Method and apparatus for fusing digital maps

Abstract

The invention relates to a method and an apparatus for fusing digital maps. The method comprises determining, for a partial area of a first and a second digital map, a set of pairwise associations between first objects from the first digital map and second objects from the second digital map, respectively, and determining a set of displacements by which the sets of first and second objects of the associations coincide with each other. Furthermore, the method comprises determining, for the partial area of the first and the second digital map, a deformation function for deforming a sub-grid from the set of displacements, and transferring map information from the partial area of the second digital map to the partial area of the first digital map by means of the deformation function.

Description

Technical Field

[0001] The present invention relates to a method and a corresponding apparatus designed to integrate information from multiple digital maps, in particular to enable the comfortable and safe automated operation of vehicles. Background Technology

[0002] The vehicle may have one or more driving functions, each configured to automatically guide the vehicle longitudinally and / or laterally. The operation of one or more driving functions may be based on a digital map of the road network traversed by the vehicle. Furthermore, sensor data from one or more environmental sensors of the vehicle, such as cameras or radar sensors, are typically taken into account during the operation of the one or more driving functions.

[0003] Digital maps used to operate driving functions can be inaccurate, which may lead to corresponding inaccuracies in the vehicle's automatic longitudinal and / or lateral guidance. For example, map inaccuracies related to the lanes a vehicle travels through can result in the vehicle not always being positioned in the center of the lane. Summary of the Invention

[0004] This article relates to the technical tasks of improving the accuracy of digital maps, in particular to enable comfortable and accurate automatic longitudinal and / or lateral guidance for vehicles.

[0005] The task is solved according to the invention. Advantageous embodiments are further described in the specification. It should be noted that additional features of dependent claims of an independent claim can form independent inventions, independent of the combination of all features of the independent claim, either without the features of the independent claim or only in combination with a subset of the features of the independent claim, which can be the subject of the independent claim, divisional application, or subsequent application. This applies in the same manner to the technical teachings described in the specification, which can form inventions independent of the features of the independent claim.

[0006] According to this description, a method for fusing map information from a first digital map and a second digital map is described. The digital map may correspond to a total grid with multiple grid points. The total grid may be divided into multiple sub-grids for corresponding partial regions of the digital map. Here, different sub-grids or different partial regions may partially overlap if necessary. Each grid point may be associated with a point or coordinate in the corresponding digital map. Therefore, deformation of the grid used for the map may lead to a corresponding deformation of the map. In particular, displacement of each grid point may lead to a corresponding displacement of the associated point or coordinate in the corresponding map.

[0007] The method includes determining a set of paired associations between a first object from the first digital map and a second object from the second digital map, respectively, for portions of first and second digital maps. The set of first objects can be determined within a portion of the first map, and for each first object, a corresponding second object exists within a portion of the second map. Exemplary objects include road sections, landmarks, and traffic signs. The associations can be implemented using an association algorithm.

[0008] Sub-mesh for a portion of the area may, if necessary, have a core region surrounded by the edge region of the sub-mesh. In a preferred example, the associated defined set includes only the first and / or second objects arranged in the core region of the sub-mesh, and preferably excludes the first and / or second objects arranged in the edge region of the sub-mesh. The core region may have 50% or less of the grid points of the entire sub-mesh.

[0009] Furthermore, the method includes determining a set of displacements (specifically displacement vectors) by which the sets of associated first and second objects coincide. For each individual pair of associations, a displacement vector can be determined by which the second object is moved to a point or coordinate of the first object.

[0010] The method also includes determining a deformation function for deforming a subgrid for a portion of the first and second digital maps based on the set of displacements. The deformation function can be determined here by optimizing an optimization problem. The optimization problem may include at least one term that depends on the (mean and / or squared) distance between the set of second objects in the deformed portion of the second digital map and the associated set of first objects in the undeformed portion of the first digital map.

[0011] The deformation function is preferably a space-morphing function. A subgrid typically comprises a matrix of grid points. The deformation function can include a matrix of spatial curvature vectors for the corresponding grid points. Here, within the scope of the deformation function, each grid point can be locally moved according to the spatial curvature vector associated with that grid point.

[0012] For each grid point in the matrix of grid points, the deformation function may include a core function that describes how the movement of the corresponding grid point with respect to the spatial curvature vector of the corresponding grid point affects one or more grid points in the surrounding environment of the corresponding grid point.

[0013] The deformation function can be applied to (arbitrary) grid points of a matrix of grid points. x 1 Corresponding description of the moving grid points f sm(x 1 ) For example, moving grid points can be determined as

[0014] Where v i These are the grid points used for subgrids. x i The spatial curvature vector; and where, It is the core function.

[0015] The core function is preferably constructed such that the value With grid point x1 and grid point x i It decreases as the distance between them increases. An exemplary core function is: or

[0016] It has an influence parameter σ, which describes a specific grid point. x i Towards grid points x 1 The degree of movement.

[0017] In a corresponding manner, deformation functions can be determined for different subgrids of the overall grid so as to enable the fusion of map information from the entire first and second digital maps.

[0018] This method can be applied to a series of sub-grids for corresponding partial regions of the first and second digital maps, respectively, including:

[0019] • For corresponding portions of the first and second digital maps, determine a set of paired associations between a first object from the first digital map and a second object from the second digital map, respectively.

[0020] • Determine the set of displacements, where the sets of first and second objects associated with each other coincide through displacement; and

[0021] • Based on the separately determined sets of displacements, deformation functions are determined for the corresponding partial regions of the first and second digital maps to deform the corresponding subgrids.

[0022] Here, directly consecutive subgrids in this series can partially overlap. Furthermore, when determining the deformation function of the directly following subgrid, the deformation function determined for the previous subgrid can be considered. Therefore, particularly efficient and accurate deformation of the first and / or second digital maps can be achieved to fuse map information from both maps.

[0023] The method also includes using a transformation function to transfer map information from a corresponding region of the second digital map to a corresponding region of the first digital map. Here, the map information may include objects recorded in a region of the second digital map but not in a region of the first digital map. Therefore, the information content of the first digital map can be improved in a more efficient and accurate manner.

[0024] The method may further include: determining a reliability index for the reliability of the determined deformation function for the corresponding partial area. The transfer of map information from the corresponding partial area of ​​the second digital map to the corresponding partial area of ​​the first digital map can be achieved in a particularly stable manner based on the reliability index.

[0025] According to another aspect, a software (SW) program is described. The software program may be designed to execute on a processor, and thereby perform the methods described in this document.

[0026] According to another aspect, a storage medium is described. The storage medium may include software programs designed to execute on a processor and thereby perform the methods described in this document.

[0027] According to one aspect, an apparatus for fusing map information from a first digital map and a second digital map is described. The apparatus is designed to determine, for partial regions of the first and second digital maps, a set of pairwise associations between a first object from the first digital map and a second object from the second digital map, respectively. Furthermore, the apparatus is designed to determine a set of displacements, wherein the sets of associated first and second objects coincide with each other through displacements. The apparatus is also designed to determine, based on the set of displacements, a deformation function for deforming a subgrid for partial regions of the first and second digital maps, and to transfer map information from a partial region of the second digital map to a partial region of the first digital map using the deformation function.

[0028] It should be noted that the various aspects described in connection with this method, especially the claims described in connection with this method, can also be used as corresponding features in the apparatus.

[0029] According to another aspect, a (road) motor vehicle (especially a passenger car, truck, bus, or motorcycle) is described, which includes the devices described in this document.

[0030] It should be noted that the methods, apparatuses, and systems described in this document can be used not only individually but also in combination with other methods, apparatuses, and systems described in this document. Furthermore, any aspect of the methods, apparatuses, and systems described in this document can be combined with each other in various ways. In particular, the features of the claims can be combined with each other in various ways. Additionally, features listed in parentheses should be understood as optional features. Attached Figure Description

[0031] Furthermore, the present invention is described in detail with the aid of embodiments.

[0032] Figure 1 Exemplary components of the vehicle are shown;

[0033] Figure 2 An exemplary digital map for a specific local area of ​​the Earth's surface is shown;

[0034] Figure 3a An exemplary total grid for digital maps is shown;

[0035] Figure 3b An exemplary deformable subgrid is shown for deforming map fragments; and

[0036] Figure 4 A flowchart illustrating an exemplary method for determining a merged digital map is shown. Detailed Implementation

[0037] As mentioned above, this document relates particularly to automatic longitudinal and / or lateral guidance of vehicles to improve the comfort and / or safety of vehicle operation. In this case, Figure 1 An exemplary vehicle 100 is shown, which includes one or more environmental sensors 102. The exemplary environmental sensors 102 are cameras, radar sensors, laser rangefinders, ultrasonic sensors, etc. The one or more environmental sensors 102 are designed to detect environmental data (i.e., sensor data) related to the environment of the vehicle 100.

[0038] Vehicle 100 also includes a position sensor 104 designed to determine, with the aid of a Global Navigation Satellite System (GNSS), such as GPS, position data (i.e., sensor data) related to the position of vehicle 100 in a world coordinate system and / or within a digital map.

[0039] In addition, vehicle 100 includes one or more vehicle sensors 103 designed to determine sensor data related to one or more measurement parameters, such as driving speed (in the longitudinal, lateral and / or height directions), steering angle and / or rotation rate (around the longitudinal axis, lateral axis and / or height axis), which are capable of determining the relative motion of vehicle 100 over time (especially between two consecutive points in time) based on mileage.

[0040] In addition, vehicle 100 may include an acceleration sensor 105 designed to determine measurements related to the (three-dimensional) acceleration vector of vehicle 100. For example, acceleration sensor 105 may include an inertial measurement unit (IMU) or part of an IMU.

[0041] Therefore, when driving on the road, vehicle 100 can detect sensor data, which includes, for example:

[0042] • Position measurement values ​​from GNSS sensor 104;

[0043] • Odometer measurements related to the motion of vehicle 100; and / or

[0044] • Camera images are used to identify landmarks in the environment surrounding vehicle 100.

[0045] Sensor data can be detected at a series of time points while the vehicle is traveling along the road. This sensor data can be used, for example, by the (control) device 101 of vehicle 100 to locate the vehicle 100 on the road. For this purpose, a SLAM (Simultaneous Localization and Mapping) algorithm can be used.

[0046] The position measurements from GNSS sensor 104 can be used to determine the position of vehicle 100 within a digital map. Information that can be used for automatic longitudinal and / or lateral guidance, such as the direction of roads traveled by vehicle 100 and / or the location of one or more landmarks, and / or the location and type of traffic signs, can be extracted from the digital map. The digital map may have inaccuracies and / or may lack information (e.g., unspecified traffic signs).

[0047] Multiple different digital maps may be available for specific local areas of the Earth's surface. For example, a map provider may offer a digital map built using sensor data from measuring vehicles. Such a digital map may have relatively high accuracy for primary traffic routes. On the other hand, such a digital map may be unavailable for secondary traffic routes, and / or the digital map may be outdated, resulting in incorrect and / or unusable information stored within it.

[0048] Additional digital maps may be available for specific local areas, and these additional digital maps are determined based on environmental data from multiple vehicles 100 and thus contain relatively recent information. On the other hand, such digital maps typically have lower accuracy.

[0049] Figure 2An overlay of a first digital map 210 and a second digital map 220 for a specific local area is shown, wherein the first digital map 210 has first objects 211, 212, and the second digital map 220 has second objects 221, 222, 223. Exemplary objects are: road segments and / or road marking segments, landmarks, traffic signs, etc.

[0050] As from Figure 2 As can be seen, the different digital maps 210 and 220 are generally not consistent. In particular, the same and / or corresponding objects 211, 221 or 212, 222 may be recorded in different locations on different digital maps 210 and 220, making direct fusion of information from two digital maps 210 and 220 generally not easily achievable. In order to still enable accurate fusion of information from multiple digital maps 210 and 220, one or both maps 210 and 220 can be modified, wherein the modification is intended to make the objects 211, 221 or 212, 222 recorded in the first digital map 210 and the second digital map 220 overlap as accurately as possible. Therefore, information related to object 223 recorded only in one of the two maps 210 and 220 can be integrated into the fused map to provide a fused map with higher information content.

[0051] In the first step, an association 230 can be determined between the set of first objects 211, 212 from the first digital map 210 and the set of second objects 221, 222 from the second digital map 220. The association 230 can describe the pairing of corresponding objects 211, 221 or 212, 222 from the two digital maps 210, 220 respectively. For example, the association 230 can be determined using a nearest neighbor algorithm, a Hungarian matching algorithm, and / or a template matching algorithm.

[0052] For each association 230, a displacement vector 231 can be determined, and the two objects 211 and 212 of the corresponding association 230 can overlap through the displacement vector. Therefore, the set of displacement vectors 231 can be determined for the set of associations 230.

[0053] In addition to the corresponding sets of associated objects 211, 212 or 221, 222, the first digital map 210 and / or the second digital map 220 may each have one or more objects 223 for which no corresponding object can be found in the other digital maps 210, 220. The measures described in this document allow the one or more objects 223 to be precisely incorporated into the merged map.

[0054] After determining the association 230 of the object sets from the two digital maps 210, 220, deformation of the first digital map 210 and / or the second digital map 220 can be implemented, wherein the deformation aims to reduce, in particular minimize, the distance dimension of the distance between the sets of associated objects 211, 212 or 221, 222. The deformation of one or both digital maps 210, 220 can be implemented in an effective and accurate manner by means of one or more deformation functions, as illustrated by way of example in Figure 3a and 3b .

[0055] Figure 3a shows an exemplary overall grid 300 for digital maps 210, 220, wherein the overall grid 300 has a plurality of grid points 301. The grid points 301 may have a specific distance from each other, for example a distance between 1 meter and 5 meters. The overall grid 300 may comprise a matrix of grid points 301.

[0056] A sub-grid 310 can be observed within the overall grid 300, wherein the sub-grid 310 has a core region 311 and an edge region 312. The core region 311 is completely surrounded by the edge region 312. For example, the sub-grid 310 may have N×M grid points 301, and the core region 311 of the sub-grid 310 may have Q×R grid points 301, wherein Q<N and R<M. In a preferred example, Q=N / 3 and R=M / 3. The core region 311 may be arranged in the middle of the sub-grid 310, and the edge region 312 may have the shape of a rectangular frame surrounding the core region 311. The edge region 312 may have a width of (N-Q) / 2 or (M-R) / 2. In a preferred example, the width of the edge region 312 is N / 3 or M / 3.

[0057] Deformation can be implemented separately for each individual sub-grid 310, so as to reduce the value of the distance dimension of the set of associated objects 211, 221 arranged in the corresponding sub-grid 310, in particular arranged in the core region 311 of the corresponding sub-grid 310. The deformation of the sub-grid 310 can herein be implemented in an effective and accurate manner by means of a deformation function. As mentioned above, the sub-grid 310 may correspond to a matrix having N×M points 301, wherein N and M have values between 5 and 20, for example. Each individual grid point 301 can be associated with a deformation function respectively, wherein the deformation function implements displacement of the corresponding grid point 301 and the grid points surrounding the corresponding grid point 301. Herein, each individual deformation function is constructed for each individual grid point 301 such that no displacement is implemented at the edge 313 of the sub-grid 310. By this condition, it can be achieved that the deformed map segment of the digital maps 210, 220 corresponding to the deformed sub-grid 310 also has no deformation at the edge of the deformed map segment.

[0058] The deformation function for a specific grid point 301 can be defined by a (spatial) curvature vector 322. This curvature vector 322 can correspond to rotation and / or translation (where translation is in...). Figure 3b (Example shown in the figure). The curvature vector 322 for a particular grid point 301 can be scaled with respect to the grid points surrounding the particular grid point 301 by a particular scaling factor K, wherein the scaling factor K typically decreases as the distance from the particular grid point 310 increases.

[0059] The displacement of grid point 301 (based on curvature vector 322) thus causes the displacement of the corresponding grid points around grid point 301 in relation to the displacement of grid point 301. Since each grid point 301 is also arranged around the other grid points 301 of subgrid 310, the position of the movement of grid point 301 is generated from the overlap of the displacements of all grid points 301 of subgrid 310.

[0060] Each grid point 301 of subgrid 310 can be summarized as a matrix.

[0061] Where n is the number of grid points 301 in subgrid 310, and where x i (in Each grid point 301 is represented by a coordinate (e.g., in a two-dimensional Cartesian coordinate system). The curvature vectors 322 used for each grid point 301 can be summarized into a matrix in a corresponding manner.

[0062] Moving grid points of subgrid 310 f sm (x) 301 can be described (when using the deformation function that implements the translation of each grid point 301) by the subsequent linear matrix equations.

[0063] in K It is a matrix containing scaling factors for each deformation function and 301 for each grid point, where

[0064] and

[0065] in It is used for grid points The scaling factor of the deformation function, where , and among them x j Grid points Surrounding grid points (of which) i =1, ..., m). Here... n It is the number of grid points 301 that are moved within subgrid 310. m It is the number of adjacent grid points 301 (where m is usually the number of adjacent grid points 301). <n)。

[0066] Used to determine scaling factor The deformation functions, especially the so-called kernel functions, can be positive definite. An example kernel function is: or

[0067] in

[0068] • σ is an influencing parameter that describes the specific grid point 301 towards specific grid points Grid points around 301 The extent of 301's movement; this influencing parameter can be determined in advance if necessary;

[0069] • It is the moving grid point 301; and

[0070] • Grid points The surrounding grid point 301.

[0071] Therefore, for each grid point of subgrid 310 301 can be used to determine the curvature of space (especially displacement). Then, the spatial curvature of all grid points 301 (especially through displacement) is considered. A deformed subgrid 310 is generated, from which corresponding deformed map segments of digital maps 210 and 220 can be determined.

[0072] As previously mentioned, different maps 210 and 220 are generally not geometrically consistent. Furthermore, the differences between maps 210 and 220 are neither known nor quantified prior to the fact. Therefore, it is generally unknown whether the differences between the two maps 210 and 220 are large or small at a given location. Therefore, for the positioning of vehicle 100, it is preferable to use only a single map. For safety reasons, map changing during the movement of vehicle 100 is generally not possible. Therefore, only data from map 210 used for positioning is accessed. Information from other maps 220 cannot be used.

[0073] The measures described in this document enable the measurement of geometric differences between different digital maps 210 and 220. Furthermore, different digital maps 210 and 220 (depending on trust level) can be merged into a single map, and / or maps 210 and 220 can be geometrically superimposed.

[0074] The merging of maps 210 and 220 can be achieved through spatial deformation. This allows space to be curved independently of maps 210 and 220, and thus geometrically modifies maps 210 and 220 and / or thereby transfers the two maps 210 and 220 to each other. Because the spatial curvature caused by spatial deformation is achieved locally, a natural and smooth transition is produced between the modified and unmodified map fragments.

[0075] The spatial curvature required to compensate for the overlap of two maps 210 and 220 can be interpreted as a scale of trust in maps 210 and 220. If the two maps 210 and 220 are consistent, or can become consistent with a small spatial curvature, then the degree of trust in the resulting curved (fused) map is generally higher than the degree of trust when space must be relatively strongly curved in order for maps 210 and 220 to transfer to each other.

[0076] Within the scope of the fusion, the first map 210 can be used as an undistorted "master map." On the other hand, the second map 220 can be used as a distorted "correction map" to enable the integration of information from the second map 220 into the first map 210. Specifically, additional information from the second map 220 that is not yet included in the master map (e.g., object 223) can be curved and continuously incorporated into the master map. The topology of the master map remains unchanged. The parameters required for correction (especially the spatial curvature 322) can be determined in the so-called "regenerative kernel Hilbert space" (RKHS).

[0077] As previously mentioned, in the first step, points 211 and 212 with coordinates from the first digital map 210 (main map) can be associated with corresponding points 221 and 222 from the second digital map 220. Typically, not all elements exist in both maps 210 and 220, so only elements (i.e., objects) that can be found in both maps 210 and 220 are associated. Objects with a single semantic meaning (e.g., a traffic light with an identifier) ​​can be associated using semantics (e.g., using the identifier). Geometric objects without semantic meaning can be associated using an association algorithm, such as...

[0078] • Nearest Neighbor Algorithm: Here, maps 210 and 220 are decomposed into points with coordinates, and the best pairings are identified respectively.

[0079] • Hungarian Matching: Here, the map is broken down into points with coordinates, and pairs are identified for a “perfect match.” The total distance can be used, for example, as an error term (Fehlerterm).

[0080] • Template matching: Maps 210 and 220 can be converted into pixel-based images by means of map rendering, and map fragments of the two maps 210 and 220 can be identified in pairs.

[0081] In the subsequent steps, geometric adaptation of maps 210 and 220 is performed, for example:

[0082] • Adapt the geometry of the first map 210 to fit the geometry of the second map 220;

[0083] • To adapt the second map 220 to the geometry of the first map 210; or

[0084] • Adapts to the geometry of two maps, 210 and 220.

[0085] Here, the geometry is adapted through geometric correction. From a topological and data format perspective, the two maps, 210 and 220, generally maintain their differences. For example:

[0086] • The bifurcation in the first map 210 remains bifurcation after geometric adaptation. It is possible that the bifurcation does not exist in the second map 220, and even after geometric adaptation with the aid of spatial deformation, the bifurcation is generally not present in the second map 220.

[0087] • The two maps 210 and 220 preferably maintain their data format completely, regardless of whether roads are stored as lines or other geometric elements. The deformation, typically based on a function learned from the different maps 210 and 220, transforms the coordinates to make the maps 210 and 220 geometrically match.

[0088] This approach is also advantageous for the following reasons:

[0089] • Geometric adaptation allows the simultaneous use of two maps, 210 and 220;

[0090] • The initial maps 210 and 220 can maintain their integrity;

[0091] • The required computation time is relatively short, and the computation is performed only locally (on relatively small map fragments);

[0092] • Coordinates that do not directly have a match (i.e., associated 230) are also adapted because a soft spatial curvature is applied. Figure 2 Object 223 is also moved, for example, so that object 223 can be integrated into the merged map in a precise manner.

[0093] A preferred implementation of the deformation using the deformation employs the so-called RKHS (Regenerative Kernel Hilbert Space) as a function space for adapting the geometry of the first and / or second maps 210, 220. This space allows for seamless geometric adaptation of the geometry. The spatial curvature performed using the RKHS can be illustrated as described herein.

[0094] Each point of subgrid 310 Point 301 is pulled onto the corresponding deformed sub-mesh 310 by means of the spatial curvature 322. Therefore, the positions of all points 301 are expressed in coordinates. The change applies, but point 301 at edge 313 of submesh 310 will not be transformed. For Figure 3b The spatial curvature shown is, for example, applicable to,

[0095] • : Moving grid points ;

[0096] • An exemplary spatial curvature vector 322 describes the movement from the initial submesh 310 to the deformed submesh 310;

[0097] • or An example of a core function of RKHS is one with an influence parameter σ = 3.3, which describes the correlation of spatial curvature.

[0098] Therefore, for each point in subgrid 310 x 301, the relevant point in curved space (i.e., in the deformed submesh 310). The following limitations apply:

[0099] Therefore, the deformation function used for the deformable sub-mesh 310 (by means of spatial deformation) can be obtained through the mesh points. The set of related curvature vectors v and core functions To define the range, using multiple grid points, spatial displacement can be represented in matrix notation as follows:

[0100] have The curvature vectors are the core matrix (with core or kernel functions for each spatial curvature) and the curvature vectors for each grid point (generalized into a matrix). .

[0101] After determining the core or kernel function, suitable grid points 301 and / or spatial displacements can be found that match the displacement vector 231 determined within the associated 230 of objects 211, 212. This can be achieved by utilizing the fact that if the observed grid points... If the grid is uniformly and grid-shaped in space (e.g., at each intersection of subgrid 310), then any possible spatial curvature can be represented using this grid, so that an infinitely dense grid does not provide a significant improvement in accuracy. A suitable choice of grid distance typically depends on the constraints of the core function and can be chosen, for example, as 0.5σ.

[0102] Using limited core functions k and limited to having grid points When the submesh of 301 is 310, the spatial curvature passes only through the spatial curvature vector. Limited. Spatial curvature vector for submesh 310. It can be determined by solving a system of linear equations.

[0103] in d It is one or more displacement vectors 231 of the set of associated objects 211, 212 arranged in the core region 311 of the subgrid 310.

[0104] The advantage of the spatial curvature described in this document is that the curvature vector 322 has a very local effect on the corresponding environment. If, for a specific map segment (e.g., for a specific subgrid 310), (using the spatial curvature vector...) v If the set of deformation functions (i.e., the set of spatial curvature vectors) has been determined, then it is usually sufficient to update the deformation functions (i.e., the set of spatial curvature vectors) for that map segment when it is changed. Complete map information is not required for this. Displacement in the core region 311 of segment 310 will only affect the edge regions 312 of segment 310 (i.e., sub-grid 310).

[0105] Therefore, it is possible to iteratively determine the deformation function for each map segment (and its corresponding sub-grid 310). For this purpose, the sub-grid 310 can be moved gradually, as in... Figure 3a As illustrated by the arrows in the diagram. The computation can also be performed in parallel, which enables the determination of the deformation function with particular efficiency.

[0106] like Figure 3a As shown, subgrids 310 can move in an overlapping manner to gradually cover the total grid 300. For example, each subgrid 310 can be moved exponentially. t To identify, among which, index tThe number increases as sub-grid 310 moves across the total grid 300. In sub-grid 310... t In, especially in subgrid 310 t The core region 311 can contain elements with displacement vectors. d t The set corresponding to 231 is associated with the set of 230. Furthermore, the curvature vector from the previous determination step used for the previous subgrid t-1. For subgrid 310 t A portion of the grid points 301 can be available. Based on this, the initial curvature vector can be determined. s t The set used to determine the submesh 310 t curvature vector v t The set. Used for submesh 310 t curvature vector v t The set can be determined using an optimization problem, wherein an exemplary optimization problem is defined by the following function:

[0107] Here,

[0108] • It is a (pre-defined) prediction function that can predict based on the current initial curvature vector. and curvature vector Determine the expected displacement vector .

[0109] • It is a weighted matrix (based on covariance).

[0110] • It is a weighted matrix (based on covariance).

[0111] The second term in the optimization problem aims to minimize the deformation of the submesh. This term typically improves the stability of the optimization method.

[0112] Therefore, the curvature vector 322 for each grid point 301 of the total grid 300 can be gradually determined. Furthermore, a covariance value can be determined for each grid point 301, which indicates the quality and / or reliability of the curvature vector 322.

[0113] Each curvature vector 322 can be used to make the various parts of two different maps 210, 220 overlap with each other and thereby provide a fused digital map with information from the two digital maps 210, 220.

[0114] During vehicle 100 operation, automatic longitudinal and / or lateral guidance can be implemented, for example, based on a first digital map 210. When needed, information 223 can be incorporated from a second digital map 220 into the first digital map 210. Here, a curvature vector 322 determined for a subgrid 310, which includes the corresponding current position of vehicle 100, can be used. Therefore, information can be integrated from the second digital map 220 into the first map 210 as if the two maps 210 and 220 were a uniquely merged digital map.

[0115] Figure 4 A flowchart is shown of a method 400 (possibly computer-implemented) for fusing map information from a first digital map 210 and a second digital map 220. Method 400 can be executed by device 101. Method 400 includes: determining 401 a set of paired associations 230 between a first object 211 from the first digital map 210 and a second object 221 from the second digital map 220, respectively, for partial regions of the first and second digital maps 210 and 220; and determining 402 a set of displacements 231 (specifically displacement vectors), wherein the associated first and second objects 211 and 221 coincide with each other (in pairs, respectively) through the displacements. A partial region may correspond to a specific region of the Earth's surface. A subgrid 310 with a matrix of grid points 301 can be defined for a partial region. Each grid point 301 can be associated with a point or coordinate of the partial region. Displacement of a grid point may result in a corresponding displacement of the associated point or coordinate. Therefore, the corresponding deformation of the partial region can be achieved through deformation of the subgrid 310.

[0116] Furthermore, method 400 includes determining, based on the set of displacements 231, a deformation function 403 for deforming subgrid 310 for a portion of the first and second digital maps 210, 220. The deformation function can be defined by a spatial curvature vector 322 for each grid point 301. The spatial curvature vector 322 can be determined by optimizing an optimization function, wherein the optimization function depends on the set of displacements 231.

[0117] Method 400 further includes transferring map information 404 from a portion of the second digital map 220 to a portion of the first digital map 210 using a deformation function. Therefore, a particularly stable and accurate fusion of the digital maps 210 and 220 can be achieved.

[0118] This invention is not limited to the embodiments shown. In particular, it should be noted that the specification and drawings are intended to illustrate the principles of the proposed methods, apparatus, and systems only by way of example.

Claims

1. A method (400) for fusing map information from a first digital map (210) and a second digital map (220), wherein, The method (400) includes: For a portion of the first digital map (210) and the second digital map (220), determine (401) a set of pairwise associations (230) between a first object (211) from the first digital map (210) and a second object (221) from the second digital map (220), respectively. Determine the set of displacements (231) (402), the set of the first object (211) and the set of the second object (221) associated with each other by the displacements. Based on the set of displacements (231), for the aforementioned partial regions of the first digital map (210) and the second digital map (220), determine (403) a deformation function for deforming the sub-grid (310), and Using the deformation function, map information is transferred (404) from a portion of the second digital map (220) to a portion of the first digital map (210).

2. The method (400) according to claim 1, wherein The subgrid (310) includes a matrix of grid points (301), and The deformation function includes a matrix of spatial curvature vectors (322) for the matrix of the corresponding grid point (301).

3. The method (400) according to claim 2, wherein, for each grid point (301) of the matrix of grid points (301), the deformation function correspondingly includes a core function that describes how the movement of the corresponding grid point (301) with the spatial curvature vector (322) of the corresponding grid point (301) affects one or more grid points (301) in the surrounding environment of the corresponding grid point (301).

4. The method (400) according to any one of claims 2 to 3, wherein, The deformation function is applied to a grid point of the matrix of grid point (301). (301) Explain the moving grid points As in These are the grid points used for the sub-grid (310). (301) spatial curvature vector, and where It is the core function.

5. The method (400) according to claim 4, wherein, The core function is constructed such that the value With grid points x 1 With grid points x i It decreases as the distance between them increases.

6. The method (400) according to any one of claims 4 to 5, wherein, The core function is: or It has an influence parameter σ, which describes a specific grid point. x i (301) Towards the grid point x 1 (301) The degree of movement.

7. The method (400) according to any one of the preceding claims, wherein The deformation function is determined through an optimization problem, and The optimization problem includes at least one term that depends on the distance between the set of second objects (221) in a portion of the second digital map (220) deformed by the deformation function and the associated set of first objects (211) in a portion of the first digital map (210) that is not deformed.

8. The method (400) according to any one of the preceding claims, wherein, The method (400) includes, for each of the corresponding sub-grids (310) of a series of partial regions for the first digital map (210) and the second digital map (220): For corresponding partial areas of the first digital map (210) and the second digital map (220), a set of paired associations (230) between a first object (211) from the first digital map (210) and a second object (221) from the second digital map (220) is determined; The set of displacements (231) is determined such that the corresponding sets of the associated first objects (211) and second objects (221) of the partial regions coincide with each other through the displacements; and Based on the respective determined sets of displacements (231), for the corresponding partial regions of the first digital map (210) and the second digital map (220), a deformation function (403) is determined to deform the corresponding subgrid (310).

9. The method (400) according to claim 8, wherein The directly continuous sub-mesh (310) of the series of sub-mesh (310) partially overlap; and When determining the deformation function of the directly following submesh (310), the deformation function determined for the previous submesh (310) is taken into account.

10. The method (400) according to any one of the preceding claims, wherein, The method (400) includes: Determine a reliability index for the reliability of the determined deformation function; and Map information is transferred (404) from a portion of the second digital map (220) to a portion of the first digital map (210) according to the reliability index.

11. The method (400) according to any one of the preceding claims, wherein, The map information includes objects (223) recorded in a portion of the second digital map (220) but not in a portion of the first digital map (210).

12. An apparatus (101) for fusing map information from a first digital map (210) and a second digital map (220), wherein, The device (101) is designed for, For a portion of the first digital map (210) and the second digital map (220), a set of paired associations (230) between a first object (211) from the first digital map (210) and a second object (221) from the second digital map (220) is determined; The set of displacements (231) is determined, and the set of the first object (211) and the second object (221) associated with each other coincides through the displacements; Based on the set of displacements (231), for a portion of the first digital map (210) and the second digital map (220), a deformation function is determined to deform the sub-grid (310); and The map information is transferred from a portion of the second digital map (220) to a portion of the first digital map (210) using the deformation function.

Images