Apparatus and method for generating feature data

Abstract

To provide a generation device of feature data, which can ensure the accuracy of a feature database.SOLUTION: A feature data generation device 300 of a feature data generation system 100 generates a feature characteristic vector as feature data indicating planimetric features from an edge pattern which is extracted from positional data of a plurality of measured points by a LIDAR technique around vehicles 200, and indicates boundaries of object features existing around the vehicles 200, and to which positions of an absolute coordinate system are added, so as to calculate a quality assurance index for assuring the quality of the generated feature characteristic vector.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a feature data generating device and a feature data generating method used for updating and generating a three-dimensional high-precision map. [Background technology]

[0002] Three-dimensional high-precision maps, which are spatial maps that record road lanes, guardrails, road signs, crosswalks, etc. at more accurate positions than conventional planar maps, are known as mobile driving environment description data used in advanced driver assistance systems (ADAS) and autonomous driving (AD). In addition, feature databases that provide feature data for three-dimensional high-precision maps are also known.

[0003] In addition, by utilizing the high-precision positioning service provided by the Quasi-Zenith Satellite System "Michibiki," which has recently been launched, it is now possible to reduce the error in the position of a moving object to a few centimeters. By using this high-precision positioning service together with a high-precision three-dimensional map, it is expected that autonomous driving will become more accurate.

[0004] In order to create a high-precision three-dimensional map, data on landmarks such as road edges, lanes, guardrails, road signs, and pedestrian crossings (hereinafter referred to as "feature data") must be collected in advance into a feature database. One method of collecting such feature data has been proposed, which involves capturing images of the surrounding area using a camera mounted on a moving object such as a car, and then analyzing the images to obtain the feature data.

[0005] For example, the system proposed by Mobileye uses a monocular camera to grasp the surrounding situation of a moving object and obtain feature data. However, a monocular camera does not provide sufficient distance measurement accuracy, making it difficult to obtain high-precision feature data worthy of constructing a high-precision 3D map.

[0006] A system has also been proposed that uses a compound-eye camera to grasp the surroundings of a moving object and obtain feature data, but this is more expensive than a monocular camera, and the position of each camera needs to be calibrated because distance measurement is performed by triangulation based on the positions of multiple cameras. Furthermore, cameras have the disadvantage of being vulnerable to environmental changes such as rain, fog, and the setting sun. Furthermore, when using a compound-eye camera, triangulation calculations need to be repeated based on multiple images obtained from each camera while the moving object is moving, which requires a huge amount of calculations, making it unrealistic to use it in practice as a tool to obtain feature data.

[0007] Furthermore, a method has been proposed in which data on features around a vehicle is collected sequentially by driving a vehicle equipped with a measurement system called a Mobile Mapping System (MMS) (see, for example, Patent Document 1), which is equipped with three-dimensional measuring devices such as cameras and laser scanners, and satellite positioning devices such as GPS. The laser scanner applies LIDAR (Laser Imaging Detection and Ranging) technology, which is a remote sensing technology that uses laser light.

[0008] LIDAR technology can obtain accurate distance position data to the measurement point using laser light without triangulation calculations. In addition, since LIDAR technology uses extremely short wavelengths, it can identify and detect smaller features of ground objects than RADAR technology, which uses radio waves. In addition, LIDAR technology does not require calibration of the position of each camera as in the case of using a compound eye camera, and since it does not use images, it is less susceptible to environmental changes such as rain and fog. Therefore, when a laser scanner using LIDAR technology is applied to a measurement system, it is possible to obtain more accurate ground object data more easily than when using a compound eye camera. [Prior art documents] [Patent documents]

[0009] [Patent Document 1] JP 2019-049466 A Summary of the Invention [Problem to be solved by the invention]

[0010] However, because the MMS laser scanner acquires the position data of all measurement points around the vehicle, acquiring feature data using the MMS requires processing of a huge amount of position data, which takes a lot of time and effort. Therefore, once feature data for a certain area has been acquired, it is almost never acquired again for that area, and even if the relevant feature, for example a road sign, is moved due to construction, the movement is not reflected in the feature database, and deterioration in the quality of the feature data and feature database over time is unavoidable.

[0011] In addition, when there are obstacles such as parked vehicles on the roadside, some areas may be hidden by the obstacles and cannot be measured. Furthermore, because conventional laser scanners perform scans at regular intervals by mechanical rotation, the target may get caught between adjacent laser beams, and the laser beam may not be irradiated on the target, resulting in the target not being recognized as a feature. Therefore, the features obtained by the MMS laser scanner do not necessarily reflect all real features, and the quality of the feature data cannot be guaranteed.

[0012] When acquiring feature data from position data of measurement points that represent the situation around the vehicle, for example, an operator judges whether a collection of position data represents a feature, and acquires polygon data or spline curve data of the collection that is judged to represent a feature as feature data. At this time, human errors such as operator's misjudgment may be reflected in the feature data, but as described above, feature data is rarely reacquired, so there is no opportunity to correct the feature data including the error in the feature database, and the quality of the feature data is not maintained.

[0013] Even if a feature database is created using feature data that does not maintain such quality, it is difficult to continue to guarantee the accuracy of the feature database, and even more so, the accuracy of three-dimensional high-precision maps that are updated or generated using the feature database.

[0014] An object of the present invention is to provide a feature data generating device and a feature data generating method that can ensure the accuracy of a high-precision three-dimensional map. [Means for solving the problem]

[0015] In order to achieve the above object, the feature data generating device of the present invention includes a feature data generating unit that generates feature data representing features from an edge pattern that indicates boundaries of features present around a moving body, the edge pattern being extracted from position data of a large number of measurement points around the moving body measured using LIDAR technology, and a quality assurance index calculating unit that calculates a quality assurance index that guarantees the quality of the generated feature data, and the quality assurance index calculating unit sets the quality assurance index lower as time passes since the generation of the feature data. and varying said quality assurance index in response to a method for obtaining the position of said feature in a terrestrial reference coordinate system. It is characterized by the above. Effect of the Invention

[0016] According to the present invention, the accuracy of a high-precision three-dimensional map can be guaranteed. [Brief description of the drawings]

[0017] [Figure 1] 1 is a block diagram showing an outline of a configuration of a feature data generating system according to an embodiment of the present invention; [Diagram 2] FIG. 2 is a diagram for explaining a database configuration of the feature data generating device of FIG. [Diagram 3] FIG. 2 is a diagram for explaining the configuration and functions of the vehicle in FIG. [Figure 4] FIG. 2 is a diagram for explaining extraction of an edge pattern by a laser scanner unit. [Diagram 5]10 is a diagram for explaining generation of feature description data by a feature description data generation unit; FIG. [Figure 6] 11 is a diagram for explaining matching of feature description data by a feature description data generating unit. FIG. [Figure 7] FIG. 1 is a diagram for explaining the concept of Mahalanobis distance. [Figure 8] 11 is a diagram for explaining corresponding feature discrimination by a class classification processing unit. FIG. [Figure 9] 11 is a diagram for explaining determination of a time change in a shape characteristic vector by a feature statistical data calculation unit. FIG. [Figure 10] 11 is a diagram for explaining the assignment of a lane ID and the like by a road feature assignment unit. FIG. [Figure 11] FIG. 2 is a diagram showing an example of a feature feature vector generated by a feature feature vector database generating unit; [Figure 12] 11 is a diagram for explaining generation of a quality assurance index by a quality assurance index calculation unit. FIG. [Figure 13] FIG. 2 is a diagram for explaining the configuration of a feature database. [Figure 14] 4 is a flowchart showing a method for generating feature data in the present embodiment. [Figure 15] 11 is a diagram for explaining an example of a method for setting a value of a quality assurance index of a feature feature vector. FIG. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0018] Hereinafter, embodiments of the present invention will be described with reference to the drawings.

[0019] Fig. 1 is a block diagram showing a schematic configuration of a feature data generation system (feature database update system) according to an embodiment of the present invention. In Fig. 1, the feature data generation system 100 includes at least one vehicle 200 (mobile body), a feature data generation device 300 consisting of, for example, a host computer, and a three-dimensional high-precision map updating device 400. The vehicle 200, the feature data generation device 300, and the three-dimensional high-precision map updating device 400 can transmit and receive data to and from each other wirelessly or via wires.

[0020] The feature data generating device 300 includes an SD map 301 (standard accuracy map), a feature description data generating unit 302, a feature feature data generating unit 303, an anomaly measurement value detecting unit 304, a class classification processing unit 305, a shape feature vector database generating unit 306, a feature statistical data calculating unit 307, a road feature assignment unit 308, a feature feature vector database generating unit 309, a crustal deformation correcting unit 310, a quality assurance index calculating unit 311, a position data remeasurement requesting unit 312, and a feature database generating unit 313. Each component of the feature data generating device 300 will be described in detail later. The feature description data generating unit 302, the feature feature data generating unit 303, the shape feature vector database generating unit 306, and the feature feature vector database generating unit 309 correspond to the feature data generating unit.

[0021] The three-dimensional high-precision map updating device 400 stores a three-dimensional high-precision map (not shown), and generates and updates features of the three-dimensional high-precision map using feature data generated by the feature data generating device 300. Note that the three-dimensional high-precision map updating device 400 may be integrated with the feature data generating device 300, and the integrated feature data generating device 300 may generate and update features of the three-dimensional high-precision map using feature data that it generates itself.

[0022] Fig. 2 is a diagram for explaining the database configuration of the feature data generation device 300 of Fig. 1. In Fig. 2, the feature data generation device 300 has a shape feature vector database 314, a feature feature vector database 315, and a feature database 316. The shape feature vector database 314 stores shape feature vectors generated by the feature feature data generation unit 303 and the shape feature vector database generation unit 306. The feature feature vector database 315 and the feature database 316 store feature feature vectors generated by the feature feature vector database generation unit 309. Each database of the feature data generation device 300 will be described in detail later.

[0023] Fig. 3 is a diagram for explaining the configuration and functions of the vehicle 200 in Fig. 1. In the feature data generation system 100 in Fig. 1, there are multiple vehicles 200, but since all of them have the same configuration, one vehicle 200 will be explained here. Note that the vehicle 200 is not limited to an automobile, and also includes, for example, a two-wheeled vehicle or an air vehicle (such as a drone).

[0024] The vehicle 200 has a laser scanner unit 201 that uses LIDAR technology, a self-position estimation unit 202 (an absolute coordinate system position assignment unit), a data transmission / reception unit 203, and a camera 209. The laser scanner unit 201 has a head 204, a scanner control unit 205, and an edge pattern extraction unit 206 (FIG. 3(A)).

[0025] 3(B), the laser scanner unit 201 irradiates laser light 207 from a head 204 to the vicinity of the vehicle including road signs 101, and receives reflected light 208 from a large number of surrounding measurement points to obtain the distance, direction, and elevation angle of each measurement point on the scanner reference coordinate system as position data. Since the acquisition of this position data continues while the vehicle 200 is traveling, the laser scanner unit 201 acquires a wide range of position data as a huge group of point data (point cloud) as the vehicle 200 travels.

[0026] The head 204 has a device that changes the direction of laser light irradiation, such as a MEMS (Micro Electro Mechanical Systems) device, and performs scanning of the periphery of the vehicle 200 with the laser light 207. The scanner control unit 205 controls the head 204 by software to realize various scanning patterns of the laser light 207. In particular, the laser light 207 is irradiated in a concentrated manner on a feature, and a large amount of position data is obtained from the feature and its surroundings. There are several types of scanning patterns for intensively irradiating the laser light 207. For example, a conical scan that draws a spiral with the laser light 207 is performed on a road sign 101 with a disk at the tip (FIG. 3(C)). In addition, a vertical scan is performed on a target 102 fixed on the ground, in which the laser light 207 is used to repeat vertical scans while shifting the laser light 207 slightly in the horizontal direction (FIG. 3(C)). This makes it possible to efficiently obtain a large amount of position data from the feature and its surroundings, and improves the accuracy of the generated feature data. Furthermore, since a large amount of unnecessary position data is not acquired from surrounding areas that are not of much value other than features, the efficiency of generating feature data can be improved.

[0027] The scanner control unit 205 intensively irradiates the laser light 207 on the feature identified in the image data of the surroundings captured by the camera 209 mounted on the vehicle 200. In order to smoothly perform the integration (fusion) of the image data and the position data described later, the laser light 207 may be intensively irradiated on the feature identified in the image data of the surroundings captured by the mounted camera 209 in this way to increase the position data of the feature, or the camera 209 may intensively image the feature identified in the position data acquired by the laser light 207 to improve the accuracy of the image data of the feature. In addition, the feature data generating device 300 may transmit the approximate position information of the feature included in the SD map 301 to the vehicle 200, and the scanner control unit 205 of the laser scanner unit 201 may perform a cueing measurement in which the laser light 207 is intensively irradiated on a location corresponding to the position information of the feature received. Furthermore, the irradiation density of the laser light 207 may be higher at night or in rainy weather than during the daytime.

[0028] Furthermore, the scan patterns realized by the scanner control unit 205 are not limited to the conical scan and vertical scan described above. For example, a rectangular spiral pattern or a horizontal scan in which a horizontal scan is repeated while being shifted slightly upward and downward can also be realized.

[0029] The edge pattern extraction unit 206 extracts a collection of position data along the contour line or line segment shape of the target feature as an edge pattern, with reference to the density of position data around the vehicle 200 and the reflection intensity of the laser light 207. As an example of a method for extracting an edge pattern, as shown on the website of AEYE, the applicant of the specification of U.S. Patent No. 10495757, there is a method for extracting an edge pattern (FIG. 4(C)) indicating the shape boundary (edge) of the target feature existing around the vehicle 200 by integrating (fusing) image data (FIG. 4(A)) of the periphery of the vehicle 200 captured by the camera 209 with a large amount of position data (FIG. 4(B)) of the periphery of the vehicle 200 obtained by scanning with the laser light 207. Note that the method for extracting the edge pattern is not limited to the fusion of image data and position data, and may be a method for extracting an edge pattern only from position data without using image data. The target features here include road structures such as guardrails and road edges (including curbs), road edges, painted markings such as white lines, signposts (including road signs and traffic lights), buildings, and other landmarks near the road.

[0030] Therefore, in this embodiment, the edge pattern is a collection of significant position data that represents the contours of the target feature. By extracting this edge pattern, it is possible to significantly reduce unnecessary position data and extract only the position data of the target feature that is a candidate for the feature. As a result, it is possible to significantly reduce the load of subsequent processing.

[0031] The self-position estimation unit 202 includes, for example, an inertial measurement unit (IMU) and odometry constituted by a gyro and an accelerometer. The self-position estimation unit 202 acquires the attitude of the vehicle 200 expressed by the roll angle, pitch angle, and yaw angle, as well as the position of the vehicle 200, the amount of translation and the movement trajectory on a horizontal plane, in an earth reference coordinate system (hereinafter referred to as an "absolute coordinate system") with high accuracy using high-precision positioning, for example, GNSS (Global Navigation Satellite System) single positioning or GNSS interferometric positioning such as PPP-RTK (Precise Point Positioning-Real Time Kinematic). Furthermore, the self-position estimation unit 202 performs coordinate conversion of the position of each piece of position data constituting the edge pattern in the scanner reference coordinate system to a position in the absolute coordinate system (hereinafter referred to as an "absolute coordinate position") based on the acquired position of the vehicle 200 in the absolute coordinate system, and assigns the absolute coordinate position to each piece of position data. Note that the satellite positioning receiver signal of the vehicle 200 may be transmitted to the feature data generating device 300, where the satellite positioning receiver signal may be made highly accurate and absolute coordinate positions may be assigned to each piece of position data.

[0032] In a satellite invisible environment where GNSS cannot be used, the self-position estimation unit 202 may estimate the absolute coordinate position of the vehicle 200 by triangulation using various reference points (e.g., Ground Control Points (GCPs)) previously installed in the vehicle 200. At this time, if the vehicle 200 traveling in a tunnel recognizes a blower, the self-position estimation unit 202 may use the position information of the blower on a design drawing as a reference point. Furthermore, the self-position estimation unit 202 may upload measurement information required for calculating the absolute coordinate position of the vehicle 200 to a server (not shown) of the vehicle 200, and the server may estimate the absolute coordinate position of the vehicle 200 by using signal information of a GNSS receiver (not shown) of the vehicle 200.

[0033] In this embodiment, the coordinate conversion of each position data from the position on the scanner reference coordinate system to the absolute coordinate position is performed after the edge pattern is extracted, but the position information of all the position data may be converted to the absolute coordinate position before the edge pattern is extracted. However, the former is preferable from the viewpoint of improving the work efficiency because the number of objects to be converted is smaller in the former.

[0034] The data transmitter / receiver 203 transmits an edge pattern in which absolute coordinate positions have been assigned to each piece of position data to the feature data generating device 300, and also receives an instruction to search (scan) features, such as a request to execute a rescan, which will be described later.

[0035] The SD map 301 of the feature data generating device 300 is, in principle, a two-dimensional map, and contains less information than a three-dimensional high-precision map, but it does contain, for example, position information and shape information of features that serve as guides for route search. The position information of features is provided to the laser scanner unit 201 (scanner control unit 205) of the vehicle 200 as described above, and the shape information of features is used by the class classification processing unit 305 as described later.

[0036] The feature description data generation unit 302 describes the edge pattern transmitted from the data transmission / reception unit 203 using a multidimensional vector, and generates feature description data for the target feature.

[0037] Here, the edge pattern may contain measurement noise, and furthermore, some position data may be missing. Therefore, the feature description data generating unit 302 distinguishes between features having a circular or polygonal shape, such as a signpost, and features that are long to both ends, such as painted marks such as white lines, guardrails, and road edges (hereinafter referred to as "linear features"), and changes the description method using multidimensional vectors.

[0038] For example, as shown in Fig. 5(A), a polygon is extracted from a road sign 101, and feature description data is generated by using the coordinates of each point and the centroid of a rectangle that contacts the extracted polygon together with various information of the edge pattern as vector elements. The centroid of a rectangle that surrounds the road sign 101 may also be used as a vector element. As shown in Fig. 5(B), feature description data is generated from the road edge (see the discrete white dots in the figure) by using position information of each position data of the edge pattern as vector elements. Since the surface properties of the target feature can be determined according to the reflection intensity of the laser light 207, the reflection intensity may be set as an index of the surface properties as a vector element.

[0039] The feature feature data generation unit 303 (matching unit) performs matching, for example, between feature description data of features that have already been registered and prepared in the feature database 316 (hereinafter referred to as "existing feature description data") and the feature description data generated by the feature description data generation unit 302 (hereinafter referred to as "new feature description data").

[0040] Specifically, the existing feature description data of the area where the new feature description data exists is read from the feature database 316, and the absolute coordinate position information of the new feature description data is compared with the absolute coordinate position information of the existing feature description data. If these pieces of information match at a predetermined rate or more, the new feature description data is set as a shape feature vector for updating the existing feature description data (hereinafter also referred to as "shape feature vector for update"). On the other hand, if the absolute coordinate position information of the new feature description data does not match at a predetermined rate or more, the new feature description data is set as a shape feature vector for generating new feature data (hereinafter also referred to as "shape feature vector for generation"). The feature feature data generation unit 303 also stores these set shape feature vectors in the shape feature vector database 314.

[0041] FIG. 6 is a conceptual diagram of matching new feature description data with existing feature description data. FIG. 6(A) shows the matching of new feature description data 601 representing road signs with existing feature description data 602, and FIG. 6(B) shows the matching of new feature description data 603 representing road edges with existing feature description data 604.

[0042] In this embodiment, edge patterns are transmitted successively from a plurality of vehicles 200 to the feature data generation device 300 to generate feature description data, and the shape feature vectors are repeatedly stored in the shape feature vector database 314. As a result, the shape feature vector database 314 stores a large number of shape feature vectors for updating the same existing feature description data and shape feature vectors for generating the same features.

[0043] However, shape feature vectors stored in this manner may contain abnormal values ​​due to measurement noise, etc., and if a shape feature vector containing such abnormal values ​​is used, accurate feature data cannot be obtained.

[0044] Correspondingly, the abnormal measurement value detection unit 304 removes abnormal values ​​from the shape feature vector. Specifically, the abnormal measurement value detection unit 304 detects and removes abnormal values ​​from each shape feature vector using the Mahalanobis distance. The Mahalanobis distance is a distance in a multidimensional space that takes into account the variance of multiple data, and in this case, position data with a large Mahalanobis distance from the center of variance of each position data constituting the shape feature vector, which is a multidimensional vector, is removed.

[0045] For example, in Fig. 7, which is shown in two dimensions for ease of understanding, the position data a and b appear to be closer to the center of distribution than the position data c and d, but when the form of distribution is taken into account, the Mahalanobis distance of the position data a and b becomes longer than the Mahalanobis distance of the position data c and d. Therefore, here, the position data a and b are removed as outliers.

[0046] In addition, since the range that can be scanned by the laser beam 207 of the laser scanner unit 201 may change depending on the traveling state of the vehicle 200 and the surrounding conditions, the edge pattern extracted by the edge pattern extraction unit 206 of the laser scanner unit 201 may not include the entire position data of the feature. In such a case, it is necessary to identify which feature the edge pattern corresponds to.

[0047] In response to this, the class classification processing unit 305 determines which feature the edge pattern corresponds to. Specifically, the edge pattern transmitted from the data transmission / reception unit 203 is compared with the shape information of the feature in the area where the edge pattern exists, stored in the SD map 301. In the example of FIG. 8, it is found that the edge pattern 800 (FIG. 8(A)) corresponds to a part (FIG. 8(B)) of a deformed five-way intersection 801, which is a feature, but does not correspond to a four-way intersection 802 (FIG. 8(C)), which is also a feature. In this case, it is determined that the edge pattern 800 corresponds to a deformed five-way intersection, and the edge pattern 800 (and the corresponding shape feature vector) is classified as a deformed five-way intersection. In addition, if the orientation of the edge pattern does not match the orientation of the feature, it is difficult to accurately compare the edge pattern with the feature, so it is preferable to consider the position and posture of the vehicle 200, correct the orientation of the edge pattern to match the orientation of the feature, and then compare the edge pattern with the feature.

[0048] By comparing such edge patterns with some of the features stored in the SD map 301, the vehicle 200 can grasp its own position in the SD map 301 and can also grasp nearby features that serve as reference points for automatic driving. Therefore, even if the vehicle 200 does not have a three-dimensional high-precision map, as long as it has the SD map, the laser scanner unit 201, and the camera 209, it can estimate its own position and recognize the surrounding environment, and can perform safe driving support and automatic driving. In particular, by using features with high values ​​of the quality assurance index described later, it is possible to more accurately estimate the own position and recognize the surrounding environment, and to perform more advanced safe driving support and automatic driving.

[0049] The shape feature vector database generation unit 306 manages a large number of shape feature vectors stored in the shape feature vector database 314. For example, a plurality of update shape feature vectors corresponding to the same existing feature description data are managed in association with each other, and these update shape feature vectors are used to update the shape feature vectors corresponding to the existing feature description data.

[0050] The timing of updating the shape feature vector may be when the number of update shape feature vectors corresponding to the same existing feature description data exceeds a predetermined number. In this case, for example, a method using the idea of ​​a sliding window may be adopted, and specifically, the shape feature vector may be updated when the number of update shape feature vectors (corresponding to the area of ​​the sliding window) accumulated over time reaches a predetermined number. At this time, in areas with high traffic volume, the number of update shape feature vectors reaches the predetermined number in a relatively short time, so the shape feature vector corresponding to the existing feature description data of the feature is updated in a short cycle. On the other hand, in areas with low traffic volume, it takes a relatively long time for the number of update shape feature vectors to reach the predetermined number, so the shape feature vector corresponding to the existing feature description data of the feature is updated in a long cycle. In other words, the update cycle (frequency) of the shape feature vector may be influenced by the traffic volume in the area where the feature is located. This allows the update cycle of the shape feature vector to be optimized, and the calculation load of the shape feature vector database generation unit 306 to be optimized. In addition, the timing of updating the shape feature vector may be when a predetermined time has passed since the update shape feature vector was first stored.

[0051] The timing of updating the shape characteristic vector is not limited to these, and may be, for example, a predetermined time having passed since the completion date of construction in the relevant area in the publicly provided construction information described below, or a predetermined time having passed since the time when the shape characteristic vector changed over time, as described below, determined based on the statistical values ​​of each shape characteristic vector.

[0052] Furthermore, the method of updating the shape characteristic vector differs depending on the type of feature corresponding to the existing feature description data. For example, when the feature corresponding to the existing feature description data is a feature having a circular or polygonal shape, such as a signpost, the shape characteristic vector database generation unit 306 calculates, for example, the average values ​​of the coordinates of each point and the centroid of a rectangle that contacts a polygon included in the existing feature description data accumulated in the shape characteristic vector database 314 and each update shape characteristic vector, and updates the shape characteristic vector corresponding to the feature using these average values.

[0053] In addition, when a predetermined number or more of shape feature vectors for generation have been stored in the shape feature vector database 314, the shape feature vector database generation unit 306 generates a new shape feature vector using the average value of the coordinates of each point and the centroid point of the rectangle bordering the polygon, as described above, if the feature corresponding to the feature description data for generation is a feature such as a signpost.

[0054] Incidentally, when storing shape characteristic vectors in shape characteristic vector database 314, road edge extension shapes and road traffic divisions may change or signposts may be moved due to road construction or the like. In such cases, the vector elements of each shape characteristic vector before and after road construction change significantly, so if each shape characteristic vector before and after road construction is lumped together and subjected to statistical processing to update or generate a shape characteristic vector, the resulting feature data may no longer be accurate.

[0055] In response to this, the feature statistical data calculation unit 307 calculates the statistical values ​​of each shape characteristic vector stored in the shape characteristic vector database 314, and determines whether each shape characteristic vector includes a time change. For example, as shown in FIG. 9A, when a road sign is x When moving to , the moving average of the number of position data existing on line A of each shape feature vector corresponding to the road sign is x In such a case, the feature statistical data calculation unit 307 calculates the road sign at time t x It is determined that the user has moved at time t xThe shape feature vector database generator 306 does not use any shape feature vectors stored before time t x Each shape feature vector stored thereafter is used to update the shape feature vector corresponding to the existing feature description data.

[0056] The statistical value calculated by the feature statistical data calculation unit 307 is not limited to the moving average value. For example, the feature statistical data calculation unit 307 may calculate the Mahalanobis distance of the position data of each shape feature vector, classify each position data into a plurality of groups based on these Mahalanobis distances, and not use each shape feature vector corresponding to a group other than the group consisting of the most recent position data in the subsequent processing. Furthermore, the covariance value and histogram of the position data of each shape feature vector may be calculated. In this case, too, the temporal change (movement) of the feature can be detected by comparing the covariance values ​​and histograms calculated at regular intervals.

[0057] Furthermore, the method of detecting the movement of road signs, changes in the extension shape of road edges, and changes in road traffic divisions is not limited to the method based on the calculation of the statistical values ​​of each shape characteristic vector as described above. Since the movement of road signs, changes in the extension shape of road edges, and changes in road traffic divisions occur due to road construction, publicly available construction information (for example, construction information provided by the Japan Road Traffic Information Center, a public interest incorporated foundation) may also be used as a reference. In this case, the shape characteristic vectors stored before the construction is completed in the relevant area are not used in subsequent processing, and the shape characteristic vectors corresponding to the existing feature description data are updated using the shape characteristic vectors obtained after the construction is completed in the relevant area.

[0058] Furthermore, the statistical values ​​of each shape feature vector and publicly available construction information may be used in combination to detect the movement of road signs, changes in the extension shape of road edges, and changes in road traffic divisions. In this case, for example, the moving average value of the number of position data on line A of each shape feature vector corresponding to a road sign changes at time t x The time when the road sign is completed is compared with the time when the construction work is completed in the corresponding area, and if the difference between the two times is within a predetermined value, the road sign isx It is determined that the movement has occurred between the boundary.

[0059] In addition, the feature data generation system 100 may monitor the movement trajectory of each vehicle 200, and may use these movement trajectories to detect movement of road signs, changes in the extension shape of road edges, and changes in traffic divisions of roads. For example, if the movement trajectory of each vehicle 200 in a certain area is y If the change occurs at time t y At time t y The shape feature vector database generator 306 does not use any shape feature vectors stored before time t y The shape feature vectors corresponding to the existing feature description data are updated using the shape feature vectors stored thereafter. Whether or not the movement trajectory of each vehicle 200 has changed is determined based on whether or not the movement trajectory changes significantly with respect to the road line shape. For example, if the movement trajectory is not smooth and appears discontinuous, it is determined that the movement trajectory of each vehicle 200 has changed.

[0060] Incidentally, the movement trajectory of each vehicle 200 also changes when, for example, a temporary obstacle (fallen object, fallen tree, etc.) occurs on the road. In this case, since there is no movement of road signs, no change in the extension shape of the road edge, or no change in the traffic section of the road, it is preferable to use each shape feature vector stored before the movement trajectory of each vehicle 200 changes in updating the shape feature vector. Whether or not the change in the movement trajectory of each vehicle 200 is due to a temporary obstacle is determined based on whether or not the movement trajectory of each vehicle 200 returns to the original movement trajectory before a predetermined time has elapsed after the movement trajectory of each vehicle 200 has changed. Specifically, if the movement trajectory of each vehicle 200 returns to the original movement trajectory before a predetermined time has elapsed after the change, it is determined that the change in the movement trajectory of each vehicle 200 is due to a temporary obstacle, and if the movement trajectory of each vehicle 200 does not return to the original movement trajectory even after the predetermined time has elapsed, it is determined that the change in the movement trajectory of each vehicle 200 is due to the movement of road signs, a change in the extension shape of the road edge, or a change in the traffic section of the road.

[0061] Furthermore, the movement of road signs, changes in the extension shape of road edges, and changes in road traffic divisions are also caused by earthquakes and landslides. Therefore, publicly provided earthquake and disaster information may also be used in combination to detect the movement of road signs, changes in the extension shape of road edges, and changes in road traffic divisions. In this case, the shape feature vectors stored before the earthquake or landslide in the relevant area ceases are not used in subsequent processing, and the shape feature vectors corresponding to the existing feature description data are updated using the shape feature vectors obtained after the earthquake or landslide in the relevant area ceases.

[0062] Note that the movement of features can occur not only due to earthquakes and landslides, but also due to crustal movement. Whether the cause of the movement of a feature is an earthquake, a landslide, or crustal movement is determined based on whether the position data of the shape feature vector acquired from each vehicle 200 changes suddenly or gradually. For example, if the position data changes gradually, it is determined that the cause of the movement of the feature is crustal movement. In this case, as will be described later, the crustal movement correction unit 310 applies crustal movement correction to the position data of each shape feature vector, and updates the shape feature vectors corresponding to the existing feature description data using all the shape feature vectors after the crustal movement correction.

[0063] When the shape characteristic vector updated or generated by the shape characteristic vector database generating unit 306 corresponds to a road edge, the road characteristic assigning unit 308 assigns the center of a pair of road edges constituting one road to a lane center line, which is a virtual feature, as shown in FIG. 10, and assigns L as a lane ID to the lane center line. x (x is a natural number).

[0064] At this time, the road feature assignment unit 308 sets nodes at both ends of the lane center line, connects each node with a transition curve, such as a spline curve or a clothoid curve, along the lane center line, and sets waypoints as dots at a predetermined interval, for example, every 5 m, on the transition curve. x(x is a natural number) is assigned to each lane ID and waypoint ID. Correspondence with nearby buildings 1000 and signposts 1001 already recorded on a three-dimensional high-precision map is set for the lane ID and waypoint ID. This makes it easier to make deliveries by autonomous driving.

[0065] When the road width is large, the road is divided into multiple lanes in the width direction according to the dividing lines, and a lane ID is assigned to the center line of each lane. As a result, even if the vehicle 200 is traveling on the same road, by referring to the lane ID, it is possible to determine whether the vehicle is traveling in the straight lane or the right turn lane, and, for example, to distinguish which traffic lights to follow.

[0066] In addition, the road feature assignment unit 308 can represent a road network constructed by multiple lane center lines as a topological map by referring to lane IDs, and can verify the validity and consistency of road connections throughout the map.

[0067] The feature feature vector database generation unit 309 generates a feature feature vector by adding the statistical values ​​calculated by the feature statistical data calculation unit 307, the lane ID and waypoint ID assigned by the road feature assignment unit 308, and even the name of the feature to the updated or generated shape feature vector. Fig. 11 shows an example of a feature feature vector generated by the feature feature vector database generation unit 309, but the vector elements that make up the feature feature vector are not limited to the example in Fig. 11. In addition, the feature feature vector database generation unit 309 stores the generated feature feature vector in the feature feature vector database 315.

[0068] However, features may move due to crustal movements, and the amount of movement may reach, for example, 10 cm to 20 cm per year. Therefore, the position of the feature at the time of measurement included in the feature feature vector shifts from the position at the time of map creation. In response to this, the crustal movement correction unit 310 performs crustal movement correction to convert the current position data (position at the time of measurement) included in the feature feature vector into the original position data (position at the time of map creation), for example, the position data in the Japanese Geodetic Datum 2011 (JGD2011), which is the geodetic reference system of the vehicle 200. This ensures the accuracy of the position data included in the feature feature vector in the official coordinate system established by the country. In particular, in areas such as Japan where large crustal movements frequently occur due to major earthquakes, ensuring the accuracy in the official coordinate system is an essential quality assurance condition for feature data.

[0069] In addition, since the feature feature vector is obtained through statistical processing of multiple shape feature vectors, the amount of information in the current position data included in the feature feature vector may be reduced. Therefore, the crustal movement correction unit 310 may correct the current position data included in the edge pattern to the original position data, rather than the current position data included in the feature feature vector. This makes it possible to perform crustal movement correction before statistical processing is performed. In other words, since the crustal movement correction can be performed before the amount of information in the position data is reduced, more accurate original position data can be obtained.

[0070] Furthermore, although feature feature vectors are used as feature data to update or generate features of the 3D high precision map in the 3D high precision map updating device 400, not all feature feature vectors are equally reliable. For example, a feature feature vector including recently acquired position data is considered to be more reliable than a feature feature vector including position data acquired in the past, and a feature feature vector with large variance of position data, which is a statistical value, is considered to be less reliable than a feature feature vector with small variance.

[0071] Therefore, the quality assurance index calculation unit 311 generates a quality assurance index, which is an index indicating the reliability of each feature feature vector, and assigns it to each feature feature vector. In this embodiment, "Quality_Index" in Fig. 11 corresponds to the quality assurance index. For example, as shown in Fig. 12(A), for a feature feature vector in which the variance of the position data (see "Variance_Value" in the figure) is equal to or greater than a predetermined value and the acquisition time of the position data (see "Time_Stamp" in the figure) is earlier than a predetermined time, a low-value quality assurance index (see "Quality_Index" in the figure) is generated.

[0072] On the other hand, a quality assurance index with a high value is generated for an object feature vector with a small variance and whose position data was acquired recently, as shown in Figure 12(B). An object feature vector with a low quality assurance index value is not used for updating or generating objects in a 3D high precision map, or even if it is used for updating or generating objects, the object is not actively used in route search or autonomous driving. When generating a quality assurance index, a histogram may be used instead of the variance, and the value of the quality assurance index is not limited to the values ​​shown in Figures 12(A) and 12(B).

[0073] The timing for generating quality assurance indicators may be based on, for example, the completion of construction work in areas where features are located in publicly provided construction information, the timing when the statistical values ​​of feature feature vectors change significantly, the timing when the movement trajectories of each vehicle 200 passing near a feature change, the time when the correction value of crustal movement (crustal correction data) is published by the Geospatial Information Authority of Japan, the timing when an earthquake that occurred in an area where features are located in publicly provided earthquake information ends, or the timing when a landslide that occurred in an area where features are located in publicly provided disaster information ends.

[0074] In addition, the quality assurance index of the feature feature vector may be low not only when the variance of the location data of the feature feature vector is large or when the location data was acquired a long time ago, but also when, for example, it is confirmed from publicly provided construction information that a certain time has not passed since construction was carried out in the area where the feature exists, when a certain time has not passed since the movement of road signs, changes in the extension shape of the road edge, or changes in the traffic division of the road were detected from the movement trajectory of each vehicle 200, when it is confirmed from publicly provided earthquake information or disaster information that a certain time has not passed since an earthquake or landslide that occurred in the area where the feature exists ended, or when the crustal correction data is not reflected in the feature data after it is published by the Geospatial Information Authority of Japan. Note that, since the feature may move by several meters when an earthquake or landslide occurs, the quality assurance index is set to a fairly low value, that is, a value indicating that the feature is unusable.

[0075] The position data remeasurement request unit 312 requests the scanner control unit 205 of the laser scanner unit 201 of each vehicle 200 to perform rescanning with the laser light 207 again toward a location where a feature corresponding to a feature feature vector whose quality assurance index value has been set low due to the various factors described above exists. The request to perform rescanning with the laser light 207 is transmitted, for example, at the timing when the value of the quality assurance index of the feature feature vector becomes low, and is received by the data transmission / reception unit 203. This makes it possible to regenerate the feature feature vector based on a new (remeasured) edge pattern, and to maintain the reliability of the feature feature vector. As a result, it is possible to improve the reliability of the feature database 316 and the three-dimensional high-precision map. In addition, by performing rescanning with the laser light 207 only on features corresponding to the feature feature vector whose quality assurance index value has been set low, it is not necessary to perform rescanning with the laser light 207 on features corresponding to the feature feature vector whose quality assurance index value remains high. This eliminates the need to perform unnecessary rescanning of the laser light 207, and improves the efficiency of updating the feature database 316 and the features of the three-dimensional high-precision map.

[0076] Furthermore, when it is considered that the quality assurance index will be significantly decreased, the quality assurance index is reset to, for example, 0, and the laser light 207 is rescanned again toward the location where the feature corresponding to the feature characteristic vector for which the quality assurance index has been reset is present. The quality assurance index is reset, for example, when construction work is completed in the area where the feature is present, or when an earthquake or landslide that occurred in the area where the feature is present ends. Such information is obtained from publicly provided construction information, earthquake information, and disaster information. In addition, the quality assurance index may be reset when the statistical value of the feature characteristic vector changes significantly, when the movement trajectory of each vehicle 200 passing near the feature changes, or when crustal correction data for crustal movement is published by the Geospatial Information Authority of Japan. Note that the feature characteristic vector before the quality assurance index is reset is not used for automatic driving or for updating or generating features in a three-dimensional high-precision map.

[0077] The quality assurance index in this embodiment, for example, guarantees two things: the position precision and the position accuracy of the feature represented by the feature feature vector. The smaller the variance of the position data of the feature feature vector, the higher the position accuracy. For example, when 2σ (standard deviation) of the position data of the center position of the feature represented by the feature feature vector is 10 cm or less, 3 points are set as the quality assurance index corresponding to the position accuracy, when 2σ is 11 cm to 20 cm, 2 points are set, when 2σ is 21 cm or more, 1 point is set, and when it is more than 21 cm, 0 point is set. In addition, when setting the quality assurance index corresponding to the position accuracy, statistical data such as RMS (Root Mean Square) may be used instead of the above-mentioned 2σ.

[0078] The accuracy is higher when the deviation (deviation bias) of the center position of the feature represented by the feature feature vector from the true value of the center position of the feature is smaller. The center position of the feature represented by the feature feature vector changes as the number of shape feature vectors used to derive the feature feature vector increases. The more accurate the measurement method is for obtaining an absolute coordinate position, the more accurate the initial value of the center position of the feature represented by the feature feature vector is, and the higher the accuracy of the feature is. The higher the accuracy, the higher the quality assurance index is set. For example, when the above-mentioned 2σ is 10 cm or less, if the initial value is an absolute coordinate position that is surveyed in accordance with the "Standards for Public Survey Work Regulations" issued by the Geospatial Information Authority of Japan, Ministry of Land, Infrastructure, Transport and Tourism and is corrected for crustal deformation, three points are set as the quality assurance index corresponding to the accuracy, as shown in Table 1 below, if the initial value is an absolute coordinate position obtained by GNSS interferometric positioning such as PPP-RTK, two points are set, and if the initial value is an absolute coordinate position obtained by GNSS single point positioning, one point is set.

[0079] That is, the quality assurance index in this embodiment is set, for example, as shown in Table 1 below.

[0080] [Table 1]

[0081] Note that, as time passes, the number of measurements of the shape feature vectors used to derive the feature feature vector increases, and the possibility of noise being included increases, so the quality assurance index corresponding to the accuracy may be set low. However, even if the number of measurements of the shape feature vector increases, the center position of the feature represented by the feature feature vector changes only gradually, so in this case, the deterioration of the quality assurance index due to the passage of time is set small.

[0082] The use of features derived from a feature feature vector may be changed depending on the quality assurance index of the feature feature vector. For example, features derived from a feature feature vector having a quality assurance index of three points may be used for autonomous driving, and features derived from a feature feature vector having a quality assurance index of two points or less may be specified to be used only for ADAS and not for autonomous driving.

[0083] The position of each vehicle 200 is acquired in real time by the self-position estimator 202 of each vehicle 200 using GNSS independent positioning or GNSS interferometric positioning. However, as described above, the satellite positioning receiver signal of each vehicle 200 may be transmitted to the feature data generating device 300, and the satellite positioning receiver signal may be processed in the feature data generating device 300 to acquire the absolute coordinate position of the vehicle 200. In this case, the satellite positioning receiver signal is post-processed, and the signal used for acquiring the position can be selected. For example, the position of the vehicle 200 can be acquired after removing the signal from the satellite causing the multipath. On the other hand, since it is possible to determine whether each signal causes the multipath by comparing it with other signals, when acquiring the position of the vehicle 200 in real time, the signal from the satellite causing the multipath cannot be easily removed. Therefore, there is a method of estimating the satellite causing the multipath by a ray tracing method using a three-dimensional high-precision map and removing the signal from the satellite. However, since this requires a large amount of calculation and uses a large amount of memory, the signal comparison and ray tracing may be usually performed offline when creating or updating the feature database 316. As a result, when processing the satellite positioning receiver signal in the feature data generation device 300, a more accurate position of the vehicle 200 can be obtained than when the position is obtained in real time at the vehicle 200, so the quality assurance index may be set higher.

[0084] Furthermore, if the feature corresponding to the feature feature vector is a road, a virtual vehicle may be driven on the road obtained from the feature feature vector, and a quality assurance index for the feature feature vector may be set according to whether or not the time displacement of the attitude (roll angle, pitch angle, and yaw angle) of the vehicle, particularly the speed and acceleration of the attitude change, deviates from the range of standard values ​​that occur in a vehicle that normally drives. Here, if the speed and acceleration of the attitude change deviate from the range of standard values, the shape of the feature (road) is deemed unnatural, and the quality assurance index is set to a low value. Alternatively, the quality assurance index for the feature feature vector may be set according to the degree of agreement between the road center (lane center) calculated from the positions of each road sign standing along the road and the running trajectory when a virtual vehicle is driven on the road obtained from the feature feature vector. Here, the lower the degree of agreement between the calculated road center and the running trajectory, the more unnatural the shape of the feature (road) is, and the lower the quality assurance index is set.

[0085] In addition, when the feature corresponding to the feature vector is a linear feature, the linear shape of the feature, which is a spline curve (y=f(x) in FIG. 15) already obtained from the feature vector, and the new measurement data (position data constituting the newly measured linear feature) ((x i , y i )) (σ in Fig. 15) i )L 1 The average value of the norm and L 2 The average value of the norm is taken as a quantitative value of the error, and the smaller the difference, the higher the quality assurance index is set. The difference at this time is, for example, shown on the horizontal axis of Table 1 above.

[0086] Furthermore, if the characters or patterns of a road sign captured by the camera 209 of each vehicle 200 remain the same, the quality assurance index of the feature vector corresponding to the road sign may be set to a high value, and if the characters or patterns of the road sign change, the quality assurance index of the feature vector after the change may be reset to an initial setting value. In particular, if the characters or patterns of a road sign change, the position data remeasurement request unit 312 may request the scanner control unit 205 of the laser scanner unit 201 of each vehicle 200 to rescan the road sign whose characters or patterns have changed.

[0087] The quality assurance indexes of the feature vectors corresponding to each feature set by the method described above may be compiled into a quality table for each region, and using this quality table, the feature data generation device 300 may suggest to the user which feature vector should be used to update or generate features in the relevant region.

[0088] Thereafter, the feature characteristic vector for which the quality assurance index has been generated is stored in the feature database 316, and the feature database 316 is updated. FIG. 13 is a diagram for explaining the configuration of the feature database 316. The feature database 316 is an object-oriented database, and stores each feature characteristic vector as an object. For example, the feature characteristic vectors of "general roads" and "expressways" are classified into the "road" class, and the vector elements of each feature characteristic vector are managed as feature values ​​(attributes). The feature data generating device 300 transmits the updated feature database 316 to the 3D high precision map updating device 400, and the 3D high precision map updating device 400 refers to the quality assurance index to extract necessary feature characteristic vectors from the feature database 316, and updates or generates features of the 3D high precision map.

[0089] FIG. 14 is a flowchart showing a method for generating feature data in this embodiment.

[0090] In FIG. 14, first, the laser scanner unit 201 of the vehicle 200 scans the surroundings with the laser light 207 (step S1401), and extracts the edge pattern of a target feature existing around the vehicle 200 (step S1402).

[0091] Next, the self-location estimation unit 202 assigns an absolute coordinate position to each piece of position data of the edge pattern (step S1403). Note that, for the purpose of reducing the processing load on the vehicle 200, the self-location estimation unit 202 may transmit the edge pattern to the feature data generation device 300 without assigning an absolute coordinate position (high accuracy), and the feature description data generation unit 302 may assign an absolute coordinate position to each piece of position data of the edge pattern.

[0092] Thereafter, the data transmitter / receiver 203 transmits the edge pattern to which the absolute coordinate positions have been assigned to the feature data generation device 300 (step S1404). At this time, instead of S1412 described later, the crustal movement correction unit 310 may perform crustal movement correction on the position data to which the absolute coordinate positions of the current period have been assigned, which is included in the edge pattern, to the position data of the original period.

[0093] Note that steps S1401 to S1404 may be repeatedly executed by one vehicle 200, or may be executed by a plurality of vehicles 200. As a result, edge patterns of a plurality of target features are transmitted to the feature data generating device 300.

[0094] Next, the feature description data generation unit 302 generates feature description data for the target feature (step S1405), and the feature feature data generation unit 303 matches the existing feature description data with the new feature description data (step S1406), and stores a shape feature vector in the shape feature vector database 314 based on the matching result (step S1407).

[0095] Then, the abnormal measurement value detection unit 304 removes abnormal values ​​from the shape feature vectors stored in the shape feature vector database 314 (step S1408), and the shape feature vector database generation unit 306 updates the shape feature vectors corresponding to the existing feature description data through statistical processing of multiple update shape feature vectors (step S1409).

[0096] Next, the feature characteristic vector database generating unit 309 generates a feature characteristic vector based on the updated shape characteristic vector (step S1410) and stores it in the feature characteristic vector database 315 (step S1411). After that, the crustal movement correction unit 310 performs crustal movement correction on the current position data included in the stored feature characteristic vector to the original position data (step S1412). Note that if crustal movement correction has been performed immediately after step S1403, step S1412 is skipped.

[0097] Next, the quality assurance index calculation unit 311 generates a quality assurance index for each feature vector (step S1413), each feature vector is stored in the feature database 316 (step S1414), the feature database 316 is updated, and this process is terminated.

[0098] According to this embodiment, an edge pattern is extracted from a large number of position data around the vehicle 200 obtained by scanning the laser light 207, and a feature characteristic vector used for updating or generating features of a three-dimensional high-precision map as feature data is generated through statistical processing of the position data of the edge pattern. That is, compared to generating feature characteristic vectors directly from a large number of position data around the vehicle 200, the effort and time can be significantly reduced, and it is possible to repeatedly generate feature characteristic vectors. This allows the feature database 316 to be repeatedly updated. As a result, errors in the feature database 316 are corrected, and the accuracy of the feature database 316 can be guaranteed. In addition, by updating or generating a three-dimensional high-precision map using the feature characteristic vectors of the updated feature database 316, the features of the three-dimensional high-precision map are kept up to date, and the accuracy of the three-dimensional high-precision map can be guaranteed.

[0099] Furthermore, in this embodiment, the feature vectors used to update the feature database 316 are generated through statistical processing of multiple shape feature vectors, so that the features updated and generated on the 3D high-precision map using the feature vectors can be made statistically valid.

[0100] Although the preferred embodiments of the present invention have been described above, the present invention is not limited to the above-described embodiments, and various modifications and changes are possible within the scope of the gist of the present invention. [Explanation of symbols]

[0101] 100 Feature Data Generation System 200 vehicles 300 Geographical feature data generator 301 SD Map 302 Feature Description Data Generation Unit 303 Feature Data Generation Unit 304 Abnormal measurement value detection unit 306 Shape Feature Vector Database Generation Unit 307 Geographical Statistical Data Calculation Department 309 Feature Vector Database Generation Unit 310 Crustal Deformation Correction Division 311 Quality Assurance Index Calculation Department 312 Position data remeasurement request section 313 Feature Database Generation Unit 314 Shape Feature Vector Database 315 Feature Vector Database 316 Feature Database

Claims

1. a feature data generation unit that generates feature data representing features from an edge pattern that indicates boundaries of features present around a moving object, the edge pattern being extracted from position data of a large number of measurement points around the moving object measured using LIDAR technology; a quality assurance index calculation unit that calculates a quality assurance index that guarantees the quality of the generated feature data, The quality assurance index calculation unit sets the quality assurance index lower the longer the time that has passed since the feature data was generated, and changes the quality assurance index depending on a method for obtaining the position of the feature in the Earth's reference coordinate system.

2. 2. The feature data generating device according to claim 1, further comprising: a step of determining whether or not to regenerate the feature data in accordance with the quality assurance index.

3. The feature data generating device according to claim 1, wherein a use of the feature data is determined according to the quality assurance index.

4. The feature data generating device according to claim 3, further comprising a step of determining whether or not the feature data is to be used for automatic driving of a mobile object in accordance with the quality assurance index.

5. The feature data generating device according to claim 3, wherein whether or not the feature data is used for updating or generating features in a three-dimensional high precision map is determined according to the quality assurance index.

6. 6. The feature data generating device according to claim 1, wherein the quality assurance index guarantees the positional precision of the feature and the accuracy of the position of the feature.

7. The feature data generating device of any one of claims 1 to 6, characterized in that the quality assurance index calculation unit generates the quality assurance index at the time when construction work being carried out in the area where the feature is located is completed, the time when the movement trajectories of multiple moving objects passing near the feature change, the time when an earthquake that occurred in the area where the feature is located ends, the time when a landslide that occurred in the area where the feature is located ends, or the time when crustal correction data is published by the Geospatial Information Authority of Japan.

8. The feature data generating device of any one of claims 1 to 6, characterized in that the quality assurance index calculation unit resets the quality assurance index at the time when construction work being carried out in the area where the feature is located is completed, at the time when the movement trajectories of multiple moving objects passing near the feature change, at the time when an earthquake that occurred in the area where the feature is located subsides, or at the time when a landslide that occurred in the area where the feature is located subsides.

9. The feature data generation device of any one of claims 1 to 8, characterized in that the quality assurance index calculation unit sets the quality assurance index low in the following cases: a predetermined time has not passed since construction was carried out in the area where the feature is located; a predetermined time has not passed since a change in the movement trajectories of multiple moving objects passing near the feature was detected; a predetermined time has not passed since an earthquake that occurred in the area where the feature is located subsided; a predetermined time has not passed since a landslide that occurred in the area where the feature is located subsided; or crustal correction data has not been reflected in the feature data after it was published by the Geospatial Information Authority of Japan.

10. The feature data generating device according to any one of claims 1 to 9, characterized in that the quality assurance index calculation unit sets the quality assurance index higher the smaller the difference between the position data constituting an already obtained linear feature and the newly measured linear feature.

11. The feature data generating device described in claim 1, characterized in that the method for obtaining the position of the feature in the Earth Reference Coordinate System includes at least one of the following methods: a method for measuring an initial value of the center position of the feature in accordance with the "Standards for Public Surveying Work Regulations" published by the Geospatial Information Authority of Japan, Ministry of Land, Infrastructure, Transport and Tourism, and correcting for crustal movement; a method for obtaining an initial value of the center position of the feature by GNSS interferometric positioning; and a method for obtaining an initial value of the center position of the feature by GNSS independent positioning.

12. An edge pattern extraction step of extracting an edge pattern indicating boundaries of target features existing around a moving object from position data of a large number of measurement points around the moving object measured using LIDAR technology; a feature data generating step of generating feature data representing a feature from the extracted edge pattern; a quality assurance index calculation step of calculating a quality assurance index for assuring the quality of the generated feature data, A method for generating feature data, characterized in that in the quality assurance index calculation step, the quality assurance index is set lower the longer the time has passed since the feature data was generated, and the quality assurance index is changed depending on a method for obtaining the position of the feature in the Earth Reference Coordinate System.

13. 13. The method for generating feature data according to claim 12, wherein in said quality assurance index calculation step, said calculated quality assurance index is assigned to said feature data.

14. The method for generating feature data described in Claim 12, characterized in that the method for obtaining the position of the feature in the Earth Reference Coordinate System includes at least one of the following methods: a method for measuring an initial value of the center position of the feature in accordance with the "Standards for Public Survey Work Regulations" published by the Geospatial Information Authority of Japan, Ministry of Land, Infrastructure, Transport and Tourism, and correcting for crustal movement; a method for obtaining an initial value of the center position of the feature by GNSS interferometric positioning; and a method for obtaining an initial value of the center position of the feature by GNSS independent positioning.

Images