EP4004581A1 - Production de données de référence non sémantiques pour déterminer la position d'un véhicule à moteur - Google Patents

Production de données de référence non sémantiques pour déterminer la position d'un véhicule à moteur

Info

Publication number
EP4004581A1
EP4004581A1 EP20735552.0A EP20735552A EP4004581A1 EP 4004581 A1 EP4004581 A1 EP 4004581A1 EP 20735552 A EP20735552 A EP 20735552A EP 4004581 A1 EP4004581 A1 EP 4004581A1
Authority
EP
European Patent Office
Prior art keywords
raw data
point cluster
point
data points
cluster
Prior art date
Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
Pending
Application number
EP20735552.0A
Other languages
German (de)
English (en)
Inventor
Constanze HUNGAR
Daniel WILBERS
Bernd Rech
Niklas KOCH
Stefan Juergens
Christian MERFELS
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
MAN Truck and Bus SE
Volkswagen AG
Original Assignee
MAN Truck and Bus SE
Volkswagen AG
Priority date (The priority date is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the date listed.)
Filing date
Publication date
Application filed by MAN Truck and Bus SE, Volkswagen AG filed Critical MAN Truck and Bus SE
Publication of EP4004581A1 publication Critical patent/EP4004581A1/fr
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/02—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00
    • G01S7/41—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S13/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01C—MEASURING DISTANCES, LEVELS OR BEARINGS; SURVEYING; NAVIGATION; GYROSCOPIC INSTRUMENTS; PHOTOGRAMMETRY OR VIDEOGRAMMETRY
    • G01C21/00—Navigation; Navigational instruments not provided for in groups G01C1/00 - G01C19/00
    • G01C21/38—Electronic maps specially adapted for navigation; Updating thereof
    • G01C21/3804—Creation or updating of map data
    • G01C21/3833—Creation or updating of map data characterised by the source of data
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S13/00—Systems using the reflection or reradiation of radio waves, e.g. radar systems; Analogous systems using reflection or reradiation of waves whose nature or wavelength is irrelevant or unspecified
    • G01S13/88—Radar or analogous systems specially adapted for specific applications
    • G01S13/93—Radar or analogous systems specially adapted for specific applications for anti-collision purposes
    • G01S13/931—Radar or analogous systems specially adapted for specific applications for anti-collision purposes of land vehicles
    • G—PHYSICS
    • G01—MEASURING; TESTING
    • G01S—RADIO DIRECTION-FINDING; RADIO NAVIGATION; DETERMINING DISTANCE OR VELOCITY BY USE OF RADIO WAVES; LOCATING OR PRESENCE-DETECTING BY USE OF THE REFLECTION OR RERADIATION OF RADIO WAVES; ANALOGOUS ARRANGEMENTS USING OTHER WAVES
    • G01S7/00—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00
    • G01S7/48—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00
    • G01S7/4802—Details of systems according to groups G01S13/00, G01S15/00, G01S17/00 of systems according to group G01S17/00 using analysis of echo signal for target characterisation; Target signature; Target cross-section
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F16/00—Information retrieval; Database structures therefor; File system structures therefor
    • G06F16/90—Details of database functions independent of the retrieved data types
    • G06F16/906—Clustering; Classification
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/23—Clustering techniques
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06F—ELECTRIC DIGITAL DATA PROCESSING
    • G06F18/00—Pattern recognition
    • G06F18/20—Analysing
    • G06F18/28—Determining representative reference patterns, e.g. by averaging or distorting; Generating dictionaries
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T7/00—Image analysis
    • G06T7/70—Determining position or orientation of objects or cameras
    • G06T7/73—Determining position or orientation of objects or cameras using feature-based methods
    • G06T7/74—Determining position or orientation of objects or cameras using feature-based methods involving reference images or patches
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
    • G06V20/00—Scenes; Scene-specific elements
    • G06V20/50—Context or environment of the image
    • G06V20/56—Context or environment of the image exterior to a vehicle by using sensors mounted on the vehicle
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10028—Range image; Depth image; 3D point clouds
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/10—Image acquisition modality
    • G06T2207/10032—Satellite or aerial image; Remote sensing
    • G06T2207/10044—Radar image
    • G—PHYSICS
    • G06—COMPUTING; CALCULATING OR COUNTING
    • G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
    • G06T2207/00—Indexing scheme for image analysis or image enhancement
    • G06T2207/30—Subject of image; Context of image processing
    • G06T2207/30248—Vehicle exterior or interior
    • G06T2207/30252—Vehicle exterior; Vicinity of vehicle

Definitions

  • the present invention relates to a computer-implemented method for generating non-semantic reference data for determining the position of a motor vehicle, with a set of raw data points being provided which depicts a predetermined surrounding area.
  • the invention also relates to a method for determining the position of a motor vehicle, a map system for determining the position of a motor vehicle, a motor vehicle with a map system, a computer system for generating non-semantic reference data for determining the position of a motor vehicle, a computer program and a computer-readable storage medium.
  • semantic structures and patterns are recognized by the vehicle sensors in the motor vehicle environment and compared with corresponding entries in a digital map of the motor vehicle.
  • semantic structures are always assigned to a predefined class, for example provided with the information about what type of object it is, for example a traffic sign or the edge of a house.
  • Landmark-based localization however, has the disadvantage that only those environmental features are used for localization that can be assigned to a more or less generic class. This limits the number of environmental features available for localization, so that in areas in which no such landmarks are available, localization cannot be carried out or can only be carried out with low accuracy.
  • the position of a motor vehicle can also be determined on the basis of satellite signals from a global navigation satellite system.
  • the accuracy of satellite receivers usually used in motor vehicles, however, is too low to enable highly automated or autonomous driving.
  • the document EP 3 290 864 A1 describes a driver assistance system for determining a vehicle position. Approximate position data of the motor vehicle are recorded using GPS signals. In addition, an image of the surroundings of the motor vehicle is recorded and compared with stored image data. By combining the information obtained in this way, the position of the motor vehicle can be determined even if the satellite reception is poor.
  • the improved concept is based on the idea of generating non-semantic reference data by clustering the raw data points depicting the environment using predetermined descriptors to describe the properties of the environment and storing corresponding feature information depending on the information gain that the respective point clusters can contribute to position determination.
  • a computer-implemented method for generating non-semantic reference data for determining the position of a motor vehicle is specified.
  • a set of raw data points in particular generated by means of an environment sensor, is provided which depicts a predefined environment area.
  • a predetermined descriptor is determined for each of the raw data points by means of a computing unit, which characterizes a property of the surrounding area at a position of the respective raw data point or at a position which corresponds to the respective raw data point in the environment.
  • At least one point cluster is generated by means of the computing unit in that, in particular by means of the computing unit, the raw data points are grouped into the point cluster depending on their respective descriptors.
  • a first point cluster of the at least one point cluster is assigned a code number that relates to information gain for determining the position of the motor vehicle, depending on the descriptors of the raw data points of the first point cluster.
  • characteristic information of the first point cluster is stored as non-semantic reference data for position determination on a memory unit.
  • the surrounding area is in particular a surrounding area of the environment sensor or a data acquisition system or a data acquisition vehicle on which the environment sensor is mounted.
  • the raw data points are generated before the motor vehicle uses the reference data to determine its position.
  • the surrounding area is therefore an area surrounding a potential position of the motor vehicle.
  • the environment sensor can be designed, for example, as a radar sensor or lidar sensor, also referred to as a laser scanner.
  • the environment sensor system In order to map the surrounding area, the environment sensor system generates a point cloud from scanning points, which are present in particular as three-dimensional coordinate tuples.
  • the raw data points that map the surrounding area are in particular such a point cloud or a part thereof.
  • the provision of the set of raw data points takes place in particular in computer-readable form so that the raw data points can be read by means of the computing unit.
  • the generation of the raw data points is not necessarily part of the method for generating non-semantic reference data according to the improved concept.
  • the fact that the raw data points are generated in advance and then provided has the particular advantage that high-precision measuring devices can be used when generating the raw data points.
  • the raw data points of the set of raw data points can be present, for example, as corresponding coordinate tuples in a predetermined reference coordinate system, for example a global coordinate system or world coordinate system, for example a geodetic coordinate system, such as WGS84.
  • a predetermined reference coordinate system for example a global coordinate system or world coordinate system, for example a geodetic coordinate system, such as WGS84.
  • the descriptor is in particular a property of the surrounding area that can be measured using the raw data points, for example a geometric property. Contain geometric properties that can serve as descriptors in particular the curvature or mean curvature of the scanned surfaces or areas in the surrounding area. Statistical properties of the raw data points or their distribution or properties of the distribution of the raw data points can also serve as descriptors. Optical properties of the environment can also be reflected by the raw data points. In particular, the raw data points can also contain intensity information in addition to the spatial coordinates, or the intensity information can be assigned to the raw data points and made available. In the case of a lidar sensor, this is, for example, the intensity of the reflected laser beam.
  • the spectral reflectivity or color of the corresponding point in the environment that corresponds to the raw data point can be determined on the basis of the intensity. These properties can also serve as descriptors.
  • the specified descriptor can also contain several values suitable as descriptors or one or several variables derived from these.
  • each raw data point of the set of raw data points is either assigned to exactly one point cluster of the at least one point cluster or the respective raw data point is sorted out or discarded, i.e. not taken into account for the generation of the reference data .
  • the raw data points of the first point cluster are in particular those raw data points of the set of raw data points that form the first point cluster as a result of the grouping.
  • the raw data points are grouped depending on the descriptors can in particular be understood to mean that the individual descriptors of the individual raw data points are used for grouping or that the descriptors of the raw data points are statistically evaluated, for example by averaging or local averaging or other analysis of the distribution of the Descriptors and the grouping is carried out based on the result of the statistical evaluation. Both Aspects can also be combined or carried out one after the other in order to generate the at least one point cluster.
  • the raw data points can initially be grouped on the basis of their individual descriptors, for example to form descriptor clusters, and cluster descriptors can then be assigned to the descriptor clusters, for example based on the descriptors of their respective raw data points, which for example correspond to statistical characteristic values of the descriptors of the raw data points of the respective descriptor cluster, for example an average value, a median and so on.
  • the descriptor clusters can then be grouped into the point clusters on the basis of further criteria, in particular on the basis of their cluster descriptors.
  • the fact that the code of the first point cluster relates to an information gain for determining the position of the motor vehicle can be understood, for example, in such a way that the code indicates what influence the use of the first point cluster for determining the position of the motor vehicle has on the accuracy of the position determination.
  • the characteristic value can, for example, indicate how clearly the first point cluster can be recognized, how many other clusters are in the immediate vicinity of the first point cluster, how different the cluster descriptor or the descriptors of the raw data points of the first point cluster compared to other point clusters of the at least one point cluster are and the like.
  • the key figure can express how distinctive a feature of the surrounding area is that is represented by the first point cluster.
  • the feature can be viewed as more prominent, the higher the distinctiveness of the feature, the higher the uniqueness or singularity of the feature and the lower the density of further features in the vicinity of the respective feature.
  • Each of the point clusters can in particular be assigned to a corresponding feature in the surrounding area or can be understood as a corresponding feature.
  • the features that are represented by the point clusters are non-semantic in nature, so they are not necessarily assigned a meaning.
  • the feature information includes, for example, a position of the first point cluster, for example a mean or characteristic position of the raw data points of the first point cluster, a spatial extent of the raw data points of the first point cluster or other geometric properties of the first point cluster or the raw data of the first point cluster.
  • the feature information can also contain the descriptors of the raw data points and / or the cluster descriptor of the first point cluster.
  • the storage of the feature information as a function of the code number can for example take place in such a way that the feature information and the associated code number are stored together or the feature information contains the code number itself.
  • the feature information of the first point cluster can, for example, only be stored if the characteristic number is above a predetermined limit value.
  • the feature information can be stored in a marked form if the code number is smaller than the predefined limit value in order to indicate that the respective point cluster is only suitable for position determination to a limited extent.
  • Storing the feature information as a function of the key figure ensures a high quality of the reference data and, in particular, enables high accuracy of the position determination based on the reference data.
  • the method according to the improved concept makes it possible to generate reference data for determining position largely independently of the content of the surrounding area, i.e. regardless of which objects or structures are located within the surrounding area.
  • the improved concept can therefore be used universally and flexibly and, in particular, independent of the presence of semantic landmarks.
  • the features represented by the point clusters can also relate to features in the environment that are not intuitively recognizable for humans.
  • the descriptors of the raw data points are used as a means to identify features in the environment as such and particularly distinctive ones
  • features for determining position or to save them as reference data For example, use is made of the fact that the distribution, in particular spatial distribution, of the descriptors of the raw data points for distinctive features, ie in particular clearly recognizable features, differs from non-distinctive features, for example objects with very complex surfaces.
  • the features are presented and used further in the form of feature information without meaningful content.
  • the method includes the acquisition of sensor measurement data based on the environment sensor system and the generation of the raw data points based on the sensor measurement data.
  • each point cluster of the at least one point cluster is assigned a respective code number, which relates to the respective information gain for determining the position of the motor vehicle by the respective point cluster, depending on the respective descriptors of the raw data points of the respective point cluster.
  • the respective feature information of the point clusters of the at least one point cluster is stored on the storage unit as a function of the respective characteristic numbers of the point clusters.
  • a spatial distribution of all point clusters of the at least one point cluster is analyzed by means of the computing unit.
  • a localization parameter is determined for the first point cluster and possibly for all further point clusters of the at least one point cluster.
  • the characteristic number of the first point cluster is determined by means of the computing unit as a function of the localization characteristic value of the first point cluster. The same applies, if necessary, to the key figures of the other point clusters.
  • the localization characteristic quantifies in particular how high the density of point clusters of the at least one point cluster is at the position of the first point cluster, that is to say in particular in the immediate vicinity of the first point cluster. The more point clusters there are in the immediate vicinity of the first point cluster, the less suitable the first point cluster is for determining the position, or the lower the gain in information that the first point cluster can contribute to determining the position.
  • the localization parameter is greater, for example, the smaller the density of point clusters at the position of the first point cluster.
  • the key figure of the first point cluster is greater, the greater the localization characteristic value for the first point cluster.
  • the preference for more strongly localized features in the surrounding area increases the reliability or accuracy of the possible position determination based on the reference data.
  • the computing unit determines a number of point clusters of the at least one point cluster that are located in a predetermined sub-area of the surrounding area in which the first point cluster is located.
  • the localization parameter is determined depending on the number of point clusters in the specified sub-area.
  • the localization parameter can additionally be determined as a function of the total number of point clusters of the at least one point cluster, for example as the ratio of the number of point clusters in the specified sub-area to the total number.
  • the surrounding area can be completely divided into a plurality of predefined partial areas, including the partial area in which the first point cluster is located.
  • the localization parameter can then be determined, for example, as a function of an average number of point clusters in the various partial areas.
  • the localization parameter can be determined as the ratio of the number of point clusters in the sub-area in which the first point cluster is located to the average number of point clusters in all sub-areas.
  • an average density of point clusters is used to determine the localization parameter.
  • a singularity characteristic value for the first point cluster is determined by means of the arithmetic unit as a function of the descriptors of the raw data points of the first point cluster and as a function of the descriptors of the raw data points of a second point cluster of the at least one point cluster.
  • the key figure is determined as a function of the singularity characteristic value, and in particular as a function of the localization characteristic value, of the first point cluster.
  • the singularity or uniqueness of a point cluster or a feature to which the point cluster corresponds can be understood as a characteristic value for a deviation of the descriptors of the raw data points of the respective point cluster or of the cluster descriptor of the respective point cluster compared to other point clusters of the at least one point cluster.
  • the descriptors or cluster descriptors of all point clusters of the at least one point cluster can be used to determine the singularity index of the first point cluster.
  • the quality of the reference data that is to say in particular the accuracy of the position determination that can be achieved by means of the reference data, can be increased further.
  • the singularity characteristic of the first point cluster is greater, the more the descriptors of the raw data points of the first point cluster differ from the descriptors of the second point cluster or the more the cluster descriptor of the first point cluster differs from the cluster descriptor of the second point cluster.
  • the singularity characteristic value and the localization characteristic value are weighted by means of the computing unit and the characteristic number becomes dependent determined by the weighted singularity value and the weighted localization value.
  • At least one descriptor cluster is generated by means of the computing unit by grouping the raw data points depending on their respective descriptors and independently of their respective spatial positions.
  • the at least one point cluster is generated by spatially grouping the raw data points, with each descriptor cluster of the at least one descriptor cluster being identical to one of the point clusters of the at least one point cluster or being separated in order to form at least two of the point clusters of the at least one point cluster.
  • a descriptor cluster In order to generate the at least one descriptor cluster, those raw data points are grouped whose descriptors have similar values. In that the spatial positions of the raw data points are not taken into account for this purpose, a descriptor cluster can also be spatially disconnected according to a predetermined criterion.
  • a descriptor cluster can therefore in particular be a subset of
  • Raw data points of the set of raw data points are understood which, according to a predetermined definition, have similar descriptors and are in particular formed using a known method for cluster analysis.
  • a descriptor cluster can accordingly represent one or more features in the surrounding area.
  • Neglecting the spatial position can, for example, lead to two spatially separated, but otherwise identical or similar objects, to raw data points of the same descriptor cluster.
  • two walls that are spatially separated from one another can be assigned to the same descriptor cluster in the surrounding area.
  • a cluster descriptor is determined by means of the computing unit for each descriptor cluster depending on the descriptors of the raw data points of the descriptor cluster, for example by statistical evaluation of the Descriptors of the raw data points.
  • the cluster descriptor can, for example, correspond to an average value of the descriptors of the raw data points of the descriptor cluster or to another statistical parameter.
  • the raw data points of a descriptor cluster can be further processed jointly, which results in a lower memory and / or computing requirement.
  • all raw data points of the set of raw data points that cannot be assigned to any of the descriptor clusters according to a predetermined rule are discarded and no longer used to generate the reference data.
  • a distinctive characteristic value for the first point cluster is determined by means of the computing unit as a function of the descriptors of the raw data points of the first point cluster.
  • the key figure is determined as a function of the distinctive characteristic value and in particular as a function of the localization characteristic value and, for example, as a function of the singularity characteristic value of the first point cluster.
  • the distinctive characteristic value can be determined differently depending on the type of descriptors or raw data points used.
  • a distribution of the values of the descriptors of the raw data points of the first point cluster can be determined and compared with predetermined criteria.
  • the distribution can be used to determine whether the descriptors of the first point cluster are distributed unimodally or multimodally, how many local maxima the distribution has, how large the maximum values of the local maxima are, how broad the distribution or individual sub-distributions of the distribution are, and so on .
  • jumps in the distribution can also be used to determine the distinctive characteristic value.
  • a histogram of the descriptors of the raw data points of the first point cluster is generated by means of the computing unit and the histogram is analyzed in order to determine the distinctive characteristic value.
  • the distinctive characteristic value describes in particular how well the first point cluster can be described, for example how clearly the feature that the first point cluster describes can be recognized as such in the environment.
  • a corresponding distinctive characteristic value is determined for each descriptor cluster of the at least one descriptor cluster.
  • the distinctive characteristic value of the first point cluster then corresponds to the distinctive characteristic value of that descriptor cluster from which the first point cluster emerges.
  • the first point cluster thus inherits the distinctive characteristic value, as it were, from the associated descriptor cluster.
  • the distribution of the descriptors of the raw data points of the first point cluster is determined by means of the arithmetic unit and the distinctive characteristic value is determined as a function of the distribution.
  • the feature information of the first point cluster is stored on the storage unit by means of the computing unit as a function of the distinguishing characteristic value.
  • the feature information can only be stored on the storage unit, for example, if the distinctive characteristic value is greater than a predetermined further limit value.
  • the distinctive characteristic value, the singularity characteristic value and the localization characteristic value are weighted by means of the arithmetic unit and the characteristic number is determined as a function of the weighted singularity characteristic value, the weighted singularity characteristic value and the weighted localization characteristic value.
  • a method for determining the position of a motor vehicle is specified.
  • Image data of the surroundings of the motor vehicle are generated by means of an environment sensor of the motor vehicle.
  • the image data with predetermined reference data for determining the position are stored, matched.
  • a position of the motor vehicle is determined as a function of a result of the comparison by means of the further computing unit.
  • the reference data for position determination were generated using a method for determining reference data for position determination according to an improved concept.
  • a map system in particular a digital map system, in particular a digital map or HD map, for determining the position of a motor vehicle is specified.
  • the map system has a further memory unit, reference data for determining the position being stored on the further memory unit, which were generated by means of a method for determining reference data for determining the position of a motor vehicle according to the improved concept.
  • the memory unit on which the feature information is stored according to the method according to the improved concept is, in particular, the further memory unit of the card system.
  • Motor vehicle specified which has a map system for determining position according to the improved concept.
  • the computer system for generating reference data for determining the position of a motor vehicle specified.
  • the computer system has a computing unit and a memory unit.
  • the arithmetic unit is set up to receive a set of raw data points which represent a predefined surrounding area.
  • the arithmetic unit is set up to determine a descriptor for each of the raw data points which characterizes a property of the surrounding area at a position of the respective raw data point.
  • the computing unit is set up to generate at least one point cluster by grouping the raw data points depending on their descriptors.
  • the arithmetic unit is set up to assign a code number that relates to information gain for determining the position of the motor vehicle to a first point cluster of the at least one point cluster, depending on the descriptors of the raw data points of the first point cluster.
  • the arithmetic unit is set up to store feature information of the first point cluster as reference data for determining the position on the memory unit as a function of the code number.
  • the fact that the processing unit is set up to receive the set of raw data points can in particular be understood in such a way that the raw data points can be read by the processing unit.
  • FIG. 1 Further embodiments of the computer system according to the improved concept result directly from the various embodiments of the method for generating non-semantic reference data for determining position according to the improved concept and vice versa.
  • a computer system according to the improved concept is set up or programmed to carry out a method for generating non-semantic reference data according to the improved concept, or the computer system carries out such a method.
  • a computer program with instructions is specified.
  • the commands cause the computer system to carry out a method for determining non-semantic reference data according to the improved concept.
  • a computer-readable storage medium is specified on which a computer program according to the improved concept is stored.
  • the invention also includes the combinations of the features of the described embodiments.
  • FIG. 1 shows an exemplary embodiment of a computer system according to the improved concept
  • FIG. 2 shows a flow diagram of an exemplary embodiment of a method for determining non-semantic reference data according to the improved concept
  • 3 shows a motor vehicle with an exemplary embodiment of a card system according to the improved concept.
  • the described components of the embodiments each represent individual features of the invention that are to be considered independently of one another, which also develop the invention independently of one another and are therefore also to be regarded as part of the invention individually or in a combination other than the one shown. Furthermore, the described embodiments can also be supplemented by further features of the invention already described.
  • a computer system 16 is shown schematically according to the improved concept.
  • the computer system 16 has a computing unit 8 and a memory unit 10.
  • a set of raw data points is stored on the storage unit 10, for example, which maps a predefined surrounding area, that is to say a potential surrounding area of a motor vehicle 6.
  • FIG. 2 shows a flow chart of an exemplary embodiment of a method for generating non-semantic reference data for determining the position of a motor vehicle 6.
  • the set of raw data points 7 is provided in that it is stored on the storage unit 10 in computer-readable form, that is to say in particular readable by the computing unit 8.
  • the set of raw data points 7 corresponds, for example, to a point cloud that was generated in advance by means of a lidar system.
  • a predefined descriptor is assigned to each of the raw data points 7 by means of the arithmetic unit 8 or the corresponding descriptor is calculated for each raw data point 7 by means of the arithmetic unit 8, the descriptor being in particular a geometric one Characterized property of the surrounding area at the position which is represented by the respective raw data point, for example a curvature or mean curvature of the surroundings or an object in the surrounding area at the position of the respective raw data point.
  • depth information can also be used as a descriptor.
  • step 2 of the method the raw data points are grouped into descriptor clusters 12a, 12b, 12c by combining raw data points 7 with similar descriptors. That is to say, descriptors of the same type are grouped into descriptor clusters 12a, 12b, 12c by a clustering method.
  • those features in the surrounding area 11 which relate to similar objects can be combined in a common descriptor cluster 12a, 12b, 12c, for example all wall-like descriptions can be combined in a descriptor cluster 12a, 12b, 12c.
  • step 2 those of the raw data 7 which cannot be assigned to any of the descriptor clusters 12a, 12b, 12d according to predetermined criteria, can also be sorted out and are then no longer used.
  • a distinctive characteristic value can be determined in particular for each of the descriptor clusters 12a, 12b, 12c, in particular based on the descriptors of the corresponding raw data points 7 of the respective descriptor cluster 12a, 12b, 12c.
  • the distinctive characteristic value quantifies how writable and recognizable the descriptors of the respective descriptor cluster 12a, 12b, 12c are.
  • step 2 for example, all descriptor clusters 12a, 12b, 12c whose distinctive characteristic values are below a predetermined limit value can be sorted out and are not considered further.
  • distributions 17a, 17b of two descriptor clusters 12a, 12b, 12c are shown schematically in FIG.
  • the distribution 17a can represent the distribution of curvature values as descriptors for an advertising column and the distribution 17b a corresponding distribution for a tree. While the distribution 17a indicates a high distinctiveness, since it has three distinct maxima, the distribution 17b may not be suitable to be used for position determination since it indicates a relatively homogeneous distribution of the descriptors and thus a low distinctive value.
  • the descriptor clusters 12a, 12b, 12c are spatially separated from one another so that corresponding point clusters 9a, 9b, 9c, 9d, 9e, 9f are generated by the computing unit 8.
  • two house walls positioned at different points in the environment can fall into the same descriptor cluster 12a, 12b, 12c, but into different point clusters 9a, 9b, 9c, 9d, 9e, 9f.
  • the point clusters 9a, 9b, 9c, 9d, 9e, 9f in particular take over the distinguishing characteristic values of the respective descriptor clusters 12a, 12b, 12c from which they each arise.
  • a uniqueness characteristic value or singularity characteristic value can also be determined for each of the point clusters 9a, 9b, 9c, 9d, 9e, 9f by means of the computing unit 8.
  • the uniqueness or singularity of a point cluster 9a, 9b, 9c, 9d, 9e, 9f in particular quantifies the difference between the respective point cluster 9a, 9b, 9c, 9d, 9e, 9f from other point clusters 9a, 9b, 9c, 9d, 9e, 9f.
  • the singularity parameter can in particular be determined by describing the difference between descriptors of the individual point clusters 9a, 9b, 9c, 9d, 9e, 9f by means of the arithmetic unit 8 according to a predetermined mathematical rule.
  • the singularity value of the individual associated point clusters 9a, 9b, 9c, 9d, 9e, 9f is lower than if there is only one house wall.
  • a localization parameter can also be assigned.
  • a spatial distribution 18 of the point clusters 9a, 9b in particular can be analyzed.
  • the surrounding area 11 can be divided into a predetermined number of sub-areas by means of the computing unit 8, and the number of point clusters that lie within the sub-areas can be determined. The more point clusters there are in a sub-area, the lower the localization parameter for the point clusters 9a, 9b, 9c, 9d, 9e, 9f in this sub-area.
  • the arithmetic unit 8 is used to determine an associated code number for each of the point clusters 9a, 9b, 9c, 9d, 9e, 9f depending on the localization characteristic value, the singularity characteristic value and the distinctiveness characteristic value, which relates to information gain for determining the position of the motor vehicle.
  • Localization characteristic value, singularity characteristic value and distinctiveness characteristic value together describe how distinctive a feature is in the surrounding area 11, which is described by the respective point cluster 9a, 9b, 9c, 9d, 9e, 9f.
  • step 5 of the method feature information for each of the point clusters 9a, 9b, 9c, 9d, 9e, 9f, in particular corresponding spatial positions, dimensions or other geometric information of the point clusters 9a, 9b, 9c, 9d, 9e, 9f and, for example, the respective The code number or the respective characteristic values of the point clusters 9a, 9b, 9c, 9d, 9e, 9f are stored on the memory unit 10 if the code number relating to the information gain of the respective point cluster 9a, 9b, 9c, 9d, 9e, 9f is greater than an associated predetermined one Limit.
  • FIG. 3 shows a motor vehicle 6 which has a further processing unit 14 and an environment sensor system 13, for example a camera, a lidar system or a radar system.
  • an environment sensor system 13 for example a camera, a lidar system or a radar system.
  • the motor vehicle 6 also has a digital map 15, in particular a further memory unit on which a digital map 15 is stored.
  • the digital map 15 contains reference data for determining position, which were generated according to a method according to the improved concept.
  • Image data of the surroundings of the motor vehicle 6 can be generated by means of the environment sensor 13 of the motor vehicle 6.
  • the computer unit 14 can compare the image data with the reference data on the digital map 15.
  • the improved concept thus provides a possibility of making available and using non-semantic reference data for determining the position of a motor vehicle.
  • high-precision raw data points can be used to generate non-semantic, distinctive reference data for digital maps in order to enable the precise localization of a motor vehicle.
  • the storage space of the digital map is significantly reduced compared to the direct storage of the raw data points.
  • the improved concept therefore makes it possible to create highly accurate, globally referenced maps in which non-semantic features are stored together with their description.

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Radar, Positioning & Navigation (AREA)
  • Remote Sensing (AREA)
  • Data Mining & Analysis (AREA)
  • Computer Networks & Wireless Communication (AREA)
  • General Engineering & Computer Science (AREA)
  • Computer Vision & Pattern Recognition (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • Artificial Intelligence (AREA)
  • Bioinformatics & Cheminformatics (AREA)
  • Bioinformatics & Computational Biology (AREA)
  • Evolutionary Biology (AREA)
  • Evolutionary Computation (AREA)
  • Databases & Information Systems (AREA)
  • Multimedia (AREA)
  • Electromagnetism (AREA)
  • Automation & Control Theory (AREA)
  • Navigation (AREA)

Abstract

Selon un procédé mis en œuvre par ordinateur et destiné à produire des données de référence sémantiques permettant de déterminer la position d'un véhicule à moteur (6), un jeu de points de données brutes (7), qui reproduit une zone environnante (11) prédéfinie, est préparé. Un descripteur prédéfini qui caractérise une propriété de la zone environnante (11) est déterminé pour chacun des points de données brutes. Au moins une grappe de points (9a, 9b, 9c, 9d, 9e, 9f) est produite, du fait que les points de données brutes (7) sont groupés en fonction de leurs descripteurs. Un indice caractéristique qui concerne un apport d'information pour effectuer la détermination de la position d'un véhicule à moteur (6) est associé à une première grappe de points en fonction des descripteurs des points de données brutes (7). En fonction de l'indice caractéristique, des informations caractéristiques de la première grappe de points sont mémorisées en tant que données de référence non sémantiques pour déterminer la position sur une unité de stockage (10).
EP20735552.0A 2019-07-23 2020-06-29 Production de données de référence non sémantiques pour déterminer la position d'un véhicule à moteur Pending EP4004581A1 (fr)

Applications Claiming Priority (2)

Application Number Priority Date Filing Date Title
DE102019119852.6A DE102019119852A1 (de) 2019-07-23 2019-07-23 Erzeugen nicht-semantischer Referenzdaten zur Positionsbestimmung eines Kraftfahrzeugs
PCT/EP2020/068242 WO2021013475A1 (fr) 2019-07-23 2020-06-29 Production de données de référence non sémantiques pour déterminer la position d'un véhicule à moteur

Publications (1)

Publication Number Publication Date
EP4004581A1 true EP4004581A1 (fr) 2022-06-01

Family

ID=71409410

Family Applications (1)

Application Number Title Priority Date Filing Date
EP20735552.0A Pending EP4004581A1 (fr) 2019-07-23 2020-06-29 Production de données de référence non sémantiques pour déterminer la position d'un véhicule à moteur

Country Status (5)

Country Link
US (1) US20220262034A1 (fr)
EP (1) EP4004581A1 (fr)
CN (1) CN114144816A (fr)
DE (1) DE102019119852A1 (fr)
WO (1) WO2021013475A1 (fr)

Family Cites Families (20)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
US7433509B1 (en) * 2001-10-09 2008-10-07 Nanometrics Incorporated Method for automatic de-skewing of multiple layer wafer for improved pattern recognition
JP4685501B2 (ja) * 2004-07-07 2011-05-18 株式会社エヌ・ティ・ティ・ドコモ チャネル割り当て方法
US8791996B2 (en) * 2010-03-31 2014-07-29 Aisin Aw Co., Ltd. Image processing system and position measurement system
US20130173225A1 (en) * 2011-04-29 2013-07-04 Thomson Licensing Method for sampling mesh models and apparatus for sampling mesh models
US20170242443A1 (en) * 2015-11-02 2017-08-24 Peloton Technology, Inc. Gap measurement for vehicle convoying
US9311899B2 (en) * 2012-10-12 2016-04-12 International Business Machines Corporation Detecting and describing visible features on a visualization
US10706384B1 (en) * 2013-02-07 2020-07-07 United Parcel Service Of America, Inc. Systems and methods for synchronized delivery
US9129189B2 (en) * 2013-08-14 2015-09-08 Qualcomm Incorporated Performing vocabulary-based visual search using multi-resolution feature descriptors
US9476730B2 (en) * 2014-03-18 2016-10-25 Sri International Real-time system for multi-modal 3D geospatial mapping, object recognition, scene annotation and analytics
US10586163B1 (en) * 2014-06-06 2020-03-10 Mmsr, Llc Geographic locale mapping system for outcome prediction
US9664779B2 (en) * 2014-07-03 2017-05-30 GM Global Technology Operations LLC Object classification for vehicle radar systems
EP3290864A1 (fr) * 2016-08-30 2018-03-07 Continental Automotive GmbH Systeme d'assistance au conducteur destine a determiner une position d'un vehicule
EP3600962A4 (fr) * 2017-03-31 2020-12-16 A^3 By Airbus, LLC Systèmes de surveillance de véhicule et procédés de détection d'objets externes
US10935978B2 (en) * 2017-10-30 2021-03-02 Nio Usa, Inc. Vehicle self-localization using particle filters and visual odometry
US11138037B2 (en) * 2017-11-02 2021-10-05 Mediatek Inc. Switch policy for hybrid scheduling in multi-processor systems
US11093759B2 (en) * 2018-03-06 2021-08-17 Here Global B.V. Automatic identification of roadside objects for localization
US10846511B2 (en) * 2018-12-20 2020-11-24 Here Global B.V. Automatic detection and positioning of pole-like objects in 3D
EP3710984A4 (fr) * 2019-01-30 2020-09-23 Baidu.com Times Technology (Beijing) Co., Ltd. Système de partition de carte pour véhicules autonomes
EP3707469B1 (fr) * 2019-01-30 2023-10-11 Baidu.com Times Technology (Beijing) Co., Ltd. Système d'enregistrement de nuages de points pour des véhicules autonomes
US11816997B2 (en) * 2021-04-29 2023-11-14 Ge Aviation Systems Llc Demand driven crowdsourcing for UAV sensor

Also Published As

Publication number Publication date
US20220262034A1 (en) 2022-08-18
DE102019119852A1 (de) 2021-01-28
WO2021013475A1 (fr) 2021-01-28
CN114144816A (zh) 2022-03-04

Similar Documents

Publication Publication Date Title
DE102018117660A1 (de) Verfahren und system zum bestimmen einer position eines fahrzeugs
DE102011111440A1 (de) Verfahren zur Umgebungsrepräsentation
EP2005361A1 (fr) Detecteur d'objets multi-sensoriel reposant sur des hypotheses et dispositif de suivi d'objets
WO2020061603A1 (fr) Procédé et dispositif d'analyse d'un flux de données de capteur et procédé de guidage d'un véhicule
DE102019216206A1 (de) Vorrichtung und Verfahren zum Bestimmen einer Kehrtwendestrategie eines autonomen Fahrzeugs
DE102019101405A1 (de) Verfahren zum Bewerten einer Positionsinformation einer Landmarke in einer Umgebung eines Kraftfahrzeugs, Bewertungssystem, Fahrerassistenzsystem und Kraftfahrzeug
DE102020214596A1 (de) Verfahren zum Erzeugen von Trainingsdaten für ein Erkennungsmodell zum Erkennen von Objekten in Sensordaten einer Umfeldsensorik eines Fahrzeugs, Verfahren zum Erzeugen eines solchen Erkennungsmodells und Verfahren zum Ansteuern einer Aktorik eines Fahrzeugs
DE102019220069A1 (de) Verlässlichere Klassifikation von Radardaten aus dynamischen Szenerien
EP4004581A1 (fr) Production de données de référence non sémantiques pour déterminer la position d'un véhicule à moteur
DE102022000849A1 (de) Verfahren zur Erzeugung einer Umgebungsrepräsentation für ein Fahrzeug
DE102017105721A1 (de) Verfahren zur Identifikation wenigstens eines Parkplatzes für ein Fahrzeug und Parkplatzassistenzsystem eines Fahrzeugs
WO2021122339A1 (fr) Procédé et dispositif permettant de fabriquer un réseau neuronal plus robuste par rapport à des perturbations antagonistes
DE102006051091B4 (de) Verfahren zur Objekterkennung von Fahrzeugen mittels Nahbereichsdetektion
DE102020200876A1 (de) Verfahren zum Verarbeiten von Sensordaten einer Sensorik eines Fahrzeugs
DE102019124498A1 (de) Verfahren zum Komprimieren von 3D-Punktdaten in einer mittels eines Detektionssystems eines Fahrzeugs erzeugten 3D-Punktwolke
DE102020213947B3 (de) Verfahren zum Auswerten einer Umgebung eines Kraftfahrzeugs mittels eines Assistenzsystems, sowie Assistenzsystem
EP3755972B1 (fr) Construire une carte geographique
EP3671125B1 (fr) Système de détection de la position et procédé de fonctionnement d'un système de détection de la position pour une unité mobile
DE102019219924B4 (de) Verfahren und Vorrichtung zum Erzeugen und Bereitstellen einer Datenbank mit darin hinterlegten Sensordatenpatches zur Verwendung beim Quilting
DE102021201557A1 (de) Verfahren zur Erfassung eines Objektes
DE102022110233A1 (de) Computer-implementiertes verfahren zur auswahl von trainingsdatensätzen aus einer vielzahl von datensätzen
DE102020204592A1 (de) Vorrichtung und Verfahren zur Verarbeitung von Radardaten und Radarsystem
DE102020208981A1 (de) Verfahren zur Schätzung der Geometrie eines Bewegungsweges
DE102022119715A1 (de) Verfahren, System und Computerprogrammprodukt zur objektiven Bewertung der Leistungsfähigkeit eines ADAS/ADS-Systems
DE102022205910A1 (de) Verfahren und Vorrichtung zum Erkennen eines Parkplatzes für ein Fahrzeug mithilfe maschineller Lernverfahren

Legal Events

Date Code Title Description
STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: UNKNOWN

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: THE INTERNATIONAL PUBLICATION HAS BEEN MADE

PUAI Public reference made under article 153(3) epc to a published international application that has entered the european phase

Free format text: ORIGINAL CODE: 0009012

STAA Information on the status of an ep patent application or granted ep patent

Free format text: STATUS: REQUEST FOR EXAMINATION WAS MADE

17P Request for examination filed

Effective date: 20220209

AK Designated contracting states

Kind code of ref document: A1

Designated state(s): AL AT BE BG CH CY CZ DE DK EE ES FI FR GB GR HR HU IE IS IT LI LT LU LV MC MK MT NL NO PL PT RO RS SE SI SK SM TR

DAV Request for validation of the european patent (deleted)
DAX Request for extension of the european patent (deleted)