JP2019179217A - Map correction method and map correction device - Google Patents

Abstract

To provide a map correction method that can locally correct an error of map data while maintaining continuity of the map data.SOLUTION: A map correction method to correct map data based on a road structure is to set a plurality of grid points on map data (S11), to detect a road structure around a vehicle and calculate an offset parameter of each grid point so as to make a difference between the detected road structure and the road structure of the map data smaller (S18), and to correct the map data using the offset parameter of each grid point (S15).SELECTED DRAWING: Figure 10

Description

  The present invention relates to a map correction method and a map correction device.

  There may be a local error in map data such as a high-precision map suitable for automatic driving. In order to correct local errors in map data, map data is divided into multiple meshes, the actual position of the points in the mesh is compared with the map data, and the points in the mesh are affine transformed to map data. There is known a method of deforming (see Patent Document 1).

JP 2004-241228 A

  However, in the method described in Patent Document 1, since map data is deformed in units of meshes, there is a problem that discontinuity occurs at the boundary between adjacent meshes when map data deformed in units of meshes are combined.

  An object of this invention is to provide the map correction method and map correction apparatus which can correct | amend the error of map data locally, maintaining the continuity of map data.

  One aspect of the present invention is to set a plurality of grid points on map data, detect the road structure around the vehicle, and reduce the difference between the detected road structure and the road structure of the map data, An offset parameter for each grid point is calculated, and map data is corrected using the offset parameter for each grid point.

  ADVANTAGE OF THE INVENTION According to this invention, the map correction method and map correction apparatus which can correct | amend the error of map data locally can be provided, maintaining the continuity of map data.

It is the schematic which shows an example of the map correction apparatus which concerns on embodiment of this invention. It is the schematic which shows an example of the map data which concern on embodiment of this invention. It is the schematic which shows an example of the map data divided | segmented into the mesh which concerns on embodiment of this invention. It is the schematic of the correction process of the map data which concerns on a comparative example. It is the schematic of the correction process of the map data which concerns on a comparative example. It is the schematic which shows an example of the correction process of the map data which concerns on embodiment of this invention. It is the schematic which shows an example of the correction process of the map data which concerns on embodiment of this invention. It is the schematic which shows an example of the correction process of the map data which concerns on embodiment of this invention. It is the schematic which shows an example of the correction process of the map data which concerns on embodiment of this invention. It is a flowchart which shows an example of the map correction method which concerns on embodiment of this invention. It is a flowchart which shows the modification of the map correction method which concerns on embodiment of this invention. It is the schematic which shows the correction process of the map data which concerns on other embodiment of this invention. It is the schematic which shows the correction process of the map data which concerns on other embodiment of this invention.

  Hereinafter, embodiments of the present invention will be described with reference to the drawings. In the following description of the drawings, the same or similar parts are affixed with the same or similar reference numerals. However, it should be noted that the drawings are schematic, and the relationship between the thickness and the planar dimensions, the ratio of the thickness, and the like are different from the actual ones. Moreover, it is a matter of course that portions having different dimensional relationships and ratios are included between the drawings. The following embodiments of the present invention exemplify apparatuses and methods for embodying the technical idea of the present invention, and the technical idea of the present invention is the material, shape, The structure, arrangement, etc. are not specified below. The technical idea of the present invention can be variously modified within the technical scope defined by the claims described in the claims.

(Map correction device)
The map correction apparatus according to the embodiment of the present invention can be mounted on a vehicle (hereinafter, the vehicle equipped with the map correction apparatus according to the embodiment of the present invention is referred to as “own vehicle”). As shown in FIG. 1, the map correction device according to the embodiment of the present invention includes a processing circuit 1, a storage device 2, a global positioning system (GPS) receiver 3, a vehicle sensor 4, an ambient sensor 5, an actuator 6, and a display 7. Is provided. The processing circuit 1, the storage device 2, the GPS receiver 3, the vehicle sensor 4, the surrounding sensor 5, the actuator 6, and the display 7 can transmit and receive data and signals by wire or wireless such as a controller area network (CAN) bus. is there.

  As the storage device 2, a semiconductor storage device, a magnetic storage device, an optical storage device, or the like can be used, and may be incorporated in the processing circuit 1. The storage device 2 includes a map storage unit 20 that stores map data.

  The map data stored in the map storage unit 20 includes, for example, high-accuracy map data (hereinafter simply referred to as “high-accuracy map”) as a correction target by the map correction apparatus according to the embodiment of the present invention. A high-precision map is suitable as a map for automatic driving. The high-accuracy map is map data with higher accuracy than map data for navigation (hereinafter simply referred to as “navigation map”), and includes more detailed lane-by-lane information than road-by-road information. For example, the high-precision map includes lane node information indicating a reference point on a lane reference line (for example, a central line in the lane) and lane link information indicating a lane section mode between the lane nodes, as lane unit information. including. The lane node information includes the identification number of the lane node, the position coordinates, the number of connected lane links, and the identification number of the connected lane link. The lane link information includes an identification number of the lane link, a lane type, a lane width, a lane boundary type, a lane shape, and a lane reference line shape. The high-precision map further includes the type and position coordinates of features such as traffic lights, stop lines, signs, buildings, utility poles, curbs, pedestrian crossings, and lane nodes that correspond to the position coordinates of the features. Feature information such as the identification number of the vehicle and the identification number of the lane link.

  Since the high-accuracy map includes node and link information in units of lanes, it is possible to specify the lane in which the host vehicle is traveling on the travel route. The high-accuracy map has coordinates that can represent the position of the lane in the extending direction and the width direction. The high-precision map has coordinates (for example, longitude, latitude, and altitude) that can represent the position in the three-dimensional space, and the lane and the feature can be described as shapes in the three-dimensional space.

  In the embodiment of the present invention, a case where a high-precision map is a correction target is illustrated, but a navigation map may be a correction target. The map data stored in the map storage unit 20 may include a navigation map. The navigation map includes road unit information. For example, a navigation map includes road node information indicating a reference point on a road reference line (for example, a center line of a road) and road link information indicating a road section mode between road nodes as road unit information. . The road node information includes an identification number of the road node, position coordinates, the number of connected road links, and an identification number of the connected road link. The road link information includes the road link identification number, road standard, link length, number of lanes, road width, and speed limit.

  The map data database is managed by a server different from the map correction apparatus according to the embodiment of the present invention, and the updated difference data of the map data is acquired through, for example, a telematics service and stored in the map storage unit 20. The map data may be updated. Moreover, you may make it acquire map data by telematics services, such as vehicle-to-vehicle communication and road-to-vehicle communication, according to the position where the own vehicle is traveling. By using the telematics service, the host vehicle does not need to have map data having a large data capacity, and the capacity of the memory can be suppressed. In addition, since the updated map data can be acquired by using the telematics service, it is possible to accurately grasp the actual driving situation such as a change in road structure and the presence / absence of a construction site. Furthermore, by using the telematics service, map data created based on data collected from a plurality of other vehicles other than the host vehicle can be used, so that accurate information can be grasped.

  The GPS receiver 3 receives radio waves from a plurality of navigation satellites, acquires the current position of the host vehicle, and outputs the acquired current position of the host vehicle to the processing circuit 1. Note that a global positioning system (GNSS) receiver other than the GPS receiver 3 may be included.

  The vehicle sensor 4 is a sensor that detects the current position and traveling state of the host vehicle. The vehicle sensor 4 can be constituted by a vehicle speed sensor, an acceleration sensor, a gyro sensor, or the like, for example, but the type and number of the vehicle sensors 4 are not limited to this. The vehicle speed sensor configured by the vehicle sensor 4 detects the vehicle speed based on the wheel speed of the host vehicle, and outputs the detected vehicle speed to the processing circuit 1. The acceleration sensor configured by the vehicle sensor 4 detects acceleration in the front-rear direction and the vehicle width direction of the host vehicle, and outputs the detected acceleration to the processing circuit 1. The gyro sensor configured by the vehicle sensor 4 detects the angular velocity of the host vehicle and outputs the detected angular velocity to the processing circuit 1.

  The ambient sensor 5 is a sensor that detects the ambient environment (surrounding situation) of the host vehicle. The ambient sensor 5 can be configured by, for example, a camera, a radar, and a communication device, but the type and number of the ambient sensors 5 are not limited to this. As a camera constituting the ambient sensor 5, a CCD camera or the like can be used. The camera may be a monocular camera or a stereo camera. The camera captures the surrounding environment of the host vehicle, and from the captured image, the relative position of the vehicle, a pedestrian, a bicycle or other object and the host vehicle, the distance between the object and the host vehicle, the lane boundary line (white line) on the road, Road structures such as curbs are detected as surrounding environment data of the vehicle, and the detected surrounding environment data is output to the processing circuit 1.

  As the radar constituting the ambient sensor 5, for example, a millimeter wave radar, an ultrasonic radar, a laser range finder (LRF), or the like can be used. The radar detects the relative position between the object and the host vehicle, the distance between the object and the host vehicle, the relative speed between the object and the host vehicle, etc. as the surrounding environment data, and processes the detected surrounding environment data. Output to circuit 1. The communication device constituting the ambient sensor 5 receives data on the surrounding environment of the host vehicle by performing vehicle-to-vehicle communication with another vehicle, road-to-vehicle communication with a roadside device, communication with a traffic information center, or the like. The received ambient environment data is output to the processing circuit 1. In the present specification, each of the GPS receiver 3, the vehicle sensor 4, and the surrounding sensor 5 can be configured as a sensor capable of detecting an actual (real) road structure around the host vehicle.

  The processing circuit 1 is a controller such as an electronic control unit (ECU) that performs arithmetic and logical operations of processing necessary for operations performed by the map correction apparatus according to the embodiment of the present invention. For example, the processor, the storage device, and the input / output I / F may be provided. The processor can correspond to a microprocessor equivalent to an arithmetic logic unit (ALU), a control circuit (control unit), a central processing unit (CPU) including various registers, and the like. The storage device built in or externally attached to the processing circuit 1 includes a semiconductor memory, a disk medium, and the like, and may include a storage medium such as a register, a cache memory, and a ROM and RAM used as a main storage device. For example, the processor can execute a program stored in advance in the storage device and indicating a series of processes necessary for the operation of the map correction apparatus according to the embodiment of the present invention.

  The processing circuit 1 divides the high-precision map stored in the map storage unit 20 into a plurality of regions (mesh), and configures a mesh model including the mesh and its end points (grid points). The processing circuit 1 further calculates an offset parameter of each grid point on the high-precision map based on an actual road structure obtained from sensors such as the GPS receiver 3, the vehicle sensor 4, and the surrounding sensor 5. The high-precision map is corrected using the offset parameter. For example, the processing circuit 1 is configured so that the difference between the road structure detected by sensors such as the GPS receiver 3, the vehicle sensor 4, and the surrounding sensor 5 and the corresponding road structure on the high-precision map is reduced. The offset parameter of the point is optimized, and the error of the high-precision map is corrected using the optimized offset parameter.

  The processing circuit 1 includes logical blocks such as a mesh setting unit 11, a self-position calculation unit 12, a parameter determination unit 13, an in-mesh interpolation unit 14, a map correction unit 15, and a vehicle control unit 16 as functional or physical hardware resources. Prepare as. The mesh setting unit 11, the self-position calculating unit 12, the parameter determining unit 13, the intra-mesh interpolating unit 14, the map correcting unit 15, and the vehicle control unit 16 constituting the processing circuit 1, a field programmable gate array (FPGA) or the like A programmable logic device (PLD) or the like may be physically configured, or a functional logic circuit or the like that is set equivalently by processing by software in a general-purpose semiconductor integrated circuit may be used.

  Further, the mesh setting unit 11, the self-position calculating unit 12, the parameter determining unit 13, the intra-mesh interpolating unit 14, the map correcting unit 15, the vehicle control unit 16 and the like constituting the processing circuit 1 are configured by a single hardware. It may be configured by separate hardware. For example, the mesh setting unit 11, the self-position calculating unit 12, the parameter determining unit 13, the intra-mesh interpolating unit 14, and the map correcting unit 15 constituting the processing circuit 1 can be configured by a car navigation system such as an in-vehicle infotainment (IVI) system. The vehicle control unit 16 constituting the processing circuit 1 can be constituted by a travel support system such as an advanced travel support system (ADAS).

  The mesh setting unit 11 reads out a high accuracy map stored in the map storage unit 20 and sets a plurality of grid points (representative points) on the high accuracy map. The mesh setting unit 11 further sets a polygon having a grid point as a vertex as a mesh (divided region).

  In the following, the case where the high-precision map MD schematically shown in FIG. 2 is corrected will be described as an example. The high-precision map MD includes data of two roads 21 and 22 that intersect each other as road structure data. The high-precision map MD can be configured in a range including the current position of the host vehicle and the past travel route of the host vehicle. The mesh setting unit 11 reads the high-accuracy map MD from the map storage unit 20 and, as shown in FIG. 3, grid points Pij (i = 1 to 4, j = 1 to 5) on the high-accuracy map MD in a grid pattern. Set to. Furthermore, the mesh setting unit 11 sets meshes Mmn (m = 1 to 3, n = 1 to 4) partitioned into rectangles by straight lines connecting the grid points Pij. A mesh model is configured by the grid points Pij and the mesh Mmn.

  FIG. 3 illustrates a case where the grid point Pij is 4 × 5 = 20 points and the mesh Mmn is 3 × 4 = 12 regions, but the number of grid points Pij and mesh Mmn is not particularly limited. The interval, the arrangement position, the size and shape of the mesh are not limited. For example, one diagonal line may be added on each mesh Mmn of FIG. 3 to set a triangular mesh partitioned by three grid points Pij. Alternatively, the grid points may be set in a triangular lattice shape to set a triangular mesh. Further, the mesh may be a pentagon or more. Also, grid points may be set irregularly and meshes having different shapes may be set.

  For convenience, FIG. 3 illustrates a case where the grid points Pij and the mesh Mmn having two-dimensional coordinates are set in two dimensions using the two-dimensional high-precision map MD as a correction target. As a correction target, grid points and meshes may be expanded in three dimensions. That is, grid points having three-dimensional coordinates may be set on the three-dimensional space constituting the high-precision map, and a three-dimensional polyhedron having the grid points as vertices may be set as a mesh.

  Here, with reference to FIG.4 and FIG.5, the correction process of the map data which concerns on a comparative example is demonstrated. As shown in FIG. 4, the high-precision map includes information on roads 31 and 32. A plurality of grid points P are set on the high-precision map, and a mesh M partitioned into rectangles by straight lines connecting the grid points P is set. Then, the points in each mesh M are affine transformed to deform the high-accuracy map in mesh units, and the deformed high-accuracy maps are combined as shown in FIG. In this case, as shown in FIG. 5, discontinuity occurs at the boundary portion between adjacent meshes M, and the road 32 is divided in the region A shown in FIG. 5, which is inappropriate for, for example, automatic driving. On the other hand, the map correction apparatus according to the embodiment of the present invention can locally correct an error in a high-precision map while maintaining the continuity of the high-precision map.

  The self-position calculation unit 12 shown in FIG. 1 performs matching between the self-position detected by the GPS receiver 3 and the high-precision map MD, and records the history of the position of the host vehicle (self-position) on the high-precision map MD. calculate. In addition, the self-position calculation part 12 may acquire the log | history of a self-position based on road-to-vehicle communication or vehicle-to-vehicle communication. The self-position calculation unit 12 calculates the movement amount and the movement direction of the own vehicle from vehicle information (odometry information) such as the wheel speed and yaw rate of the own vehicle detected by the vehicle sensor 4, that is, the odometry A position history may be calculated. The self-position calculation unit 12 may calculate a travel locus of the host vehicle by approximating the points constituting the self-position history by a curve, and may use this as the one corresponding to the self-position history.

  The parameter determination unit 13 is based on the actual road structure, and an offset parameter (hereinafter referred to as “grid”) for correcting the high-precision map MD, which is unique to each grid point Pij on the high-precision map MD shown in FIG. It is called (parameter). ”Uij (i = 1 to 4, j = 1 to 5) is determined (calculated). The actual road structure includes a history of the self-position corresponding to the actual center line of the lane acquired from the GPS receiver 3 or the like, and a lane boundary line or a road boundary line detected by the surrounding sensor 5. Lane boundaries or road boundaries include white lines and curbs. The grid parameter Uij of each grid point Pij can be composed of parameters based on known coordinate conversion means such as vectors, matrices, parallel movement amounts, affine transformations, and the like. Note that a common grid parameter Uij may be calculated between adjacent grid points Pij.

  For example, the parameter determination unit 13 compares the high-precision map MD shown in FIG. 3 with the actual road structure, and calculates an error for the actual road structure that exists in the high-precision map MD. The parameter determination unit 13 uses an optimization method to minimize the error of the corrected high-precision map MD when the high-precision map MD is corrected based on the calculated error of the high-precision map MD. Optimize the grid parameter Uij. As an optimization method, for example, a general method such as a steepest descent method or a Gauss-Newton method can be used.

  Here, an example of the optimization process of the grid parameter Uij will be described with reference to FIG. 6 while focusing on the two meshes M11 and M12 that are part of the high-precision map MD shown in FIG. The road 21 included on the meshes M11 and M12 includes two lanes L11 and L12. In FIG. 6, the center line L1 of the lane L11 corresponding to the traveling lane of the host vehicle on the high-precision map MD is indicated by a one-dot chain line. In addition, an actual self-position history (running locus) L2 calculated by the self-position calculation unit 12 is converted into a coordinate system of the high-precision map MD and is indicated by a two-dot chain line. Here, because the high-accuracy map MD includes an error with respect to the actual road structure, the center line L1 of the lane L11 on the high-accuracy map MD and the actual self calculated by the self-position calculating unit 12 Consider a case where the position history L2 is misaligned.

  For example, when the high-precision map MD is corrected, the parameter determination unit 13 sets each grid so that the error between the center line L1 of the lane L11 on the high-precision map MD after correction and the history L2 of the self-position is minimized. The grid parameter Uij of the point Pij is determined. For example, the difference (error) between the constituent points p11 to p15 of the center line L1 of the lane L11 on the high-accuracy map MD and the constituent points p21 to p25 that are the nearest points on the history L2 of the self-position with respect to the constituent points p11 to p15. The square sum of squares may be calculated, and the grid parameter Uij may be determined so as to minimize the calculated sum.

  Note that the constituent points p11 to p15 of the center line L1 of the lane L11 on the high-precision map MD may be actual constituent points of the center line L1, or virtual points obtained by curve interpolation of the constituent points of the actual center line L1. It may be a point. The constituent points p21 to p25 on the self-position history L2 may be actual points on the actual self-position history L2, or virtual points obtained by curve interpolation of the constituent points on the actual self-position history L2. But you can.

  Instead of using the self-position history L2 calculated by the self-position calculation unit 12, or in addition to the self-position history L2, the parameter determination unit 13 uses the lane boundary as the actual road structure detected by the vehicle sensor 4. The grid parameter Uij may be optimized by comparing a line, a road boundary line, a curbstone, and the like with data corresponding to the lane boundary line, the road boundary line, the curbstone, etc. on the high-precision map MD. That is, when the high-precision map MD is corrected, the parameter determination unit 13 detects the actual lane boundary line, road boundary line, curbstone, etc. detected by the vehicle sensor 4 and the corresponding lane of the high-precision map MD after correction. The grid parameter Uij may be optimized so that the difference in data such as a boundary line, a road boundary line, or a curb is minimized.

  In FIG. 6, the error between the center line L1 of the lane L11 and the history L2 of the self-position in the area within the meshes M11 and M12 on the high-precision map MD is illustrated, but the error used for optimizing the grid parameter Uij The area for calculating is not limited to this. For example, the parameter determination unit 13 calculates the sum of errors of each mesh Mmn (the error of the entire high-precision map MD) and optimizes the grid parameter Uij so as to minimize the error of the calculated high-precision map MD as a whole. May be used. Alternatively, the parameter determination unit 13 may optimize the grid parameter Uij so as to minimize an error in a part of the high-precision map MD.

  Further, the parameter determination unit 13 may optimize the grid parameter Uij for each grid point Pij by using errors in different areas in the vicinity of the grid points Pij in the high-precision map MD. For example, when optimizing the grid parameter U22 of the grid point P22 common to the meshes M11, M12, M21, and M22 shown in FIG. 3, errors in the meshes M11, M12, M21, and M22 are used. On the other hand, when optimizing the grid parameter U23 of the grid point P23 common to the meshes M12, M13, M22, and M23, errors in the areas of the meshes M12, M13, M22, and M23 are used.

  Moreover, the parameter determination part 13 may optimize the grid parameter Uij for every mesh Mmn using the difference | error of a mutually different area | region near each mesh Mmn of the high precision map MD. For example, when the grid parameters U22, U23, U32, and U33 of the grid points P22, P23, P32, and P33 that divide the mesh M22 shown in FIG. 3 are optimized, the meshes M11, M12, and M13 around the mesh M22 are optimized. , M21, M22, M23, M31, M32, and M33 area errors are used. On the other hand, when optimizing the grid parameters U23, U24, U33, U34 of the grid points P23, P24, P33, P34 that divide the mesh M23, the meshes M12, M13, M14, M22, M23 around the mesh M23, The error in the area of M24, M32, M33, and M34 is used.

  The parameter determination unit 13 determines grid parameters U11, U12, U13, P21, P22, and P23 of grid points P11, P12, P13, P21, P22, and P23 that divide the meshes M11 and M12, for example, as shown in FIG. To do. In FIG. 7, the positions of the grid points P11, P12, P13, P21, P22, and P23 corrected by using the grid parameters U11, U12, U13, P21, P22, and P23 (coordinate conversion) are schematically illustrated by broken lines. It shows. Grid points P12 and P22 shared by the adjacent meshes M11 and M12 before correction are also shared by the corrected meshes M11 and M12.

  The intra-mesh interpolation unit 14 shown in FIG. 1 configures a high-precision map MD in each mesh Mmn in units of mesh Mmn based on the position coordinates of each mesh Mmn and the grid parameter Uij determined by the parameter determination unit 13. The offset parameter (hereinafter referred to as “mesh parameter”) of the constituent point to be calculated is interpolated for each constituent point. As a mesh parameter interpolation method, for example, linear interpolation or the like can be used. The constituent points are, for example, points constituting a road structure such as lane boundaries on the high-precision map MD. The high-precision map MD in the two-dimensional space has two-dimensional coordinates, and the three-dimensional in the high-precision map in the three-dimensional space. Has coordinates.

  For example, as shown in FIG. 8, a case is considered where the mesh parameter U1 of the constituent point P1 in the mesh M11 is interpolated. Also in FIG. 8, the positions of the grid points P11, P12, P13, P21, P22, and P23 corrected by using the grid parameters U11, U12, U13, P21, P22, and P23 (coordinate conversion) are schematically illustrated by broken lines. It shows.

  In this case, the mesh parameter U1 of the constituent point P1 in the mesh Mmn is interpolated by linear interpolation based on the distances from the grid points P11, P12, P21, and P22 that define the mesh M11 and the grid parameters U11, U12, U21, and U22. To do. For example, the mesh parameter U1 can be interpolated based on the distances from the four grid points P11, P12, P21, and P22 and the grid parameters U11, U12, U21, and U22. For example, when linear interpolation is performed using only two grid points P11 and P12, the mesh parameter U1 can be interpolated by calculating a weighted average corresponding to the distance from the two grid points P11 and P12. .

  The grid parameters U11, U12, U21, and U22 of the grid points P11, P12, P21, and P22 that divide the constituent point P1 in the mesh M11 may be defined by a two-dimensional or three-dimensional vector or matrix. For example, the grid parameters U11, U12, U21, and U22 of the grid points P11, P12, P21, and P22 that define the constituent point P1 in the mesh M11 may be defined by a two-dimensional or three-dimensional translation amount. For example, when the grid parameters U11, U12, U21, and U22 shown in FIG. 8 are common and move in parallel, the mesh parameter U1 can also be interpolated by setting the value to move in parallel.

Further, the grid parameters U11, U12, U21, U22 of the grid points P11, P12, P21, P22 that divide the constituent point P1 in the mesh M11 may be defined by parameters of known coordinate conversion means such as affine transformation. . The affine transformation is a transformation that combines translation and linear transformation. For example, the affine transformation in the case where the position coordinates (x, y) of the constituent point P1 in the mesh M11 is transformed into (x ′, y ′) is expressed by the following equation (1).

  In equation (1), the linear transformation matrix parameters a, b, c, d and the translation vector parameters e, f are the coordinates of the grid points P11, P12, P21, P22 and the grid parameters U11, U12. , U21, U22 can be obtained by multiple regression analysis or the like.

  Further, as shown in FIG. 8, a case is considered where the mesh parameter U2 of the constituent point P2 on the boundary between the meshes M11 and M12, which is shared by the meshes M11 and M12, is calculated. In this case, the mesh parameter U2 is interpolated using an interpolation method such as wire diameter interpolation so that the adjacent meshes M11 and M12 have the same value. As a result, the position of the corrected component point P2 indicated by the broken line in FIG. 8 is located on the straight line connecting the corrected grid points P12 and P22 indicated by the broken line, so that the adjacent meshes M11 and M12 are continuous. Sexuality can be secured.

  Note that different interpolation methods may be used for the configuration point P1 in the mesh M11 and the configuration point P2 on the boundary line between the adjacent meshes M11 and M12. For example, different interpolation methods may be used between constituent points near the boundary lines of adjacent meshes M11 and M12 and other constituent points. FIG. 8 illustrates the case where the mesh parameters U1 and U2 of the constituent points P1 and P2 in the mesh M11 are calculated, but all the meshes M11 in the mesh M11 including the constituent points on the boundary of the adjacent meshes M11 and M12 are illustrated. The mesh parameters of the constituent points can be calculated in the same way. The mesh parameter of each constituent point in the mesh M11 can be constituted by a parameter based on a known coordinate transformation means such as a vector, a matrix, a parallel movement amount, an affine transformation or the like. The mesh parameters in the other mesh Mmn can be calculated in the same manner as the mesh parameters in the mesh M11.

  The map correction unit 15 shown in FIG. 1 uses the grid parameters Uij determined by the parameter determination unit 13 and the mesh parameters calculated by the intra-mesh interpolation unit 14 to determine the component points in the grid points Pij and the mesh Mmn. By converting the position coordinates, the high-precision map MD is deformed and corrected. For example, the map correction unit 15 transforms the high-precision map MD shown in FIG. 3 as schematically shown in FIG. In FIG. 9, the boundary of the mesh Mmn is continuous, and the continuity of the road etc. which straddles the adjacent mesh Mmn can be ensured. The high-precision map MD corrected by the map correction unit 15 is stored in the map storage unit 20.

  The vehicle control unit 16 illustrated in FIG. 1 generates a planned travel route of the host vehicle based on the high-precision map MD before or after correction by the map correction unit 15. For example, the vehicle control unit 16 specifies the position of the host vehicle on the high-accuracy map MD, and generates a scheduled travel route so as to be drawn in the lane based on the position of the host vehicle. The scheduled travel route may be generated so as to pass through the center in the lane. The vehicle control unit 16 calculates the control amount of the actuator 6 so that the host vehicle travels on the planned travel route. The calculated control amount is transmitted to the actuator 6.

  The vehicle control unit 16 may perform automatic driving that automatically travels without involving the occupant, or may perform automatic driving that controls at least one of driving, braking, and steering. That is, in this specification, the automatic driving includes a case where all control of driving, braking, and steering of the host vehicle is executed without involving an occupant (driver), and at least driving, braking, and steering of the host vehicle are performed. This includes the case where one control is performed. The automatic driving may be lane departure prevention control using a high-precision map MD.

  The vehicle control unit 16 may display the high-precision map MD on the display 7. Alternatively, the vehicle control unit 16 may present guidance information or the like resulting from the high-precision map MD to the driver on an information presentation device such as the display 7 or another speaker.

  The actuator 6 controls traveling of the host vehicle in accordance with a control signal from the vehicle control unit 16. The actuator 6 can be composed of, for example, a drive actuator, a brake actuator, and a steering actuator. The drive actuator constituting the actuator 6 is composed of, for example, an electronically controlled throttle valve, and controls the accelerator opening of the host vehicle based on a control signal from the vehicle control unit 16. The brake actuator that constitutes the actuator 6 is composed of, for example, a hydraulic circuit, and controls the braking operation of the brake of the host vehicle based on a control signal from the vehicle control unit 16. A steering actuator constituting the actuator 6 controls the steering of the host vehicle based on a control signal from the vehicle control unit 16.

  The display 7 is, for example, a liquid crystal display (LCD). The display 7 can display the high-precision map MD based on the control signal from the processing circuit 1.

  Next, an example of the map correction method according to the embodiment of the present invention will be described with reference to the flowchart of FIG.

  In step S11, the mesh setting unit 11 reads the high accuracy map MD from the map storage unit 20, and sets the grid points Pij (i = 1 to 4, j = 1 to 5) on the high accuracy map MD. In step S12, the mesh setting unit 11 sets meshes Mmn (m = 1 to 3, n = 1 to 4) partitioned into polygons such as rectangles at the grid points Pij.

  In step S13, the parameter determination unit 13 sets grid parameters Uij (i = 1 to 4, j = 1 to 5) for each grid point Pij on the high-precision map MD. In addition, the parameter determination part 13 may set arbitrary values (for example, 0) as an initial value of the grid parameter Uij, when the high precision map MD is before correction | amendment.

  In step S <b> 14, the intra-mesh interpolation unit 14 calculates the mesh parameters of the constituent points in each mesh Mmn based on the grid parameter Uij set by the parameter determination unit 13.

  In step S15, the map correction unit 15 uses the grid parameters Uij determined by the parameter determination unit 13 and the mesh parameters calculated by the intra-mesh interpolation unit 14 to determine the component points in each grid point Pij and mesh Mmn. By converting the position coordinates, the high-precision map MD is deformed and corrected.

  In step S15, the map correction unit 15 determines whether or not the correction of the high-precision map MD has been completed. For example, the number of corrections of the high-precision map MD by the map correction unit 15 is counted, and when the number of corrections is equal to or greater than a predetermined threshold, it is determined that the correction of the high-precision map MD is completed, and the number of corrections of the high-precision map MD is predetermined. It may be determined that the correction of the high-precision map MD is not completed when the threshold value is less than the threshold value. Alternatively, it is determined that the correction of the high-precision map MD has been completed when the sum of errors of the high-precision map MD is less than a predetermined threshold, and high when the sum of errors of the high-precision map MD is greater than or equal to a predetermined threshold. It may be determined that the correction of the accuracy map MD has not been completed. If it is determined that the correction of the high-precision map MD has been completed, the processing is completed. On the other hand, if it is determined that the correction of the high-precision map MD has not been completed, the process proceeds to step S16.

  In step S16, the self-position calculation unit 12 performs matching between the self-position detected by the GPS receiver 3 and the high-precision map MD corrected by the map correction unit 15, and the self-position calculation unit 12 on the corrected high-precision map MD. A history of the vehicle position (self-position) is calculated. Note that if the road structure to be compared with the corrected high-precision map MD is a road structure other than the history of the self-location, step S16 may be omitted.

  In step S17, the parameter determination unit 13 calculates the error of the corrected high-precision map MD based on the actual road structure such as the center line of the lane corresponding to the history of the self position calculated by the self position calculation unit 12. The grid parameter Uij of each grid point Pij is corrected (optimized) so as to minimize. Thereafter, returning to the procedure of step S14, the intra-mesh interpolation unit 14 corrects the mesh parameters of the constituent points in each mesh Mmn based on the grid parameters Uij corrected by the parameter determination unit 13.

  As described above, according to the embodiment of the present invention, a plurality of grid points Pij are set on the high-precision map MD that is an example of a map. Then, the road structure around the host vehicle is detected, the grid parameter Uij of the grid point Pij is calculated (optimized) based on the detected road structure, and the high-precision map MD is corrected using the calculated grid parameter Uij. Thereby, the error of the high accuracy map MD can be locally corrected while maintaining the continuity of the high accuracy map MD. For this reason, compared with the comparative example shown in FIG.4 and FIG.5, the continuity of the boundary part of the mesh Mmn can be maintained. Therefore, a high-accuracy map MD suitable for application to an automatic driving system can be generated without causing a break in the road straddling the adjacent mesh Mmn.

  Further, the high-accuracy map MD is divided into a plurality of meshes Mmn partitioned with a plurality of grid points Pij as vertices, and grid parameters Uij of the respective grid points Pij are calculated. The grid point Pij of the divided mesh Mmn (for example, the grid point P12 of the mesh M11) is the same as any grid point Pij of the adjacent mesh Mmn (for example, the grid point P12 of the mesh M12). Thereby, the error of the high-precision map MD can be locally corrected while maintaining the continuity of the boundary portion of the mesh Mmn defined by the grid points Pij.

  Any two grid points Pij (for example, two grid points P12 and P22 of the mesh M11) among the grid points Pij of the divided mesh Mmn are any two grid points Pij (for example, the mesh M12 of the adjacent mesh Mmn). The two grid points P12 and P22) are the same. Thereby, the error of the high-precision map MD can be locally corrected while maintaining continuity between the boundary portions of the mesh Mmn divided by the grid points Pij, that is, between the two grid points Pij.

  The mesh parameter on the boundary of the divided mesh Mmn is the same as the mesh parameter on the boundary in the adjacent mesh Mmn. Thereby, the error of the high-precision map MD can be locally corrected while maintaining continuity between the boundary portions of the mesh Mmn divided by the grid points Pij, that is, between the two grid points Pij.

  Further, the high-precision map MD is divided into a plurality of meshes Mmn that are partitioned with a plurality of grid points Pij as vertices, and the mesh parameters of the constituent points in the mesh Mmn are calculated using the grid parameters Uij of each grid point Pij. To do. As a result, the grid parameter Uij of each grid point Pij can be optimized and the high-precision map MD divided into the meshes Mmn can be corrected. Can be corrected appropriately.

  Further, the mesh parameters of the constituent points on the boundary of the adjacent mesh Mmn are interpolated so that the adjacent meshes Mmn are the same. Thereby, since the composing point shared by the adjacent mesh Mmn before the correction is shared by the adjacent mesh Mmn after the correction, the continuity of the adjacent mesh Mmn boundary can be ensured.

  Furthermore, the grid parameter Uij of each grid point Pij is defined by a two-dimensional or three-dimensional vector or matrix. Thereby, since the grid parameter Uij is expressed by a two-dimensional or three-dimensional vector or matrix, not only the two-dimensional position information but also the three-dimensional position information can be corrected.

  Furthermore, the grid parameter Uij of each grid point Pij is defined by a two-dimensional or three-dimensional translation amount. Accordingly, since the grid parameter Uij is expressed by a vector amount, the mesh Mmn can be translated and distorted.

  Furthermore, the grid parameter Uij of each grid point Pij is defined by parameters of coordinate transformation means such as affine transformation. Thereby, since the grid parameter Uij is represented by a transformation matrix, in addition to the above-described parallel movement and deformation, deformation with a greater degree of freedom such as rotation and contraction can be performed.

  Furthermore, the grid parameter Uij of each grid point Pij is optimized by comparing the self-position history obtained from the sensor such as the GPS receiver 3 with the corresponding position on the high-precision map MD. As a result, the high-accuracy map MD and the actual measurement data of sensors such as the GPS receiver 3 are compared and optimized, so that the high-accuracy map MD can be corrected along the actual measurement data. Furthermore, the grid parameter Uij of each grid point Pij is optimized so as to minimize the difference between the self-position history obtained from the sensor such as the GPS receiver 3 and the corresponding position on the corrected high-precision map MD. To do. Thereby, the difference between the history of the self-position and the corresponding position on the corrected high-precision map MD can be minimized.

  Furthermore, the grid parameter Uij of each grid point Pij is optimized by comparing the road structure detected by the vehicle sensor 4 and the position on the high-precision map MD corresponding to the road structure. Thereby, the high-accuracy map MD and the actual measurement data of the sensors such as the vehicle sensor 4 are compared and optimized, so that the high-accuracy map MD can be corrected along the actual measurement data. Furthermore, the grid parameter Uij of each grid point Pij is optimized so as to minimize the difference between the road structure detected by the vehicle sensor 4 and the position on the corrected high-precision map MD corresponding to the road structure. Thereby, the difference between the road structure detected by the vehicle sensor 4 and the corresponding position on the corrected high-precision map MD can be minimized.

  Furthermore, the grid parameter Uij of each grid point Pij is optimized by comparing the lane boundary line or road boundary line as the road structure detected by the vehicle sensor 4 and the corresponding position on the high-precision map MD. . Thereby, the high-accuracy map MD and the actual measurement data of the sensors such as the vehicle sensor 4 are compared and optimized, so that the high-accuracy map MD can be corrected along the actual measurement data. Further, the grid parameter Uij of each grid point Pij is optimized so as to minimize the difference between the lane boundary line or road boundary line detected by the vehicle sensor 4 and the corresponding position on the corrected high-precision map MD. To do. Thereby, the difference between the lane boundary line or road boundary line detected by the vehicle sensor 4 and the corresponding position on the corrected high-precision map MD can be minimized.

  Furthermore, since the high-precision map MD is 3D data, and the grid points are set in the 3D space, the mesh Mmn is defined as a polyhedron in the 3D space. Can also be appropriately corrected.

  Furthermore, by using the high-precision map MD including road lane information as a correction target, the high-precision map MD suitable for application to an automatic driving system can be corrected appropriately while maintaining the continuity of the mesh Mmn. .

(Modification)
Next, a modification of the map correction method according to the embodiment of the present invention will be described with reference to the flowchart of FIG.

  In step S21, the mesh setting unit 11 reads the high-precision map MD from the map storage unit 20, and sets the grid points Pij on the high-precision map MD. In step S22, the mesh setting unit 11 sets a mesh Mmn that is partitioned into polygons such as rectangles at the grid points Pij.

  In step S23, the parameter determination unit 13 sets a grid parameter Uij for each grid point Pij on the high-precision map MD. In addition, the parameter determination part 13 may set arbitrary values (for example, 0) as an initial value of the grid parameter Uij, when the high precision map MD is before correction | amendment.

  In step S24, the parameter determination unit 13 corrects (optimizes) the grid parameter Uij of each grid point Pij so as to minimize the error of the corrected high-precision map MD based on the actual road structure. After step S24, the procedure for calculating the mesh parameters by the intra-mesh interpolation unit 14 corresponding to step S14 shown in FIG. 10 is omitted, and the process proceeds to step S25.

  In step S <b> 25, the map correction unit 15 deforms and corrects the high-precision map MD based on the grid parameter Uij determined by the parameter determination unit 13. In step S26, the map correction unit 15 determines whether or not the correction of the high-precision map MD has been completed. If it is determined that the correction of the high-precision map MD has been completed, the processing is completed. On the other hand, if it is determined that the correction of the high-accuracy map MD has not been completed, the procedure returns to step S24, and the optimization of the grid parameters Uij for each grid point Pij is repeated.

  After step S24, the procedure corresponding to step S14 shown in FIG. 10, that is, the intra-mesh interpolating unit 14 determines the component points in each mesh Mmn based on the grid parameter Uij modified by the parameter determining unit 13. A procedure for modifying the mesh parameters may be performed. In this case, in step S25, the map correction unit 15 deforms and corrects the high-precision map MD based on the grid parameter Uij determined by the parameter determination unit 13 and the mesh parameter calculated by the intra-mesh interpolation unit 14. To do.

(Other embodiments)
As mentioned above, although this invention was described by embodiment, it should not be understood that the statement and drawing which form a part of this indication limit this invention. From this disclosure, various alternative embodiments, examples and operational techniques will be apparent to those skilled in the art.

  For example, as shown in FIG. 12, the parameter determination unit 13 shown in FIG. 1 divides each mesh Mmn for each mesh Mmn (m = 1 to 2, n = 1 to 2), and grid points Pij (i = Consider the case of determining grid parameters Uij (i = 1 to 3, j = 1 to 3) of 1 to 3, j = 1 to 3). Since the high-accuracy map is deformed for each of the meshes M11, M12, M21, and M22, the grid point P22a that partitions the mesh M21 and the grid point P22b that partitions the mesh M22 are separated and individually determined, and the adjacent mesh M21, The boundary of M22 is discontinuous.

  On the other hand, as shown in FIG. 13, the parameter determination unit 13 determines between the adjacent meshes M21 and M22 based on the position coordinates of the grid points P22a and P22b and the grid parameters Uij held by the grid points P22a and P22b. The grid point P22 to be shared may be determined. For example, the grid point P22 may be set to be a midpoint between the grid point P22a and the grid point P22b illustrated in FIG.

  As described above, the present invention naturally includes various embodiments not described herein. Therefore, the technical scope of the present invention is defined only by the invention specifying matters according to the scope of claims reasonable from the above description.

DESCRIPTION OF SYMBOLS 1 ... Processing circuit 2 ... Memory | storage device 3 ... GPS receiver 4 ... Vehicle sensor 5 ... Ambient sensor 6 ... Actuator 7 ... Display 11 ... Mesh setting part 12 ... Self-position calculation part 13 ... Parameter determination part 14 ... Inter-mesh interpolation part 15 ... map correction unit 16 ... vehicle control unit 20 ... map storage unit

Claims (17)

In a map correction method for correcting map data based on a road structure,
A plurality of grid points are set on the map data,
Detecting the road structure around the vehicle,
Calculating an offset parameter of each grid point so that a difference between the detected road structure and the road structure of the map data is reduced;
The map correction method, wherein the map data is corrected using an offset parameter of each grid point. Dividing the map data into a plurality of meshes partitioned by the plurality of grid points;
The map correction method according to claim 1, wherein any grid point of the divided mesh is the same as any grid point of an adjacent mesh. The map correction method according to claim 2, wherein any two grid points of the divided mesh are the same as any two grid points of adjacent meshes. The map correction method according to claim 2 or 3, wherein the offset parameter on the boundary of the divided mesh is the same as the offset parameter on the boundary of an adjacent mesh.   4. The map correction method according to claim 2, wherein the offset parameters of the constituent points on the adjacent mesh boundaries are interpolated to make the adjacent meshes the same. 5.   The map correction method according to claim 1, wherein the offset parameter of each grid point is defined by a vector or a matrix.   The map correction method according to claim 1, wherein the offset parameter of each grid point is defined by a parallel movement amount.   The map correction method according to claim 1, wherein the offset parameter of each grid point is defined by an affine transformation parameter.   The offset parameter of each grid point is optimized by comparing the history of the position of the vehicle and the corresponding position on the map data. The map correction method described.   The offset parameter of each grid point is optimized so as to minimize the difference between the history of the position of the vehicle and the corresponding position on the corrected map data. Map correction method.   The offset parameter of each grid point is optimized by comparing the road structure detected by a vehicle sensor and a position on the map data corresponding to the road structure. 9. The map correction method according to any one of 8 above.   The offset parameter of each grid point is optimized so as to minimize a difference between the road structure detected by the vehicle sensor and a position on the map data corresponding to the road structure. Item 12. The map correction method according to Item 11.   By comparing the lane boundary line or road boundary line detected by the vehicle sensor and the position on the map data corresponding to the lane boundary line or road boundary line as the road structure, each grid point The map correction method according to claim 11 or 12, wherein an offset parameter is optimized.   The offset parameter of each grid point is optimized so as to minimize the difference between the lane boundary line or road boundary line detected by the vehicle sensor and the corresponding position on the corrected map data. The map correction method according to claim 13.   The map correction method according to claim 1, wherein the map data is three-dimensional data, and the grid points are set in a three-dimensional space.   The map correction method according to claim 1, wherein the map data is high-precision map data including road lane information. A map correction device for correcting map data based on a road structure,
A sensor for detecting the road structure around the vehicle;
A plurality of grid points are set on the map data, and an offset parameter of each grid point is calculated so that a difference between the detected road structure and the road structure of the map data is reduced, and each grid point is calculated. And a controller that corrects the map data using a point offset parameter.

Images