Computer-implemented method and system for classifying traffic conditions

Abstract

The invention relates to a computer-implemented method and a system (1) for classifying predetermined traffic conditions included in a data set of environmental data of a motor vehicle, comprising: applying (S2) a digraph (G) to a first data set (DS1), nodes (11) of the digraph (G) divide the first data set (DS1) into at least one segment of the movement behavior of the own vehicle (10) and / or of the companion vehicle (12) relative to the vehicle environment according to a first condition (14) that is satisfied within a time interval, and edges (15) of the digraph (G) represent the connection between the respective nodes (11); classifying (S3) the predetermined traffic condition if all the determined segments satisfy a second condition (16) of the predetermined traffic condition; and outputting (S4) a second data set (DS2) having respective start and end times of the category (K) representative of the predetermined traffic condition and / or of the segment representative of the predetermined traffic condition.

Description

Technical Field

[0001] The invention relates to a computer-implemented method for classifying predetermined traffic situations comprised by a data set of surrounding data of a motor vehicle.

[0002] The invention further relates to a system for classifying a predetermined traffic situation comprised by a data set of surrounding data of a motor vehicle. Background Art

[0003] In order to create the test scenarios for the simulation, test drives are carried out. The sensor data obtained in this way is then abstracted into logical scenarios.

[0004] The input data are raw data, ie sensor data from real measurement drives in the form of radar echo recordings, 3D point clouds from lidar measurements and image data. The result data are simulated driving scenarios, which include the environment on the one hand and the trajectory on the other hand.

[0005] “Szenario-Optimization for Automatic and Autonomous Driving Systems (Florian Hauer, Bernd Holzmüller, arXiv:1901.05680)” discloses a method for verifying and validating automated and autonomous driving systems, in particular for finding suitable test scenarios for virtual assurance.

[0006] The test method provides for the adaptation of the metaheuristic search in order to optimize the scenario. For this purpose, a suitable search space and a suitable quality function must be established. The parameterized scenarios are derived based on an abstract description of the functionality and application of the system.

[0007] It is assumed that certain parameters have a strong influence on the situation. For example, the situation in which emergency braking occurs is mainly determined by the speed of the two road users. Therefore, it is crucial for the user to understand the distribution of these parameter values ​​and know which parts are not covered by the data or simulation. The missing data points must be collected in reality or in the simulation.

[0008] Therefore, it is desirable for users to perform efficient evaluation of the generated data sets in order to identify critical situations and to test in slightly altered versions of the same simulation scenario.

[0009] Furthermore, most scenarios created for experimenting with autonomous driving functions are determined a priori and are based on expert knowledge of which scenarios are needed for which testing purposes.

[0010] On the other hand, methods based on real data are based on machine learning algorithms, which are difficult to generalize to situations requiring different similar traffic conditions.

[0011] Therefore, there is a need to improve existing methods for analyzing driving scenario datasets so that effective identification and classification of critical situations can be achieved. Summary of the invention

[0012] It is therefore an object of the present invention to provide a method for classifying predetermined traffic situations comprised by a data set of surrounding data of a motor vehicle, which method enables efficient recognition and classification of relevant critical traffic situations.

[0013] This object is achieved according to the invention by a computer-implemented method for classifying predetermined traffic situations comprised by a data set of surrounding data of a motor vehicle having the features of claim 1 .

[0014] Furthermore, the object is achieved according to the invention by a system for classifying predetermined traffic situations comprised by a data set of surrounding data of a motor vehicle having the features of claim 14 .

[0015] The invention relates to a computer-implemented method for classifying predetermined traffic situations comprised by a data set of surrounding data of a motor vehicle.

[0016] The method comprises providing a first data set, in particular a data stream, of sensor data of a travel of the own vehicle and / or a companion vehicle, which is recorded by at least one vehicle-side surroundings recognition sensor and comprises a plurality of traffic situations.

[0017] In addition, the method includes applying a directed graph to the first data set, wherein the nodes of the directed graph divide the first data set into at least one segment of the movement behavior of the self-vehicle and / or companion vehicle relative to the vehicle environment according to a first condition satisfied within a time interval, and the edges of the directed graph represent connections between the corresponding nodes.

[0018] Furthermore, the method comprises: if all determined segments satisfy a second condition of the predetermined traffic condition, classifying the predetermined traffic condition and outputting a second data set having a category representing the predetermined traffic condition and / or corresponding start times and end times of the segments representing the predetermined traffic condition.

[0019] The invention further relates to a system for classifying a predetermined traffic situation comprised by a data set of surrounding data of a motor vehicle.

[0020] The system comprises a data memory configured to provide a first data set, in particular a data stream, of sensor data of a trip of the own vehicle and / or a companion vehicle recorded by at least one vehicle-side surroundings recognition sensor and comprising a plurality of traffic situations.

[0021] In addition, the system includes a first computing unit, which is configured to apply a directed graph to a first data set, wherein the nodes of the directed graph divide the first data set into at least one segment of the movement behavior of the self-vehicle and / or companion vehicle relative to the vehicle environment according to a first condition satisfied within a time interval, and the edges of the directed graph represent connections between the corresponding nodes.

[0022] The system includes a second computing unit and a data output unit, wherein the second computing unit is configured to classify the predetermined traffic condition if all determined segments meet a second condition of the predetermined traffic condition, and the data output unit is configured to output a second data set having a category representing the predetermined traffic condition and / or corresponding start times and end times of the segments representing the predetermined traffic condition.

[0023] Furthermore, the invention relates to a computer program having a program code for carrying out the method according to the invention for classifying a predetermined traffic situation comprised by a data set of surrounding data of a motor vehicle, when the computer program is executed on a computer.

[0024] The idea of ​​the present invention is to improve the data selection and, if necessary, based thereon, in a subsequent step, to improve the parameter extraction of the traffic situation contained in the data set of sensor data of the travel of the own vehicle and / or the companion vehicle recorded by a first plurality of vehicle-side surroundings recognition sensors.

[0025] Thus, relevant situations can be automatically searched for in the data and only the correspondingly extracted traffic situations can then be brought into the simulation.

[0026] Therefore, the present invention provides a scenario selection method which can be used for identifying traffic situations from real-world data.The proposed method is graph-based and is able to meaningfully define traffic situations that should be identified in the provided raw measurement data.

[0027] The graph-based structure is intrinsically linked to situation similarity—similar situations are mapped onto corresponding similarity graphs.

[0028] The graph-based model (Vorlage) can be easily modified to recognize different scenarios, and graphs can be reused and combined to describe more complex traffic behaviors. Thus, multiple graphs used in parallel can describe the entire data flow with the help of blocks of recognized traffic situations (e.g. cut-ins).

[0029] The overlay of the data stream can be viewed directly with the identified traffic conditions. In addition, the user can specify the model to be identified without writing code. The user can decide how simple or detailed the model should be.

[0030] In this case, a traffic situation may be understood to mean a small, basic and / or larger scene or traffic scene involving a plurality of road users.

[0031] Other embodiments of the present invention are the technical solutions of other dependent claims and the following description with reference to the drawings.

[0032] According to a preferred embodiment of the invention, it is provided that predetermined traffic situations to be classified are predetermined, and the directed graph respectively determines segments of the predetermined traffic situation. Thus, it is advantageously possible to determine or classify binary in accordance with the regulations whether the segment included in the corresponding traffic situation is present in the first data set.

[0033] According to another preferred embodiment of the invention, each node has at least one input and exactly one output, the at least one input of each node being given via an edge to another node or via a first data set. The node can thus process one or more inputs of the preceding layer, so that the vehicle behavior of a higher logical or abstract layer can be determined with the aid of each subsequent layer.

[0034] According to another preferred embodiment of the invention, each node has an algorithm which is applied to the input data of the node and which classifies whether a first condition of the movement behavior of the own vehicle and / or the companion vehicle relative to the vehicle environment is met.

[0035] The classification here is a binary classification, ie it is determined whether a predetermined condition relating to the segment is true or false.

[0036] According to another preferred embodiment of the invention, the algorithm of the corresponding node outputs the corresponding start and end times of the segments representing the movement behavior when the first condition is met. This makes it possible to advantageously determine the relevant segments in which a specific vehicle behavior occurs.

[0037] According to another preferred extension of the present invention, it is stipulated that the segment of the movement behavior of the own vehicle and / or the companion vehicle relative to the vehicle environment is a time interval that includes the start time and the end time of the movement behavior of the own vehicle and / or the companion vehicle relative to the vehicle environment included in the first data set, in particular the data stream.

[0038] Thus, starting from a first data set of a data stream comprising sensor data of a plurality of traffic situations, a corresponding time segment can be determined which includes the relevant movement behavior of the own vehicle and / or the companion vehicle.

[0039] According to another preferred extension scheme of the present invention, it is stipulated that the directed graph has a first layer including at least one node, the input of the node is a first data set, the second layer of the directed graph including at least one first node and a second node divides the first data set into at least one segment of the movement behavior of the own vehicle and / or companion vehicle relative to the vehicle environment according to a first condition, and if the combination of the determined time intervals meets the second condition of the predetermined traffic condition, the third layer of the directed graph including at least one node classifies the predetermined traffic condition using the time intervals output by the nodes of the second layer.

[0040] The corresponding layers of the graph thus fulfill the task of continuously segmenting the first data set according to corresponding regulations in order to check, based on the determined segments, whether a higher-level traffic situation is covered by the data.

[0041] According to another preferred embodiment of the present invention, the second condition of the predetermined traffic situation specifies that the segments of the own vehicle and / or at least one segment of the movement behavior of the companion vehicle determined according to the first condition are carried out in a predetermined sequence and / or within a predetermined time interval. This ensures that the determined segments occur in a corresponding sequence, which characterizes the predetermined traffic situation.

[0042] According to another preferred embodiment of the invention, the movement behavior of the vehicle and / or the companion vehicle relative to the vehicle environment includes all vehicle actions that are detectable in the first data set and represent the movement behavior. Thus, all data recorded by different types of sensors and events detectable in the data can be used to identify the predetermined traffic situation.

[0043] According to another preferred extension of the present invention, the movement behavior of the self-vehicle and / or the companion vehicle relative to the vehicle environment is the lateral behavior and / or longitudinal behavior of the self-vehicle relative to the traffic infrastructure and / or at least one companion vehicle and / or the lateral behavior and / or longitudinal behavior of the companion vehicle relative to the traffic infrastructure and / or at least one self-vehicle, and the movement behavior of the self-vehicle and / or the companion vehicle relative to the vehicle environment includes lane keeping, lane changing, turning, constant or changing acceleration and the resulting speed, flashing brake lights, passing objects in the vehicle environment and / or recognizing traffic signs.

[0044] Thus, the movement behavior of the own vehicle detectable in the first data set relative to the traffic infrastructure and / or the at least one companion vehicle and vice versa includes a plurality of different behaviors of the participating vehicles.

[0045] According to another preferred embodiment of the present invention, the driving sensor data of the own vehicle and / or the companion vehicle recorded by at least one vehicle-side environment recognition sensor are position data of a GNSS sensor, IMU data, camera data, lidar data, radar data and / or ultrasonic data. The sensor data can therefore be obtained from a plurality of different data sources.

[0046] According to another preferred embodiment of the invention, a virtual test for verifying the automated driving function of the motor vehicle is performed based on the output second data set. Thus, the determined traffic situation can form the basis for a virtual test for verifying the automated driving function of the motor vehicle.

[0047] According to another preferred extension of the present invention, the output of the directed graph and the output of another directed graph are used to classify another predetermined traffic situation, and if the combination of the output of the directed graph and the output of the other directed graph meets the third condition of the other predetermined traffic situation, the other predetermined traffic situation is classified. Therefore, a combination of multiple determined traffic situations can be used in an advantageous manner to determine another traffic situation.

[0048] Features of the computer-implemented method described herein for classifying predetermined traffic conditions comprised by a data set of surrounding data of a motor vehicle can also be applied to the system according to the invention for classifying predetermined traffic conditions comprised by a data set of surrounding data of a motor vehicle and vice versa. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] For a better understanding of the present invention and its advantages, reference is now made to the following description taken in conjunction with the accompanying drawings.

[0050] The invention is explained in more detail below with reference to exemplary embodiments, which are shown in the schematic drawings of the accompanying drawings. The drawings are as follows:

[0051] Figure 1a A flow chart showing a computer-implemented method for classifying a predetermined traffic situation comprised by a data set of surrounding data of a motor vehicle according to a preferred embodiment of the present invention;

[0052] Figure 1b A schematic diagram showing a cut-in scene according to a preferred embodiment of the present invention;

[0053] Figure 2 A diagram showing a method for determining a cut-in scene according to a preferred embodiment of the present invention;

[0054] Figure 3 A schematic diagram showing a trajectory of a vehicle according to a preferred embodiment of the present invention;

[0055] Figure 4A schematic diagram showing a companion vehicle trajectory according to a preferred embodiment of the present invention;

[0056] Figure 5 A schematic diagram showing vehicle maneuvers during different time intervals according to a preferred embodiment of the present invention;

[0057] Figure 6 Schematic diagram showing different scenarios according to a preferred embodiment of the present invention;

[0058] Figure 7 A diagram showing a method for determining a cut-in scene according to a preferred embodiment of the present invention;

[0059] Figure 8 A diagram showing a method for determining a drift-out scenario according to a preferred embodiment of the present invention;

[0060] Fig. 9 A diagram showing a method for determining an overtaking manoeuvre according to a preferred embodiment of the present invention; and

[0061] Fig.10 A schematic diagram of a system for classifying a predetermined traffic situation comprised by a data set of surrounding data of a motor vehicle according to a preferred embodiment of the present invention is shown.

[0062] Unless otherwise indicated, the same reference numerals refer to the same elements of the drawings. DETAILED DESCRIPTION

[0063] Figure 1a A flow chart of a computer-implemented method for classifying a predetermined traffic situation comprised by a data set of surrounding data of a motor vehicle is shown according to a preferred specific embodiment of the present invention.

[0064] The method comprises providing S1 a first data set DS1 , in particular a data stream, of sensor data of the travel of own vehicle 10 and / or companion vehicle 12 , which is recorded by at least one vehicle-side surroundings recognition sensor and comprises a plurality of traffic situations.

[0065] In addition, the method includes applying S2 a directed graph G to a first data set DS1, wherein the nodes 11 of the directed graph G divide the first data set DS1 into at least one segment of the movement behavior of the self-vehicle 10 and / or the companion vehicle 12 relative to the vehicle environment according to a first condition 14 satisfied within a time interval, and the edges 15 of the directed graph G represent connections between corresponding nodes 11.

[0066] Furthermore, the method comprises: classifying S3 the predetermined traffic situation if all determined segments satisfy the second condition 16 of the predetermined traffic situation; and outputting S4 a second data set DS2 having a category K representing the predetermined traffic situation and / or corresponding start and end times of the segments representing the predetermined traffic situation.

[0067] According to the Cut-In-Szenario Figure 1b The example shown in FIG. 1 is used to illustrate the solution according to the present invention. In this scenario, the vehicle 10 is traveling in the left lane until it reaches another vehicle or companion vehicle 12 traveling in the middle lane and then changes lanes to the lane of the other vehicle.

[0068] In order to simplify the description of scenarios such as cut-ins, Allen interval algebra is used within the scope of the following description, with the help of which relations between time intervals can be defined.

[0069] In the context of a scene, the basic motion behavior of a vehicle is used as a time interval for describing the motion of the vehicle. For example, the motion of the vehicle 10 can be divided into three intervals: an interval for keeping the lane, followed by an interval for changing lanes, followed by another interval for keeping the lane.

[0070] By applying Allen interval algebra to the time intervals used to describe the vehicle motion, the cut-in scenario can be described in three simple statements:

[0071] Self-driving sequence:

[0072] Lane keeping meets lane changing meets lane keeping.

[0073] Companion vehicle sequence:

[0074] Lane keeping.

[0075] Furthermore, the ego vehicle sequence occurs during the companion vehicle sequence.

[0076] On the basis of this description, a computational graph G is created in order to determine whether a scenario given as input can be categorized or classified as an entry scenario. Figure 2 The resulting graph G for identifying cut-in scenes is shown in FIG.

[0077] The graph G comprises nodes 11, which are connected by edges 15. Each node 11 has one or more inputs and always produces one output. The inputs of a node 11 are defined by the connected edges 15 of other nodes 11. Each node 11 performs a test that returns the time interval of the fragment in which the test of the input time interval is true.

[0078] The graph G includes different types of nodes 11. The first layer 17a nodes 11 of the graph G are input data based on the motion data of the vehicle, such as the vehicle trajectory. The second layer 17b of the graph G are nodes 11 that analyze the trajectory and divide it into time intervals, such as "lane keeping".

[0079] These time intervals are then combined by subsequent nodes 11 to check whether the temporal correlation between the segmented time intervals corresponds to, for example, an "encounter" or "meet" scene description. After passing through the last node 11 of the graph G, the identified time interval of the cut-in scene is output.

[0080] The creation of graph G is described below. The first step for determining the scene or traffic situation is to segment the trajectory of the traffic participant into time intervals in order to describe the movement semantically. For this purpose, a node 11 is used which implements a segmentation algorithm, which can segment the trajectory into lateral time intervals or segments, for example into a "lane keeping" segment and a "lane change" segment for a cut-in scenario.

[0081] For the vehicle 10 in the cut-in example in FIG1 , the trajectory can be divided into Figure 3 As described above, the segmentation of the trajectory results in three time intervals: "lane keeping", "lane change" and another interval "lane keeping".

[0082] If the same segmentation is performed for the trajectory of the companion vehicle 12 in this example, only one “lane keeping” segment ( Figure 5 ).

[0083] After the segmentation, the time intervals are reassembled with the help of nodes 11, which are checked based on Allen interval algebra. The first node 11 based on Allen interval algebra (which is used for the graph G of the cut-in scene) is "Meet". This node 11 "Meet" has two inputs: the time intervals "Lane Keeping" and "Lane Change" from the segmentation node 11 of the previous layer of the graph G.

[0084] The check of the node 11 "encountering" provides a time interval in which the "lane change" time interval directly follows the "lane keeping" time interval. Figure 5 In the example in , this is the case for the first time interval of the segmentation results of "lane keeping" and "lane changing". The output of this node 11 is the combined time interval of the "lane keeping" and "lane changing" time intervals.

[0085] The same process applies to the next "Meet" node 11, which checks whether another "Lane Keeping" time interval follows directly after the output of the first "Meet" node 11. The last node 11 "During" of the graph G checks whether the time interval of the output of the "Meet" node 11 (which describes the sequence of the ego vehicle 10) is within the same time interval as the identified "Lane Keeping" time interval of the companion vehicle 12.

[0086] If we only consider the temporal description of the vehicle motion, we can see that it is possible to determine a number of other scenarios with the aid of the graph G, whereas the graph G itself is only designed to identify the cut-in scenario ( Figure 6 ). In all four scenarios, the ego vehicle and companion vehicle perform the same motion if the additional scenario parameters are ignored.

[0087] Therefore, an additional node 11 must be added to the graph G in order to identify additional scene parameters, which in the case of a cut-in scene are spatial parameters used to describe the relative position of the vehicle and the relative lane in which the vehicle is traveling. In order to distinguish the cut-in scenes, two additional conditions can be added:

[0088] 1) The host vehicle and the companion vehicles 10 and 12 arrive at the destination in the same lane.

[0089] 2) The vehicle 10 arrives at the destination ahead of the companion vehicle 12 .

[0090] If we transform these two conditions and add them to the graph for detecting cut-in scenes, we end up with a graph like Figure 7 As shown.

[0091] In the second graph G, there is again a first layer 17a with nodes 11 “Same lane” and “Own vehicle in front of companion vehicle”, which perform a segmentation based on the vehicle trajectories and output the time intervals in which both vehicles are in the same lane and the time intervals in which own vehicle 10 is in front of companion vehicle 12 .

[0092] In the second layer 17b of the second graph G, the time intervals are calculated: whether the detected time intervals of the first graph G are ended by the time intervals output by the "same lane" and "self-vehicle in front of companion vehicle" nodes. The last node 11 "during" finally checks whether the obtained time intervals occur in the same duration. If there is a time interval left at the end, the cut-in scene is recognized.

[0093] exist Figure 8 In the example, only three nodes 11 of the graph G used to identify the cut-in scene need to be replaced to identify the drift-out scene, that is, "front" is replaced by "back", and "end" is replaced by "start".

[0094] exist Fig. 9 2 shows how a graph G once defined for identifying outgoing and cutting-in scenarios can be redefined as nodes 11 and then combined in order to identify more complex traffic scenarios, such as overtaking scenarios.

[0095] Fig.10 A schematic diagram of a system for classifying a predetermined traffic situation comprised by a data set of surrounding data of a motor vehicle according to a preferred embodiment of the present invention is shown.

[0096] System 1 includes a data memory 20 which is configured to provide a first data set DS1 , in particular a data stream, of sensor data of the travel of own vehicle 10 and / or companion vehicle 12 recorded by at least one vehicle-side surroundings recognition sensor and comprising a plurality of traffic situations.

[0097] In addition, the system 1 includes a first computing unit 22, which is configured to apply a directed graph G to a first data set DS1, wherein the nodes 11 of the directed graph G divide the first data set DS1 into at least one segment of the movement behavior of the self-vehicle 10 and / or the companion vehicle 12 relative to the vehicle environment according to a first condition 14 satisfied within a time interval, and wherein the edges 15 of the directed graph G represent connections between the corresponding nodes 11.

[0098] In addition, the system 1 includes a second computing unit 24 and a data output unit 26, wherein the second computing unit is configured to classify the predetermined traffic condition if all determined segments meet the second condition of the predetermined traffic condition, and the data output unit is configured to output a second data set DS2 having a category K representing the predetermined traffic condition and / or corresponding start times and end times of the segments representing the predetermined traffic condition.

[0099] Reference numerals list

[0100] 1 System

[0101] 10 Self-driving

[0102] 11 nodes

[0103] 12 Companion Vehicles

[0104] 14First condition

[0105] 15 sides

[0106] 16 Second Condition

[0107] 18 The Third Condition

[0108] 20 Data storage

[0109] 22 First computing unit

[0110] 24 Second computing unit

[0111] 26 data output units

[0112] A Algorithm

[0113] G-graph

[0114] K Category

[0115] DS1 First Dataset

[0116] DS2 Second Dataset

[0117] S1-S4 Method Steps

Claims

1. A computer-implemented method for classifying a predetermined traffic situation comprised by a data set of environmental data of a motor vehicle, The following steps are involved: Providing (S1) a first data set (DS1), in particular a data stream, of sensor data of the driving of the own vehicle (10) and / or a companion vehicle (12), recorded by at least one vehicle-side environment recognition sensor, comprising a plurality of traffic situations; Applying (S2) a directed graph (G) to a first data set (DS1), wherein the nodes (11) of the directed graph (G) divide the first data set (DS1) into at least one segment of the movement behavior of the ego vehicle (10) and / or the companion vehicle (12) relative to the vehicle environment according to a first condition (14) satisfied in a time interval, and the edges (15) of the directed graph (G) represent connections between corresponding nodes (11); If all determined segments satisfy the second condition (16) of the predetermined traffic condition, the predetermined traffic condition is classified (S3); and A second data set (DS2) having a category (K) representing the predetermined traffic situation and / or corresponding start time and end time of a segment representing the predetermined traffic situation is output (S4).

2. The computer-implemented method according to claim 1, in, A predetermined traffic situation to be classified is predetermined, and the directed graph (G) respectively determines a segment of the predetermined traffic situation.

3. The computer-implemented method according to claim 1 or 2, in, Each node (11) has at least one input and exactly one output, the at least one input of each node (11) being given via an edge of another node (11) or via a first data set (DS1).

4. A computer-implemented method according to any one of the preceding claims, in, Each node (11) has an algorithm (A) applied to input data of the node (11), the algorithm (A) classifying whether a first condition (14) of the movement behavior of the ego vehicle (10) and / or the companion vehicle (12) relative to the vehicle environment is met.

5. The computer-implemented method of claim 4, in, The algorithm (A) of the corresponding node (11) outputs the corresponding start time and end time of the segment representing the movement behavior when the first condition (14) is met.

6. A computer-implemented method according to any one of claims 1 to 4, in, The segment of the movement behavior of the self-vehicle (10) and / or the companion vehicle (12) relative to the vehicle environment is a time interval that includes the start time and the end time of the movement behavior of the self-vehicle (10) and / or the companion vehicle (12) relative to the vehicle environment, which is included in the first data set (DS1), in particular the data stream.

7. The computer-implemented method of claim 6, in, The directed graph (G) has a first layer (17a) including at least one node (11), the input of which is a first data set, and a second layer (17b) of the directed graph (G) including at least one first node (11) and a second node (11) divides the first data set (DS1) into at least one segment of the movement behavior of the vehicle (10) and / or the companion vehicle (12) relative to the vehicle environment according to the first condition (14), and if the determined combination of time intervals meets the second condition (16) of the predetermined traffic condition, the third layer (17c) of the directed graph (G) including at least one node (11) classifies the predetermined traffic condition by using the time interval output by the node (11) of the second layer (17b).

8. The computer-implemented method of claim 7, in, The second condition (16) of the predetermined traffic situation states that the segments of the own vehicle (10) and / or at least one segment of the movement behavior of the companion vehicle (12) determined according to the first condition (14) are performed in a predetermined sequence and / or within a predetermined time interval.

9. A computer-implemented method according to any one of the preceding claims, in, The movement behavior of the ego vehicle (10) and / or companion vehicle (12) relative to the vehicle environment includes all vehicle actions that are detectable in a first data set (DS1) and represent the movement behavior.

10. A computer-implemented method according to any one of the preceding claims, in, The movement behavior of the self-vehicle (10) and / or the companion vehicle (12) relative to the vehicle environment is the lateral behavior and / or longitudinal behavior of the self-vehicle (10) relative to the traffic infrastructure and / or at least one companion vehicle (12) and / or the lateral behavior and / or longitudinal behavior of the companion vehicle (12) relative to the traffic infrastructure and / or at least one self-vehicle (10). The movement behavior of the self-vehicle (10) and / or the companion vehicle (12) relative to the vehicle environment includes lane keeping, lane changing, turning, constant or changing acceleration and the resulting speed, flashing brake lights, passing objects in the vehicle environment and / or recognizing traffic signs.

11. A computer-implemented method according to any one of the preceding claims, in, The driving sensor data of the own vehicle (10) and / or companion vehicle (12) recorded by at least one vehicle-side environment recognition sensor are position data of a GNSS sensor, IMU data, camera data, lidar data, radar data and / or ultrasonic data.

12. A computer-implemented method according to any one of the preceding claims, in, A virtual test for verifying an automated driving function of the motor vehicle is performed based on the output second data set (DS2).

13. A computer-implemented method according to any one of the preceding claims, in, The output of the directed graph (G) and the output of another directed graph (G) are used to classify another predetermined traffic condition, and if the combination of the output of the directed graph (G) and the output of the other directed graph (G) satisfies the third condition (18) of the other predetermined traffic condition, the other predetermined traffic condition is classified.

14. System (1) for classifying a predetermined traffic situation comprised by a data set of environmental data of a motor vehicle, include: a data memory (20) configured to provide a first data set (DS1), in particular a data stream, of sensor data of the travel of the own vehicle (10) and / or of a companion vehicle (12), recorded by at least one vehicle-side environment recognition sensor, comprising a plurality of traffic situations; - a first computing unit (22) configured to apply a directed graph (G) to a first data set (DS1), wherein the nodes (11) of the directed graph (G) segment the first data set (DS1) into at least one segment of a movement behavior of the ego vehicle (10) and / or the companion vehicle (12) relative to the vehicle environment, respectively, according to a first condition (14) satisfied within a time interval, and the edges (15) of the directed graph (G) represent connections between corresponding nodes (11); a second computing unit (24) configured to classify the predetermined traffic situation if all determined segments satisfy a second condition of the predetermined traffic situation; and - a data output unit (26) configured to output a second data set (DS2) having a category (K) representing a predetermined traffic situation and / or a corresponding start time and end time of a segment representing a predetermined traffic situation. 15 . A computer program having a program code for carrying out the method according to claim 1 for classifying a predetermined traffic situation comprised by a data set of surrounding data of a motor vehicle when the computer program is executed on a computer.

Images