cost calculated according to the above formula for each of the three-dimensional point group data belonging to the same classification.

  In addition, the relative relationship estimation unit 470 determines, for each piece of block three-dimensional point cloud data, a partial point group of the block three-dimensional point cloud data that is a reference virtual block three-dimensional point estimated for the partial point group. Deformation is performed according to the deformation amount representing the relative positional relationship with the group data.

  The overall relationship estimation unit 472 for each segment of each point, for each category, the relative relationship estimation unit 470 for each of the segment three-dimensional point cloud data of the segment traveling data classified into the classification by the traveling data classification unit 468. For each of the pairs of the section 3D point cloud data deformed by the above and the other section 3D point cloud data classified and deformed in each classification, a deformation amount representing a relative positional relationship between the pairs is set. calculate.

  For each section of each point, the traveling data integration unit 474 performs, for each classification, the relative relationship estimation unit 470 for each section three-dimensional point cloud data of the section traveling data classified into the classification. A deformation amount that is estimated for the group data and represents a relative positional relationship with the virtual section 3D point cloud data serving as a reference, and is estimated for the section 3D point group data by the overall relationship estimation unit 472. A map is obtained as a map data by deforming according to a deformation amount representing a relative positional relationship with the virtual section 3D point cloud data as a whole reference, and integrating the plurality of section 3D point group data which have been deformed. Register in the database 476.

<Operation of Map Generating Device 410 According to Fourth Embodiment>
Next, the operation of the map generation device 410 according to the fourth embodiment will be described. A vehicle on which the map generation device 410 is mounted travels, the 3D point cloud data around the vehicle is acquired by the external field observation sensor 312, and the position / direction / speed information of the vehicle at each time is measured by the position / motion measurement sensor 14. At this time, the travel locus generation unit 160 generates a travel locus of the vehicle based on the vehicle position, direction, and speed information measured at each time by the position / motion measurement sensor 14. Further, the travel data generation unit 462 is based on the three-dimensional point cloud data around the vehicle at each time acquired by the external field observation sensor 312 and the vehicle travel trajectory generated by the travel trajectory generation unit 160. The three-dimensional point cloud data along the traveling locus is generated and combined with the traveling locus to give absolute coordinates to the three-dimensional point cloud data. Then, the travel data generation unit 462 generates travel data representing the 3D point cloud data to which the absolute coordinates are assigned by combining the 3D point cloud data and the travel locus. Next, the travel data accumulation unit 164 stores each travel data generated by the travel data generation unit 462 in the travel database 466.

  When a plurality of travel data is stored in the travel database 466, the map generation device 410 executes a map generation processing routine shown in FIG. 23 for each point at a predetermined timing.

  In step S100, the travel data classification unit 468 classifies each of the travel data corresponding to the target point stored in the travel database 466 according to the position of the road surface represented by the three-dimensional point cloud data. Step S100 is realized by the travel data classification process routine shown in FIG.

  Next, in step S602, the relative relationship estimation unit 470 for each section of the target point, for each section three-dimensional point cloud data of the section traveling data, for each classification output in step S100. For each pair of the three-dimensional point group data and the other three-dimensional point group data, a deformation amount representing the relative positional relationship between the pairs is calculated. Then, the relative relationship estimation unit 470 calculates, for each classification, the relative positional relationship calculated for each of the pairs with respect to each of the three-dimensional point cloud data of the section traveling data classified into the classification. Is estimated based on the deformation amount indicating the relative positional relationship between the three-dimensional point cloud data and the reference virtual three-dimensional point cloud data. Step S602 is realized by a relative relationship estimation processing routine shown in FIG.

  Next, in step S604, the relative relationship estimation unit 470 estimated the partial point cloud of the block 3D point cloud data for each block 3D point cloud data and the partial point cloud in step S602. Deformation is performed according to a deformation amount representing a relative positional relationship with a partial point group of virtual section three-dimensional point group data serving as a reference.

  In step S606, the overall relationship estimation unit 472 transforms each segment three-dimensional point cloud data of the segment travel data in step S604 for each segment of the target point for each classification output in step S100. For each pair of the divided three-dimensional point cloud data and each of the other divided three-dimensional point cloud data classified into a plurality of classifications, a deformation amount representing a relative positional relationship between the pairs is set. calculate. Then, the overall relationship estimation unit 472 calculates, for each division of the target point, for each of the pairs for each classification, for each of the three-dimensional point cloud data of the division traveling data classified into the classification. Based on the deformation amount representing the relative positional relationship, the deformation amount representing the relative positional relationship between the section three-dimensional point cloud data and the virtual section three-dimensional point group data serving as the entire reference is estimated. Step S606 is realized by the overall relationship estimation processing routine shown in FIG.

  Next, in step S608, the travel data integration unit 474 for each section of the target point, for each classification, for each of the section three-dimensional point cloud data of the section travel data classified into the classification in step S602. Deformation is performed according to the estimated deformation amount and the deformation amount estimated in step S606, and the deformed plurality of section three-dimensional point group data is integrated.

  In addition, about the other structure and effect | action of the map production | generation apparatus 410 which concern on 4th Embodiment, since it is the same as that of 1st or 3rd Embodiment, description is abbreviate | omitted.

  As described above, according to the map generation device according to the present embodiment, a plurality of three-dimensional point cloud data representing road surfaces registered along the traveling locus of the vehicle, and sensors mounted on the vehicle For each of the corresponding 3D point cloud data based on a database storing a plurality of 3D point cloud data generated based on the surrounding environment information obtained from the 3D point cloud data and other 3D point data For each pair with the group data, a deformation amount representing the relative positional relationship between the pair is calculated, and based on the deformation amount representing the relative positional relationship calculated for each pair, the three-dimensional point group data And a deformation amount representing a relative positional relationship between the reference virtual three-dimensional point cloud data, and each of the plurality of three-dimensional point cloud data is estimated for the three-dimensional point cloud data. A virtual 3D point cloud The data is deformed according to the deformation amount representing the relative positional relationship with the vehicle, and a plurality of deformed three-dimensional point cloud data is integrated and registered in the map data, thereby generating a measurement error for the absolute position of the vehicle. Map data can be generated with high accuracy.

  Note that the fourth embodiment may be applied to the second embodiment.

[Fifth Embodiment]
Next, a fifth embodiment of the present invention will be described. In addition, about the part which becomes the same structure as 1st or 2nd embodiment, the same code | symbol is attached | subjected and description is abbreviate | omitted.

  In the map generation system according to the fifth embodiment, the map generation device is not mounted on the vehicle, and the map data is generated based on vehicle information transmitted from a plurality of vehicles. The form is different.

<Configuration of Map Generation System 500 According to Fifth Embodiment>
As shown in FIG. 26, a map generation system 500 according to the fifth embodiment of the present invention receives an in-vehicle device 502 provided in a vehicle and vehicle information transmitted from the in-vehicle device 502, and receives vehicle information. And a map generation device 510 for generating map data according to the above. The in-vehicle device 502 and the map generation device 510 are connected via a network 509 such as the Internet.

[In-vehicle device 502]
The in-vehicle device 502 includes an external observation sensor 12 that measures the surrounding environment of the vehicle, a position / motion measurement sensor 14 that measures position information and movement information of the vehicle, the surrounding environment information acquired by the external observation sensor 12, and a position. A computer 504 that transmits vehicle position information and motion information measured by the motion measurement sensor 14 to the map generation device 510 as vehicle information is provided. The in-vehicle device 502 is mounted on the vehicle.

  The computer 504 includes a CPU, a RAM, and a ROM that stores a program for executing each processing routine, and is functionally configured as follows. The computer 504 includes an information acquisition unit 506 that acquires, as vehicle information, ambient environment information acquired by the external field observation sensor 12 and vehicle position information and motion information measured by the position / motion measurement sensor 14. A communication unit 508 that transmits the vehicle information acquired by the unit 506 to the map generation device 510.

  The information acquisition unit 506 acquires, as vehicle information, the surrounding environment information at each time acquired by the external field observation sensor 12 and the position information and motion information of the vehicle at each time measured by the position / motion measurement sensor 14. .

  The communication unit 508 transmits the vehicle information acquired by the information acquisition unit 506 to the map generation device 510.

[Map generator 510]
The map generation device 510 is configured by a server including a CPU, a ROM that stores a program for realizing each processing routine described later, a RAM that temporarily stores data, a memory as a storage unit, a network interface, and the like. Functionally, it receives vehicle information transmitted from the in-vehicle device 502, registers the received vehicle information in the vehicle information database 514, a vehicle information database 514 in which the vehicle information is stored, Based on the vehicle information stored in the vehicle information database 514, a travel locus generation unit 160 that generates a travel locus of the vehicle, the vehicle information stored in the vehicle information database 514, and the vehicle generated by the travel locus generation unit 160 A road surface image representing the road surface along the vehicle's travel locus is generated based on the travel locus of A travel data generation unit 162 that generates travel data representing a combination of a plane image and a travel locus, and each of the travel data generated by the travel data generation unit 162 is stored in the travel database 166 in association with the position of the travel locus. A travel data storage unit 164 that stores the travel data generated by the travel data generation unit 162, and each travel data stored in the travel database 166 for each point. A travel data classifying unit 168 for classifying according to the position of the road surface represented by the road surface image, and for each point, for each of the travel data belonging to the classification, a virtual road surface image that serves as a reference for the road surface image A relative relationship estimation unit 170 that estimates the amount of deformation representing the relative positional relationship between the , An overall relationship estimation unit 172 that estimates a deformation amount representing a relative positional relationship between a road surface image and a virtual road surface image serving as an overall reference, and a road image of travel data stored in the travel database 166 Are transformed according to the deformation amount estimated for the road surface image by the relative relationship estimation unit 170 and the deformation amount estimated for the road surface image by the overall relationship estimation unit 172, and a plurality of deformed road surface images are integrated. The travel data integration unit 174 registered in the map database 176 and the map database 176 in which the map data registered by the travel data integration unit 174 are stored are provided. The travel data integration unit 174 is an example of an integration unit, and the travel data classification unit 168 is an example of a classification unit.

  Communication unit 512 receives vehicle information transmitted from in-vehicle device 502 of the vehicle. In addition, the communication unit 512 stores the received vehicle information in the vehicle information database 514.

  Each vehicle information received by the communication unit 512 is stored in the vehicle information database 514. The vehicle information represents the surrounding environment information at each time acquired by the external field observation sensor 12 of the vehicle, and the position information and motion information at each time of the vehicle measured by the vehicle position / motion measurement sensor 14.

  For each piece of vehicle information stored in the vehicle information database 514, the travel locus generation unit 160 generates a vehicle travel locus based on the vehicle position, direction, and speed information at each time in the vehicle information.

  For each piece of vehicle information stored in the vehicle information database 514, the travel data generation unit 162 is based on the surrounding image of the vehicle at each time of the vehicle information and the vehicle travel locus generated by the travel locus generation unit 160. Then, a road surface image representing a road surface along the traveling locus of the vehicle is generated, and absolute coordinates are given to the road surface image in combination with the traveling locus.

  In addition, about the other structure and effect | action of the map generation system 500 which concern on 3rd Embodiment, since it is the same as that of 1st Embodiment, description is abbreviate | omitted.

  As described above, according to the map generation device according to the present embodiment, based on the vehicle information obtained from a plurality of vehicles, even when a measurement error occurs for the absolute position of the vehicle, Map data can be generated with high accuracy.

    Note that the fifth embodiment may be applied to the third or fourth embodiment.

  Note that the present invention is not limited to the embodiment described with reference to the drawings, and various modifications can be applied without departing from the scope of the invention.

  The program of the present invention may be provided by being stored in a storage medium.

10, 210, 310, 410, 510 Map generation device 12, 312 External field observation sensor 14 Position / motion measurement sensor 16, 216, 316, 416, 504 Computer 160 Travel locus generation unit 162, 362, 462 Travel data generation unit 164 Travel Data storage unit 166, 466 Travel database 168, 468 Travel data classification unit 170, 270, 470 Relative relationship estimation unit 172, 472 Overall relationship estimation unit 174, 274, 474 Travel data integration unit 176, 476 Map database 364 Solid object determination unit 366 Grouping unit 368 Moving object determination unit 370 Two-dimensional image generation unit 464 Data generation unit 500 Map generation system 502 In-vehicle device 506 Information acquisition unit 508, 512 Communication unit 509 Network 514 Vehicle information database

Claims (9)

Based on a database in which a plurality of track data registered along a travel locus of a moving body and a plurality of track data generated based on surrounding environment information obtained from sensors mounted on the mobile body are accumulated. For each of the corresponding track data, a deformation amount representing a relative positional relationship between the pair is calculated for each pair of the track data and other track data, and is calculated for each of the pairs. It was based on the deformation amount representing the relative positional relationship, relative to estimate said road data, the amount of deformation that represents the relative positional relationship between the reference and the Do that run path data calculated in accordance with a plurality of track data Relationship estimation means;
Each of the plurality of road data, the relative relationship estimated by the estimating means for said track data, is deformed according to the deformation quantity representing the relative positional relationship between the reference and Do that run path data, deformed Integration means for integrating the plurality of track data and registering the map data;
A map generation device including :
The relative relationship estimation means averages the deformation amount representing the relative positional relationship calculated for each of the pairs, and estimates the deformation amount representing the relative positional relationship with the reference track data. ,
Map generator . Based on a database in which a plurality of track data registered along a travel locus of a moving body and a plurality of track data generated based on surrounding environment information obtained from sensors mounted on the mobile body are accumulated. For each of the corresponding track data, a deformation amount representing a relative positional relationship between the pair is calculated for each pair of the track data and other track data, and is calculated for each of the pairs. It was based on the deformation amount representing the relative positional relationship, relative to estimate said road data, the amount of deformation that represents the relative positional relationship between the reference and the Do that run path data calculated in accordance with a plurality of track data Relationship estimation means;
Each of the plurality of road data, the relative relationship estimated by the estimating means for said track data, is deformed according to the deformation quantity representing the relative positional relationship between the reference and Do that run path data, deformed An integrated means for integrating the plurality of runway data and registering the map data in map data ,
A classifying means and an overall relationship estimating means;
The classification means classifies each of the corresponding track data according to the position of the road surface represented by the track data,
The relative relationship estimation means, for each class of the track data classified by the classification means, for each of the track data belonging to the classification, for each pair of the track data and other track data, For each partial image of the runway data, a deformation amount representing the relative positional relationship of the partial images between the pairs is calculated, and the deformation amount representing the relative positional relationship of the partial images calculated for each of the pairs. On the basis of the partial image of the road data, the amount of deformation representing a relative positional relationship between the road data and the reference road data is estimated, and the road data is converted into the estimated amount of deformation. Deform according to
The overall relationship estimation means, for each of the track data, for each classification, for each pair of the modified track data and the other track data that is classified in the classification and the other track data. A deformation amount representing a relative positional relationship between the pair is calculated, and calculated based on the road data and a plurality of road data based on the deformation amount representing the relative positional relationship calculated for each of the pairs. Estimate the amount of deformation that represents the relative positional relationship with the entire reference runway data,
The integration unit is configured to determine, for each partial image, between the partial image of each of the plurality of track data and the reference track data estimated for the partial image of the track data by the relative relationship estimation unit. And a deformation amount representing a relative positional relationship between the overall road estimation data estimated by the overall relationship estimating means and the overall reference road data, and deformed. Integrating the partial images of the plurality of runway data and registering it in map data;
Map generator. The relative relationship estimation means calculates, for each partial image of the track data, a deformation amount representing a relative positional relationship of the partial images between the pairs for each pair of the track data and other track data. , based on the deformation amount representing the relative positional relationship between the partial images calculated for each of said pairs, for each partial image of the runway data, and the road data, run path data and that Do with the reference Estimate the amount of deformation representing the relative positional relationship between
The integration means, for each of the partial images, the partial image of each of the plurality of road data, the estimated by the relative relationship estimating means for partial image of the runway data, run path data and that Do with the reference The map generation device according to claim 1, wherein the map generation device is deformed according to a deformation amount representing a relative positional relationship between the two, and the partial images of the plurality of deformed runway data are integrated and registered in map data. The overall relationship estimating means, in response to said number of the track data belonging to the classified the classification by the classification unit, estimating a deformation amount that represents the relative positional relationship between the Do that run channel data and reference of the whole The map generation device according to claim 2 . The vehicle further includes travel locus generating means for generating a travel locus of the mobile body based on the position information and motion information of the mobile body acquired by the vehicle state detection means for acquiring the position information and motion information of the mobile body,
The map generation according to any one of claims 1 to 4 , wherein the database accumulates the plurality of travel route data registered along the travel trajectory of the moving body generated by the travel trajectory generation means. apparatus. The traveling road data is a road surface image representing a road surface registered along a traveling locus of the moving body, and a road surface image generated based on surrounding environment information obtained from a sensor mounted on the moving body. The map generation device according to any one of claims 1 to 5 . The travel path data is 3D point cloud data registered along the travel trajectory of the mobile object, and is generated based on the surrounding environment information obtained from a sensor mounted on the mobile object. It is group data, The map production | generation apparatus of any one of Claims 1-5 . A computer that stores a plurality of track data registered along the track of a moving object and that stores a plurality of track data generated based on surrounding environment information obtained from sensors mounted on the moving object. Based on each of the corresponding track data, for each pair of the track data and other track data, a deformation amount representing a relative positional relationship between the pair is calculated, and for each of the pairs based on the deformation amount representing the calculated the relative positional relationship, estimation and the road data, the amount of deformation that represents the relative positional relationship between the Do that run path data and calculated reference in accordance with the plurality of track data relative relationship estimating means, and each of the plurality of road data was estimated for the road data by the relative relationship estimation means, the phase between the reference and Do that run path data Depending on the deformation amount representing the positional relationship is deformed, by integrating the plurality of track data modification, a program for functioning as an integrated unit to be registered in the map data,
The relative relationship estimation means averages the deformation amount representing the relative positional relationship calculated for each of the pairs, and estimates the deformation amount representing the relative positional relationship with the reference track data. ,
Program . Based on a database in which a plurality of track data registered along a travel locus of a moving body and a plurality of track data generated based on surrounding environment information obtained from sensors mounted on the mobile body are accumulated. For each of the corresponding track data, a deformation amount representing a relative positional relationship between the pair is calculated for each pair of the track data and other track data, and is calculated for each of the pairs. It was based on the deformation amount representing the relative positional relationship, relative to estimate said road data, the amount of deformation that represents the relative positional relationship between the reference and the Do that run path data calculated in accordance with a plurality of track data relationship estimation hand stage,
Each of the plurality of road data, the relative relationship estimated by the estimating means for said track data, is deformed according to the deformation quantity representing the relative positional relationship between the reference and Do that run path data, deformed Integration means for integrating the plurality of track data and registering the map data ;
Classification means, and
A program for functioning as an overall relationship estimation means ,
The classification means classifies each of the corresponding track data according to the position of the road surface represented by the track data,
The relative relationship estimation means, for each class of the track data classified by the classification means, for each of the track data belonging to the classification, for each pair of the track data and other track data, For each partial image of the runway data, a deformation amount representing the relative positional relationship of the partial images between the pairs is calculated, and the deformation amount representing the relative positional relationship of the partial images calculated for each of the pairs. On the basis of the partial image of the road data, the amount of deformation representing a relative positional relationship between the road data and the reference road data is estimated, and the road data is converted into the estimated amount of deformation. Deform according to
The overall relationship estimation means, for each of the track data, for each classification, for each pair of the modified track data and the other track data that is classified in the classification and the other track data. A deformation amount representing a relative positional relationship between the pair is calculated, and calculated based on the road data and a plurality of road data based on the deformation amount representing the relative positional relationship calculated for each of the pairs. Estimate the amount of deformation that represents the relative positional relationship with the entire reference runway data,
The integration unit is configured to determine, for each partial image, between the partial image of each of the plurality of track data and the reference track data estimated for the partial image of the track data by the relative relationship estimation unit. And a deformation amount representing a relative positional relationship between the overall road estimation data estimated by the overall relationship estimating means and the overall reference road data, and deformed. Integrating the partial images of the plurality of runway data and registering it in map data;
Program .