Computer-implemented method and system for analysing a traffic scene

Abstract

The proposal relates to a computer-implemented method for analysing a traffic scene comprising mutually influencing traffic flows, in which map information about mutually influencing routes in the traffic scene is provided and in which state information comprising at least information pertaining to the respective position and pertaining to the respective state of motion is determined for an ego vehicle and at least one other party to the traffic scene. According to the invention, the map information and the state information of the ego vehicle are used to determine at least one possible route of the ego vehicle as an ego route (3). Additionally, the map information and the state information are used to determine at least one route of the other party that influences the ego route (3) as a conflict route (4). At least one zone graph per conflict route (4) is then determined for the ego vehicle. Each zone graph comprises at least one conflict zone (33), the conflict zone (33) being determined on the basis of the region of overlap between the ego route (3) and the conflict route (4). The traffic scene is finally analysed on the basis of the at least one zone graph, specifically from the point of view of the ego vehicle.

Description

[0001] Description

[0002] title

[0003] -in the Procedure and system for one

[0004] Traffic scene

[0005] State of the art

[0006] The invention relates to a computer-implemented method for analyzing a traffic scene with influencing traffic flows.

[0007] It is known to base such an analysis on map information that is made available, for example, via an on-board server of an ego vehicle, a server external to the vehicle, or a cloud. The use of map information generated on the fly based on aggregated scene-specific environmental information is also known. The map information usually includes information about influencing routes in the traffic scene, such as intersections, junctions, or lane merging, as well as information on traffic regulations, in particular right-of-way rules. For the analysis of a traffic scene, information about the state of the ego vehicle and, if applicable, other participants in the traffic scene is also usually determined in order to take their positions and movement states into account.

[0008] In der Veröffentlichung, M. Butz et al., "SOCA: Domain Analysis for Highly Automated Driving Systems," 2020 IEEE 23rd International Conference on Intelligent Transportation Systems (ITSC), 2020, pp. 1 -6, doi:

[0009] 10.1109 / ITSC45102.2020.9294438., the so-called SOCA method is presented. Using the SOCA method, traffic situations are analyzed with the aim of determining boundary conditions for or requirements regarding the behavior of an automated ego-vehicle in the respective traffic situation. To do this, an abstract description of the traffic situation to be analyzed is first created. This description uses so-called zone graphs. A zone graph abstracts the traffic situation to be analyzed by representing the real road situation using a corresponding abstract traffic infrastructure element (static road geometry) with different zones that are relevant to the driving intention of the ego-vehicle but are initially not specified in terms of their size or position. The different zones can represent different map areas, possible traffic flows, objects, etc.Based on this abstract description of the traffic situation, the possible developments or behavior of the road users involved are then determined and morphologically analyzed to determine boundary conditions for the behavior of the ego vehicle in the analyzed traffic situation. It is noteworthy that the results or boundary conditions obtained in this way initially apply to all traffic situations with the same zone graph. A specification is only achieved by conditioning the results with the situation-specific parameters of the analyzed traffic situation.

[0010] The application of the SOCA method in the context of behavioral planning of an ego vehicle requires the definition of different zones within a traffic situation or scene. These zones must be defined with sufficient geometric precision for the method to deliver meaningful results. To date, the definition of the different zones has mostly been done manually, based on digital map information. However, this is very complex both in terms of initial creation and maintenance, and is impractical for real-world applications.

[0011] Disclosure of the invention

[0012] The measures according to the invention enable an automatic definition of concrete geometric zones of a zone graph for an ego vehicle in a traffic scene with influencing traffic flows. According to the invention, at least one possible route of the ego vehicle is determined as an ego route using the map information and the status information of the ego vehicle. Furthermore, at least one route of the further participant that influences the ego route is determined as a conflict route using the map information and the status information. At least one zone graph is then determined for each conflict route for the ego vehicle, wherein each zone graph comprises at least one conflict zone that is determined on the basis of the overlap area between the ego route and the conflict route. According to the invention, the traffic scene is analyzed for the ego vehicle, specifically on the basis of the at least one zone graph determined in this way.

[0013] As already mentioned, the method according to the invention serves to analyze traffic scenes with influencing traffic flows, i.e. with overlapping areas between possible traffic flows. Such overlapping areas require special attention, since conflicts can arise here between the participants in the traffic scene. The method according to the invention is characterized in that not all possible overlapping areas of possible traffic flows in a given traffic scene are considered, but only overlapping areas that could be relevant for the ego vehicle in the given traffic scene. For this purpose, according to the invention, only zone graphs are created for the ego route of the ego vehicle and all possible resulting conflict routes of other participants in the traffic scene in order to examine only the traffic flows on these routes.

[0014] The method according to the invention can be used particularly advantageously in the context of behavior planning for at least partially automated vehicles that drive in lanes. Behavior planning always requires an analysis of the current traffic scene of the ego vehicle. This analysis is based on an automatic division of the ego vehicle's surroundings into areas or zones that impact the ego vehicle's behavior planning in different ways. Thus, based on information about intersecting or overlapping traffic flows along the route of the automated vehicle, the geometric zones relevant for driving decisions are automatically generated. Using these zones, rules for the possible, permissible, and / or desired behavior of the ego vehicle are then defined or specified and taken into account in behavior planning.

[0015] Such rules or conditions are also used in the description of classic requirements for automated driving in a given Operational Design Domain (ODD). In this context, too, a generic, abstract description of traffic situations that can occur in various concrete traffic scenarios is required.

[0016] Preferably, within the scope of the method according to the invention, scene-specific information is aggregated, which primarily serves to determine information about the status of the ego vehicle and, if applicable, other participants in the traffic scene, but can also be used to generate map information. Furthermore, the scene-specific information can be used to determine at least one ego route of the ego vehicle and / or to determine conflict routes of other participants. The scene-specific information can be both in-vehicle and external sensor data and GPS data. Advantageously, the aggregated scene-specific information also includes the results of environment detection with object detection, recognition of turn signal signals, traffic signs, traffic lights, road markings, right-of-way rules, etc.

[0017] It proves particularly advantageous to consider a destination and / or a navigation specification for the ego vehicle when determining the ego route, as used, for example, in automated and semi-automated driving. This generally allows the analysis of the traffic scene to be limited to a smaller number of zone graphs.

[0018] Particularly in lane-bound driving, it often proves useful and advantageous to define a pre-conflict zone and / or a post-conflict zone for the ego vehicle's zone graph in addition to a conflict zone. A pre-conflict zone is located directly in front of the conflict zone in the direction of travel of the ego vehicle, and a post-conflict zone is located directly behind the conflict zone in the direction of travel of the ego vehicle. This allows for a more detailed analysis of traffic scenes with interfering multi-lane roadways.

[0019] Regardless, the analysis of the traffic scene can in some cases also be improved by determining a first approach zone and / or a first exit zone on the ego-vehicle route as a supplement to the zone graph. Such a first approach zone is always located in front of the conflict zone in the direction of travel of the ego-vehicle and, if applicable, immediately in front of the pre-conflict zone, while such a first exit zone is always located behind the conflict zone in the direction of travel of the ego-vehicle and, if applicable, immediately behind the post-conflict zone.

[0020] Analogous to the first approach zone and the first exit zone on the ego route, the zone graph can also include at least one second approach zone and at least one second exit zone on the conflict route. Such a second approach zone would then be located in front of the conflict zone in the direction of travel of the other participant, while such a second exit zone would be located behind the conflict zone in the direction of travel of the other participant.

[0021] In a preferred embodiment of the method according to the invention, movement corridors for the ego vehicle and the other participants in the traffic scene are determined based on the map information. These corridors then form the basis for the geometric definition of the individual zones of the zone graph, in particular the conflict zone, the pre-conflict zone, and the post-conflict zone, as well as the approach zone and the exit zone. For example, the lane in which a participant is moving could be defined across its entire width as the movement corridor of the respective participant, or even several parallel, co-directional lanes could be defined if the possibility of changing lanes is to be taken into account.

[0022] The conflict zone could then simply be defined as the overlap area between the movement corridor of the ego vehicle on the ego route and the movement corridor of the other participant on the conflict route.

[0023] The pre-conflict zone and the post-conflict zone could each be defined as the overlap area between the movement corridor of the ego vehicle on the ego route and a movement corridor that directly borders the movement corridor of the other participant on the conflict route, but within which no other participant has been detected. This could, for example, be another lane that runs parallel to the movement corridor of the other participant, but in which no other participant is currently located.

[0024] The first and second approach zones as well as the first and second exit zones could each be defined as a non-overlapping area of ​​the movement corridor of the ego vehicle on the ego route or of the other participant on the conflict route, with the approach zone being located in front of the conflict zone in the direction of travel and the exit zone being located behind the conflict zone in the direction of travel. When defining the length of the approach zones and / or the exit zones, the respective vehicle length and / or speed are advantageously taken into account. Furthermore, geographical information about the course of the ego route can be considered, such as gradient, curvature, visual obstructions, etc., and traffic regulations, in particular right-of-way rules and the permitted maximum speed.

[0025] It should be expressly noted here that individual zones can also consist of several connected or non-connected polygons to represent a specific geometry of streets and intersections. Furthermore, in special cases, the zone graph defined according to the invention can also include additional zones to account for the diversity of street and intersection situations. In addition to the method described above, the present invention also relates to a computer-implemented system for analyzing a traffic scene with interfering traffic flows.Such a system comprises an input stage for aggregating scene-specific information, a provision stage for providing map information, in particular map information about influencing routes in the traffic scene, and a processing stage for the scene-specific information for determining state information of an ego vehicle and at least one other participant in the traffic scene.

[0026] The system according to the invention further comprises an evaluation stage designed to use the map information and the status information to determine at least one possible route of the ego vehicle as an ego route and to determine at least one route of the further participant that influences the ego route as a conflict route. Furthermore, the invention provides an abstraction stage designed to determine at least one zone graph for the ego vehicle for each conflict route, which zone graph comprises at least one conflict zone. Furthermore, the system according to the invention comprises an analysis stage for analyzing the traffic scene for the ego vehicle based on the at least one zone graph.

[0027] In a preferred embodiment of the system according to the invention, the provision stage is further configured to determine movement corridors based on the map information. Furthermore, the abstraction stage is configured to use these movement corridors as the basis for the geometric definition of the individual zones of the zone graph, in particular the conflict zone, a pre-conflict zone, a post-conflict zone, as well as approach zones and an exit zone.

[0028] Drawings

[0029] Exemplary embodiments and advantageous developments of the invention are explained in more detail below in conjunction with the figures. Figure 1 illustrates the inventive concept for analyzing a traffic scene with interfering traffic flows using a block diagram.

[0030] Fig. 2a to 2c illustrate the inventive geometric definition of the individual zones of a zone graph using the example of a four-way intersection.

[0031] Description of implementation examples

[0032] In the block diagram of Fig. 1, the individual blocks 10 to 15 each represent a component of a system 1 according to the invention or a functional block of the method according to the invention for analyzing a traffic scene with interfering traffic flows. Interfering traffic flows arise on routes where conflict situations can arise between participants in the traffic scene. These include, for example, intersections, junctions, and lane mergings, as well as pedestrian and bicycle crossings.

[0033] It is essential that the analysis of the traffic scene is always performed for an ego vehicle, for example, as part of behavior planning for the ego vehicle. Another key aspect of the method according to the invention is that the geometric zones required for the analysis of the traffic scene are determined automatically, based on map information about influencing traffic flows along the ego vehicle's route.

[0034] For this purpose, the system 1 comprises a provision stage 11, which, in the exemplary embodiment described here, has access to digital map material, such as ASAM OpenDRIVE, Ianelet2, or OpenStreetMap, which is stored locally in the ego vehicle or can be retrieved from a cloud. Furthermore, map information 110 is made available to the provision stage 11 from additional data sources, such as from traffic monitoring using location-based cameras via a Car2X-connected cloud infrastructure. In addition, scene-specific information or the results of an environment perception 120 based thereon are supplied to the provision stage 11 from an input stage 10.

[0035] Provisioning level 11 evaluates all of this map information to determine all possible traffic flows or routes where conflicts between road users could arise. If necessary, traffic flows are also grouped together, for example if they can be assigned to a common infrastructure element such as an intersection. To determine the influencing routes, information about the existence, extent and course of roads is used in particular, with the geometries being represented in the form of polygons, for example. In addition to geographical map information, traffic rule information is also used in route determination, such as information on permitted directions of travel, right of way rules and permitted directions for turning. Provisioning level 11 also uses the map information available to it to determine movement corridors on the individual routes.

[0036] As already mentioned, system 1 has an input stage 10. This stage aggregates scene-specific information 100 and processes and evaluates it for the purpose of environmental perception. For this purpose, the sensors, sensor data processing, and evaluation of a perception component of the ego vehicle can advantageously be used.

[0037] In a processing stage 12 downstream of the input stage 10, which here comprises two function blocks 121 and 122, status information of the ego vehicle - function block 121 - and status information of at least one other participant in the traffic scene - function block 122 - are determined on the basis of the aggregated scene-specific information. This status information includes at least information on the respective position and the respective movement state of the ego vehicle and the other participant. The processing stage 12 with the two function blocks 121 and 122 is followed by an evaluation stage 13 with three function blocks 131, 132 and 133. This evaluation stage 13 also receives the map information from the provision stage 11, which is illustrated by the arrow 130. According to the invention, in the function block

[0038] 131 of the evaluation stage 13, using the map information and the status information of the ego vehicle, determines at least one possible route of the ego vehicle as the ego route. In the embodiment described here, the destination and / or a navigation specification 140 for the ego vehicle are also taken into account. In the function block

[0039] 132 of evaluation stage 13, at least one route of the additional participant is determined based on the map information and the status information of the additional participant, and possibly also the status information of the ego vehicle. It should be noted at this point that environmental perception information, such as turn signal signals or the illumination of brake lights, can also be used to determine the possible ego routes and routes of the additional participant. In any case, the route information generated by function blocks 131 and 132 is combined in function block 133 to determine those routes of the additional participant that influence a possible ego route of the ego vehicle. Such routes are referred to as conflict routes in the context of the present invention.

[0040] In an abstraction level 14 of system 1, at least one zone graph is then determined for the ego vehicle for each conflict route, again using the map information provided by provision level 11, and in particular the definitions of the movement corridors. This is illustrated by arrow 150. The basis for defining a zone graph here is the movement corridor of the ego route and the movement corridor of the respective conflict route. First, the overlap area between the movement corridor of the ego route and the movement corridor of the conflict route is determined and defined as the conflict zone of the zone graph. Based on this, the movement corridor of the ego route can then be further divided.If another movement corridor runs parallel to the movement corridor of the conflict route, but no other participant was detected on it, then the overlap area between this additional movement corridor and the movement corridor of the ego route is defined as a pre-conflict zone - in the direction of travel of the ego vehicle immediately in front of the conflict zone - or as a post-conflict zone - in the direction of travel of the ego vehicle immediately behind the conflict zone.

[0041] Furthermore, non-overlapping areas of the ego-route's movement corridor before and after the conflict zone can be defined as approach zones and exit zones, respectively. When defining the length of these zones, other variables are advantageously taken into account, such as the length of the ego-vehicle, the vehicle speed and the associated braking distance, safety distances, route-dependent maximum speeds, expected maneuvers of the ego-vehicle and other road users, corresponding behavior models, etc.Analogous to the definition of a first approach zone and a first exit zone on the ego route, a second approach zone and a second exit zone can also be defined as non-overlapping areas of the movement corridor of the other participant, whereby the second approach zone would then be located on the conflict route in front of the conflict zone, while the second exit zone would be located on the conflict route behind the conflict zone.

[0042] The traffic scene is analyzed based on at least one zone graph defined in this way. For this purpose, System 1 includes an analysis level 15.

[0043] The implementation described above in conjunction with Fig. 1 enables a fully automated determination of the zones of a traffic scene relevant to an ego vehicle. These zones, especially in conjunction with the SOCA method, are a key to argumentable, secure behavior planning.

[0044] Figures 2a to 2c each show an intersection area 20 of two two-lane roads 21 and 22, with the first road 21 running north to south, while the second road 22 runs west to east. The geometric extent of the individual lanes of the two roads 21 and 22, available as map information, each defines a movement corridor. Figures 2a to 2c show three different zone graphs generated using the method according to the invention.

[0045] Fig. 2a shows an ego route 3, which, as described above, was determined based on map information, status information of an ego vehicle and, if applicable, other road users, as well as navigation instructions. Ego route 3 runs in the right lane of the first road 21 from north to south and crosses both lanes of the second road 22. Furthermore, Fig. 2a shows a conflict route 4 of another vehicle. Conflict route 4 was also determined based on map information and status information. It runs in the right lane of the second road 22 from west to east and crosses both lanes of the first road 21.

[0046] To analyze the underlying traffic scene, a zone graph for the ego vehicle was defined based on ego route 3, namely on the basis of the movement corridor 30 on ego route 3 and the movement corridor 40 on conflict route 4. The zone graph here includes a conflict zone 33, a pre-conflict zone 32, as well as a first approach zone 31 and a first exit zone 35 on ego route 3. In addition, the zone graph also includes a second approach zone 41 and a second exit zone 45 for the other participant on conflict route 4.

[0047] Conflict zone 33 is formed by the overlap area between the two movement corridors 30 and 40 on ego route 3 and conflict route 4. Pre-conflict zone 32 lies in the direction of travel of the ego vehicle immediately in front of conflict zone 33 and is formed by the overlap area of ​​movement corridor 30 with the free lane of second road 22. The first approach zone 31 was defined as an overlap-free section of movement corridor 30 of the ego vehicle in front of pre-conflict zone 32, and the first exit zone 35 was defined as an overlap-free section of movement corridor 30 of the ego vehicle that lies behind conflict zone 33 in the direction of travel and directly adjacent to it. The second approach zone 41 and the second exit zone 45 for the other vehicle on conflict route 4 were also defined accordingly.

[0048] Considered here is the case of an ego vehicle approaching a four-way intersection while another vehicle is on the intersecting road and approaching the intersection from the right—from the ego vehicle's perspective. The ego vehicle first passes through approach zone 31 before the intersection, then through pre-conflict zone 32 in the intersection, until it reaches conflict zone 33 in the intersection. After passing through conflict zone 33, the ego vehicle then moves away from the intersection through exit zone 35. The other vehicle approaches conflict zone 33 via approach zone 41 and, after passing through conflict zone 33, moves away via exit zone 45.

[0049] In contrast to Fig. 2a, Fig. 2b considers the case where an ego vehicle is approaching a four-way intersection while another vehicle is on the intersecting road and is approaching the intersection area from the left—from the perspective of the ego vehicle. The ego vehicle first passes through an approach zone 31 before the intersection area and then directly reaches the conflict zone 33 in the intersection area. A pre-conflict zone is not provided here; instead, a post-conflict zone 34 is provided in the intersection area. The post-conflict zone 34 lies, in the direction of travel of the ego vehicle, immediately behind the conflict zone 33 and is formed by the overlap area of ​​the movement corridor 30 with the free lane of the second road 22. After passing through the post-conflict zone 34, the ego vehicle then moves away from the intersection area through an exit zone 35.

[0050] Finally, Fig. 2c considers the case where an ego vehicle approaches a four-way intersection in order to cross it in a straight line. Another vehicle approaches the intersection area - from the perspective of the ego vehicle - from the right in order to turn right onto the first street 21, specifically onto the movement corridor 30 of the ego route. The movement corridor 40 of the other vehicle is curved in accordance with the turning process. The ego vehicle first passes through an approach zone 31 before the intersection area and then a pre-conflict zone 32, which is defined here as the overlap area between the movement corridor 30 of the ego route and both lanes of the second street 22 and also extends across the intersection area to the junction of the movement corridor 40 of the conflict route with the movement corridor 30 of the ego route.Conflict zone 33 is defined here as the overlap area between movement corridor 30 of the ego route and movement corridor 40 of the conflict route in the intersection area, i.e., behind the intersection area in the direction of travel of the ego vehicle. Conflict zone 33 is adjoined by a post-conflict zone 34. The approach zone 41 of the other vehicle extends across the curved area of ​​movement corridor 40 to conflict zone 33.

[0051] Finally, it should be noted that the presented method can be used both during the development phase of a vehicle to annotate existing map material and online in the vehicle. When used in the vehicle, it is also conceivable to perform the determination of interfering traffic flows and / or the calculation of zones in a cloud system and transmit the calculated zones to the vehicle.

Images