Method, apparatus, computer program and storage medium for detecting road nodes

Abstract

The invention relates to a method for detecting a road node, comprising the following steps: receiving position data describing a corresponding position change process of a motor vehicle driving on a road comprising the road node and driving through the road node; solving a bending change process corresponding to the position change process; ascertaining a derivative, in particular a numerical derivative, corresponding to the bending change process in order to obtain a corresponding change of the bending change process with respect to the corresponding position change process; the road nodes are detected on the basis of the respective derived derivatives of the curve change process. The invention also relates to an apparatus, a computer program and a machine-readable storage medium.

Description

Technical Field

[0001] The invention relates to a method, a device, a computer program and a machine-readable storage medium for detecting road nodes. Background Art

[0002] Publication US 2021 / 0001877 discloses a method for determining lane connectivity at traffic intersections for high-resolution maps.

[0003] Publication US 2022 / 0057231 363 A1 discloses a method for identifying map calibration errors.

[0004] Publication US2022 / 0383024 discloses a method for identifying and encoding road passages based on image data. Summary of the invention

[0005] The object of the present invention is to provide a concept for efficiently detecting road nodes.

[0006] This object is achieved by the subject matter according to the invention. Advantageous embodiments of the invention are the subject matter of further developments.

[0007] According to a first aspect, a method for detecting road nodes is provided, comprising the following steps: receiving position data, which describes a corresponding position change process (Positionsverlauf) of a motor vehicle traveling on a road including the road node and passing through (durchfahren) the road node; obtaining a corresponding bending change process (Krümmungsverlauf) of the position change process; obtaining a corresponding derivative of the bending change process, in particular a numerical derivative, for obtaining a corresponding change of the bending change process with respect to the corresponding position change process; and detecting the road node based on the respectively obtained derivatives of the bending change process.

[0008] According to a second aspect, an apparatus is provided, which is configured to perform all the steps of the method according to the first aspect.

[0009] According to a third aspect, there is provided a computer program comprising instructions which, when executed by a computer, for example by an apparatus according to the second aspect, cause the computer to perform the method according to the first aspect.

[0010] According to a fourth aspect, there is provided a machine-readable storage medium having stored thereon a computer program according to the third aspect.

[0011] The invention is based on and includes the recognition that the above-mentioned object is achieved in the following way: from a large number of motor vehicles traveling on a road, corresponding motor vehicle position change processes are used to detect road nodes, the road comprising road nodes and in which the motor vehicles travel through these road nodes. In detail, it is provided that the corresponding bending change process of the position change process is determined. This means that the first-order derivative of the position change process is determined. Subsequently, the derivative of the bending change process is determined to obtain the corresponding change of the bending change process with respect to the corresponding position change process. In other words, the second-order derivatives of the position change process are respectively calculated. The road nodes are detected based on the respectively calculated derivatives of the bending change process.

[0012] The background is that the change in the curvature course tells whether the road turns or straightens again. In other words, the starting point of the road curve and the end point of the road curve can be effectively identified. When driving through the road node from different directions, each motor vehicle thus drives a curve respectively. This means that "multiple curves are within a narrowly defined range or area" strongly indicates that there is a road node at this location or in this area. Therefore, the road node can be effectively detected based on the derivative obtained of the curvature course.

[0013] Thus, unlike the prior art described above, the method does not require image data to identify road nodes. It is sufficient to use the corresponding position change process. That is, the position change process describes or indicates the position of the motor vehicle on the road during its corresponding driving and when passing through the road node. The position data can be called lane data, which can be translated into English as "trace data". It is lane data because the lane through which the motor vehicle drives (abfahren) is described by these data.

[0014] A road node in the sense of this specification is a node of road traffic. A node of road traffic is a construction facility used as a connection of roads. This means that at a road node, multiple roads intersect and / or one road merges into another road.

[0015] Road nodes are, for example, road junctions, roundabouts, merges, entrances to country roads / national highways / motorways, exits to country roads / national highways / motorways or motorway intersections.

[0016] In one specific embodiment of the method, it is provided that a road curve is detected based on the respectively ascertained derivative of the curvature course, wherein road nodes are detected based on the detected road curve.

[0017] This achieves the technical advantage that, for example, road curves can be efficiently detected, based on which road nodes, which have already been explained above, can be efficiently detected.

[0018] When the term “curve” is used in this description, this is always also to be understood as a road curve, even if this is not explicitly stated.

[0019] In one specific embodiment of the method, it is provided that for each road curve the number of entrances and exits into or out of the road curve is ascertained, wherein the road nodes are detected based on the ascertained number of entrances and exits.

[0020] This achieves the following technical advantage, for example, that road nodes can be detected efficiently. This embodiment is based on the recognition that, compared to simple road curves, the number of entrances and exits of road nodes is usually different.

[0021] In one embodiment of the method, it is provided that when the number of entrances and exits is greater than or equal to 3, it is determined that the road curve belongs to a road node.

[0022] This achieves the following technical advantage, for example, that road nodes can be detected efficiently.

[0023] That is, this means that when the number of the detected entrances and exits of the road curve is greater than or equal to 3, the road curve is determined to be a road node, so that the road node is effectively detected.

[0024] A road curve usually has only one exit and one entrance. This means that it is possible to enter the road curve and exit again. If, on the other hand, there is still a third possibility to enter the detected road curve, this must therefore be a node, because here another road merges into the road curve or intersects with the road curve.

[0025] Therefore, whether a detected road curve is a road node can be effectively determined by the number of entrances and exits.

[0026] In one specific embodiment of the method, it is provided that detected road curves having an abnormal course are filtered out, wherein road nodes are detected based on the road curves that have not been filtered out.

[0027] This achieves the technical advantage that, for example, implausible results are filtered out and are therefore no longer used for the detection of road nodes, so that road nodes can be detected reliably.

[0028] “Untrustworthy” means, for example, that the distance between the entrance and exit of the road curve is greater than or equal to a predetermined distance threshold.

[0029] In one specific embodiment of the method, it is provided that a polygon is determined for each (non-filtered) road curve, which represents the corresponding road curve, wherein road nodes are detected based on the determined polygon.

[0030] This achieves the technical advantage that, for example, road curves can be represented efficiently, so that road nodes can be efficiently detected based thereon. By representing the road curves as polygons, the corresponding calculations can be performed particularly efficiently.

[0031] The advantage of describing road curves by polygons is in particular that the polygons describe only the spatial extent of the road nodes and therefore make this spatial extent independent of the exact connection used for the detection of the drive-through. In particular, this also allows other spatially defined structures to be assigned to the detected road nodes, such as roads that pass through such polygons (e.g. defined by lines).

[0032] In one embodiment of the method, it is provided that the number of entrances and exits entering or leaving the corresponding polygon is respectively determined for correspondingly determining the number of entrances and exits entering or leaving the road curve, wherein the road node is detected based on the determined number of entrances and exits entering or leaving the corresponding polygon.

[0033] This achieves the technical advantage that, for example, road nodes can be detected efficiently. The general remarks made above regarding the number of entrances and exits to a road curve apply analogously to the number of entrances and exits into or out of a polygon describing a road curve.

[0034] For example, it is set accordingly that when the number of entrances and exits entering or leaving a polygon is greater than or equal to 3, the polygon belongs to a road node.

[0035] In one embodiment of the method according to the first aspect, it is provided that a corresponding degree of discreteness (Streuungsmaβ) of the values ​​of the respectively determined derivatives of the bending change process is determined, wherein values ​​of the respectively determined derivatives of the bending change process that exceed the corresponding degree of discreteness are determined, wherein the values ​​that exceed the corresponding degree of discreteness are respectively grouped into intervals with positive bending changes and with negative bending changes, wherein the road curve is determined as a pair of directly consecutive intervals each having a positive bending change and a negative bending change.

[0036] This achieves the technical advantage that, for example, curves in road course changes can be effectively identified. The background is that in the case of a positive curve change, it can be assumed that the road turns, ie appears as a curve. In the case of a negative curve change, it can be assumed that the curved road is pointing straight again in terms of its course of change.

[0037] "Road bend" in the sense of this specification means that the road is curved in terms of its course, ie the road course is curved. "Directing straight" or "straightening" means that the curved road course becomes straight again.

[0038] In one specific embodiment of the method according to the first aspect, it is provided that a corresponding standard deviation of the values ​​of the respectively ascertained derivative of the bending profile is ascertained, wherein a corresponding degree of dispersion is ascertained based on the corresponding standard deviation.

[0039] This achieves the following technical advantage, for example, that the corresponding degree of discretization can be efficiently determined.

[0040] For example, the corresponding degree of dispersion is the standard deviation of the values ​​of the respectively determined derivatives of the bending profile.

[0041] In one implementation of the method according to the first aspect, it is provided that the corresponding standard deviation is scaled using a predetermined scaling factor not equal to 1 to obtain the scaled standard deviations respectively, wherein the corresponding discrete degrees are obtained based on the respectively scaled standard deviations.

[0042] This achieves the technical advantage that, for example, an adjustment of the determined standard deviation is possible in order to enable a finer adjustment of the detection. In an embodiment used in production, this process can also be inactive and, for example, a fixed, predefined threshold value is used.

[0043] For example, the dispersion is equal to the scaled standard deviation.

[0044] In one specific embodiment of the method according to the first aspect, it is provided that the points within the road curve are clustered, wherein a corresponding polygon is created in such a way that it includes all the points of each cluster (Cluster n).

[0045] This achieves the technical advantage that, for example, corresponding polygons can be efficiently created.

[0046] The clustering advantageously makes it possible firstly to create defined intersections (general road nodes), since a plurality of individual detections of possible intersections (general road nodes) can be combined on the basis of the individual driving throughs.

[0047] In one embodiment of the device according to the second aspect, it is provided that the device is programmed to execute the computer program according to the third aspect.

[0048] The technical functions of the device are derived analogously from the corresponding technical functions of the method, and vice versa. Method features are thus derived analogously from corresponding device features, and vice versa.

[0049] The method is carried out, for example, by means of the device.

[0050] The method is, for example, a computer-implemented method.

[0051] A position in the sense of this specification tells, for example, the latitude and longitude. The position also tells, for example, the altitude, in particular the altitude relative to sea level or relative to a standard zero point.

[0052] The embodiments and examples described here can be combined with one another in any desired manner, even if this is not explicitly stated. BRIEF DESCRIPTION OF THE DRAWINGS

[0053] The present invention will be further explained in detail with the help of preferred implementation examples. The accompanying drawings show:

[0054] Figure 1 A flow chart showing a method according to the first aspect,

[0055] Figure 2 A device according to the second aspect is shown,

[0056] Figure 3 A computer program according to the fourth aspect is shown,

[0057] Figure 4 shows a diagram showing the bending process and the derivative of the bending process,

[0058] Figure 5 Show according to Figure 4 A partial enlargement of the diagram,

[0059] Figure 6 and Figure 7 shows the detected road curves belonging to the road nodes,

[0060] Figure 8 and Fig. 9 The detected road nodes and

[0061] Fig.10 A block diagram is shown. DETAILED DESCRIPTION

[0062] Figure 1 A flowchart showing a method for detecting road nodes includes the following steps:

[0063] receiving 101 position data describing a corresponding position change process of a motor vehicle travelling on a road comprising a road node and passing through the road node,

[0064] Obtain the corresponding bending change process of the position change process 103,

[0065] Obtaining 105 a corresponding derivative of the bending change process, in particular a numerical derivative, for obtaining a corresponding change of the bending change process with respect to the corresponding position change process,

[0066] Road nodes are detected 107 based on the respectively determined derivatives of the curvature course.

[0067] For a detected road node, for example, a position of the detected road node is determined.

[0068] The detected road nodes are added to a digital road map, for example.

[0069] Figure 2 An apparatus 201 is shown, which is configured to perform all steps of the method according to the first aspect.

[0070] Figure 3 A machine-readable storage medium 301 is shown on which a computer program 303 is stored. The computer program 303 comprises instructions which, when executed by a computer, cause the computer to perform the method according to the first aspect.

[0071] Figure 4 A diagram 401 is shown which illustrates a bending profile and a derivative of the bending profile with respect to the distance, ie with respect to the position profile.

[0072] The distance traveled by the associated motor vehicle is plotted in meters on an abscissa 403 of the diagram 401 . The curvature profile and the derivative of the curvature profile are plotted in arbitrary units on an ordinate 405 of the diagram 401 .

[0073] The distance, ie the distance travelled, can be ascertained, for example, based on position data.

[0074] Reference numeral 407 shows a curvature profile which has been additionally smoothed, wherein the smoothed curvature profile is denoted by reference numeral 409 .

[0075] Reference numeral 411 illustrates a change in the curvature profile 409 according to the smoothed curvature profile 409 .

[0076] Figure 5 Display basis Figure 4 An enlarged view of chart 401 .

[0077] Reference numeral 415 designates an interval with positive values ​​for the curvature change. Reference numeral 417 designates a second interval with negative values ​​for the curvature change. Intervals 415, 417 each form a pair of directly consecutive intervals each with a positive and a negative curvature change, so that the associated road segment describes a curve.

[0078] At approximately x-coordinate "1350", it would appear that there is also a pair of directly successive sections with a positive and a negative change in curvature, so that the associated road segment describes a curve. However, the distance between the positive section and the negative section is too great here, so that the drive-through is excluded as unlikely. The threshold for "too great" can be an adjustable parameter.

[0079] Figure 6 A road map 601 is shown with detected road curves superimposed on it. For clarity, not all Figure 6 The road curves shown in FIG. 60 are all labeled with their own reference numerals. For some, reference numeral 603 is used. Therefore, it can be determined that the detected road curve 603 belongs to a plurality of road nodes, because the number of exits and entrances to the road curve 603 is greater than or equal to 3.

[0080] Figure 7 Display another road map 701, similar to Figure 6 , the road map is superimposed with a plurality of detected road curves 703. For example, a traffic circle 705 is displayed, which is clearly indicated by the detected road curve 703. The number of entrances and exits into the detected road curve 703 is, for example, currently 3, so that the detected road curve 703 can be defined or determined as belonging to a road node, so that the road node, that is, the traffic circle 705, is detected here.

[0081] Figure 8 A further road map 801 is shown. Reference numeral 803 designates areas, which can also be referred to as geographical areas, within which road nodes 805 are detected according to the approach described here.

[0082] Fig. 9 Another road map 901 is shown, in which similar Figure 8 , a plurality of geographical ranges 903 are entered for which road nodes 905 are detected according to the scheme described herein.

[0083] Fig.10 A block diagram 1001 is shown which exemplarily explains the approach described herein.

[0084] A large amount of lane data 1003 is received from a large number of motor vehicles, which travel on roads that include a road node or also include a plurality of road nodes, wherein the motor vehicles at least partially pass through the road nodes. This lane data 1003 is position data in the sense of this specification.

[0085] Lane data preprocessing occurs according to function block 1005. Lane data 1003 are enriched with data on the bending angle, i.e. the bending of the position change process, for example. Further, smoothing of the bending change process occurs. The corresponding enriched lane data are further processed according to function block 1007. For example, the derivative of the steering angle is calculated for the lane data. For example, further, an outlier value or a deviation value of the bending change is defined as such a value: the value exceeds the standard deviation of all data after scaling with a predetermined scaling factor, for example. The deviation values ​​are then grouped for detecting road curves. The detected road curves are then divided into time boxes (German "Zeitbox" or "Zeitabschnitt") so that they extend no more than a user-defined number of seconds.

[0086] The "division into time boxes" ("timeboxing" in English) is provided to trim the trajectory segments that describe the road nodes into the most specific segments with the strongest curvature. As a result, the polygons generated later by clustering and observation analysis of the clusters become narrower and better describe the road nodes.

[0087] The detected road curve is further processed according to function block 1009 , wherein the points of the detected road curve are respectively clustered, wherein the clusters are stored.

[0088] The clusters are then further processed according to function block 1011. The number of ingress and egress points into and out of the clusters is obtained, wherein clusters whose corresponding number is less than 3 are filtered out. Furthermore, abnormal changes in clusters are filtered out. The remaining clusters, i.e., the clusters that are not filtered out, are detected road nodes 1013.

[0089] The solution described here is based in particular on the following recognition: Road nodes are geographical areas in which multiple roads intersect or one road merges into another road. The roads belonging to the road nodes are therefore connected to each other. Therefore, road traffic in such areas has a much higher turning rate than outside such areas. By identifying and grouping points of the position change process with high turning angles during the position change process of motor vehicles traveling through such areas, road nodes can be effectively identified and defined.

[0090] The method according to the first aspect comprises, for example, one or more of the following steps:

[0091] 1. Data preprocessing: The lane data (i.e., the position change process) describing the geographic location (i.e., longitude and latitude, and optionally altitude) is enriched with data on the steering angle, for example. This means that the steering angle of the position change process is determined for each position, which can also be called a bending angle, i.e., the first-order derivative of the position change process. The steering angle is in particular a discrete approximation of the curvature of the position change process. This value is in particular calculated for each data point, i.e., each position.

[0092] For example, the (especially numerical) derivative of the steering angle, ie the bending change, is calculated for the entire data set, ie for all positions.

[0093] For example, outlier values ​​or values ​​of curvature changes are defined as values ​​that exceed the standard deviation of all data scaled by a predetermined scaling factor, for example. These deviant values ​​or outlier values ​​are identified in the curvature changes data or values.

[0094] For example, for each curve, in particular when using a spacing metric, outliers (i.e. outlier or deviant values) are grouped into (particularly almost continuous) intervals with (completely) negative values ​​or (completely) positive values. The term "almost continuous" here means that there may be a limited number of missing values ​​between two outliers in an interval.

[0095] For example, a curve is defined as a pair of intervals with outliers grouped according to negative and positive changes in curvature, respectively. Thus, the curve is defined in particular by a first interval comprising grouped outliers with positive changes in curvature, wherein the first interval signals that the road begins to curve, wherein the curve is further defined by a second interval directly following the first interval and comprising grouped outliers with negative changes in curvature, wherein the second interval signals that the road straightens out again. In particular, the road curve is identified by connecting the two intervals to each other with the help of a distance metric. The curves are then divided into time bins such that they extend no more than a user-defined number of seconds.

[0096] For example, clustering of the points within each road curve is performed, in particular using a standard clustering algorithm, such as DBSCAN (Density-Based Spatial Clustering of Applications with Noise; German: Dichtebasierte Clusteranalyse mit Rauschen, Density-based spatial cluster analysis with noise). For example, a polygon is created that contains all the points in each cluster.

[0097] In order to distinguish road curves from road nodes, it is provided, for example, that road curves are filtered according to the number of inlets and outlets into the road curves, in particular in the polygons belonging to the road curves. A road node comprises 3 or more inlets and outlets, whereas a road curve has only 1 or 2 inlets and outlets.

[0098] The solution described here makes it possible, in particular, to obtain knowledge about the state of road nodes. The solution described here makes it possible, in particular, to be able to distinguish between roads and road nodes. Based on this knowledge and this ability, an ADAS system (ADAS: Advanced Driver-Assistance System, German: fortschrittliche Fahrerassistenzsysteme, advanced driver assistance system) can better navigate on roads and road nodes in both scenarios in an advantageous manner. The ADAS system can benefit from knowledge about the state of road nodes and about the distinction between roads and road nodes, because it can adjust its driving parameters based on the knowledge of "how to distinguish driver behavior (especially described by position change processes and corresponding derivatives) within the boundaries of road nodes" to better and more effectively deal with this situation.

Claims

1. A method for detecting a road node (705, 805, 809, 1013), comprising the following steps: Receiving (101) position data (1003) describing a corresponding position change process of a motor vehicle traveling on a road including a road node (705, 805, 809, 1013) and passing through the road node (705, 805, 809, 1013), Find the corresponding bending change process of the position change process (103), Obtaining (105) a corresponding derivative of the bending change process, in particular a numerical derivative, for obtaining a corresponding change of the bending change process with respect to the corresponding position change process, Road nodes (705, 805, 809, 1013) are detected (107) based on the respectively determined derivatives of the bending process.

2. The method according to claim 1, wherein: A road curve is detected based on the respectively determined derivatives of the curvature course, wherein road nodes are detected based on the detected road curve (705, 805, 809, 1013).

3. The method according to claim 2, wherein: For each road curve, the number of entrances and exits entering or leaving the road curve is determined, wherein road nodes are detected based on the determined number of entrances and exits (705, 805, 809, 1013).

4. The method according to claim 3, wherein: When the number of entrances and exits is greater than or equal to 3, it is determined that the road curve belongs to the road node (705, 805, 809, 1013).

5. The method according to any one of claims 2 to 4, wherein: Detected road curves having abnormal change processes are filtered out, wherein road nodes are detected based on road curves that are not filtered out (705, 805, 809, 1013).

6. The method according to claim 2 , wherein the method is based on claim 5 with respect to the road curves that are not filtered out, wherein: A polygon is obtained for each road curve (not filtered out), the polygon representing the corresponding road curve, wherein road nodes are detected based on the obtained polygon (705, 805, 809, 1013).

7. The method according to claim 6, in relation to which claim 3 or 4 is relied upon, wherein The number of entrances and exits entering or leaving the corresponding polygon is respectively calculated to correspondingly calculate the number of entrances and exits entering or leaving the road curve, wherein the road node (705, 805, 809, 1013) is detected based on the calculated number of entrances and exits entering or leaving the corresponding polygon.

8. The method according to any one of claims 2 to 7, wherein: The corresponding discrete degrees of the values ​​of the respectively determined derivatives of the bending change process are determined, wherein the values ​​of the respectively determined derivatives of the bending change process that exceed the corresponding discrete degrees are determined, wherein the values ​​that exceed the corresponding discrete degrees are respectively grouped into intervals (415, 417) with positive bending changes and with negative bending changes, wherein the road curve is determined as a pair of directly consecutive intervals (415, 417) having positive bending changes and negative bending changes, respectively.

9. The method according to claim 8, wherein: The corresponding standard deviation of the values ​​of the respectively determined derivative of the bending profile is ascertained, wherein the corresponding degree of dispersion is ascertained based on the corresponding standard deviation.

10. The method according to claim 9, wherein: The corresponding standard deviations are scaled using a predetermined scaling factor that is not equal to 1 to obtain scaled standard deviations respectively, wherein the corresponding discrete degrees are obtained based on the scaled standard deviations respectively.

11. The method according to any one of claims 2 to 10, to which claim 3 or 4 is attached, wherein: The points within the road curve are clustered, wherein corresponding polygons are created in such a way that they include all the points of each cluster.

12. An apparatus (201) configured to carry out all the steps of the method according to one of the preceding claims.

13. Computer program (303) comprising instructions which, when the computer program (303) is executed by a computer, cause the computer to perform the method according to one of claims 1 to 11.

14. A machine-readable storage medium (301) on which a computer program (303) according to claim 13 is stored.

Images