Method and system for dynamically reducing the complexity of 3-D maps - Patents.com

Abstract

To provide a method and a system for simplifying the complexity of a three dimensional map.SOLUTION: The invention relates to a method for dynamically reducing complexity of a three-dimensional map in an own vehicle, the method comprising: a step (S1) of recording surrounding environment of the own vehicle by using at least one surrounding environment detection sensor; a step (S2) of supplying a three-dimensional map with height information; a step (S3) of estimating the location of the own vehicle in the three-dimensional map; and a step (S4) of generating a reduced three-dimensional map by reducing the supplied three-dimensional map by using a reduction algorithm, where the reduced three-dimensional map preferably comprises a road, where the road is currently associated with the own vehicle or, on the other hand, is associated with a traffic participant associated with the own vehicle.SELECTED DRAWING: Figure 1

Description

[Technical field]

[0001] The present invention relates to a method and system for dynamically reducing the complexity of a 3D map. [Background technology]

[0002] Many automated driving systems (AD systems) are not configured to function in environments where multiple elevation levels exist, for example, in multi-storey car park buildings or interchanges. While the concept of multiple levels exists, drivable paths between the levels are often not taken into account.

[0003] A common solution is to do a large map shift of the entire environment model from one level to the next. This can be problematic as it leads to a lack of consistency in the environment model. Also, there are environments where the different levels cannot be clearly defined. For example, there is a parking building with multiple levels that consists only of ramps.

[0004] Another solution is to directly transmit a multi-layer 3D map, and all downstream AD components need to process this complex 3D map and plan and execute actions using this complex 3D map with vertically overlapping roads, thus significantly increasing the complexity of all downstream AD components, which results in, for example, increased technology and computational costs.

[0005] Furthermore, Patent Document 1 discloses a method for displaying roads, which selects and displays roads to be used for navigation based on a multi-layered three-dimensional map. [Prior art documents] [Patent documents]

[0006] [Patent Document 1] Korean Patent No. 101051310 Summary of the Invention [Problem to be solved by the invention]

[0007] It is an object of the present invention to provide a method and system for simplifying the complexity of 3D maps. [Means for solving the problem]

[0008] This problem is solved by the independent claims 1 and 7. Further advantageous configurations and embodiments are the subject matter of the dependent claims.

[0009] The first consideration was that, based on the high complexity of the 3D maps, downstream systems for assisted and / or automated driving need to have very high computational power to be able to process large amounts of data. Also, due to different hierarchies with different heights, mis-interventions may occur if a road or a certain object or feature is classified as relevant but not related to the height level at which the ego-vehicle is traveling.

[0010] Therefore, according to the invention, there is provided a method for dynamically reducing the complexity of a 3D map in an ego vehicle, comprising the steps of: - recording the surroundings of the vehicle using at least one surroundings detection sensor; - providing a three-dimensional map having different height levels; - estimating the position of the vehicle in a three-dimensional map; - generating a reduced 3D map by reducing the created 3D map using a reduction algorithm, the reduced 3D map preferably comprising roads, which are currently associated with the ego vehicle or, on the other hand, are associated with other traffic participants associated with the ego vehicle.

[0011] The surrounding environment detection sensor may be, for example, a camera, a radar sensor, a lidar sensor or an ultrasonic sensor. It is also conceivable to use multiple similar and / or different sensors to record the surrounding environment.

[0012] The 3D map may for example be provided via a database, in which the 3D map with the corresponding height information is stored and can be retrieved by the vehicle. It is also conceivable to fuse the 3D map based on the map information with sensor data of at least one surrounding environment detection sensor, from which a surrounding environment representation can then be created and evaluated.

[0013] Other environment representations are generated depending on the environment detection sensor used: when using a camera, either a monocular or a stereo camera, the environment representation is an image or an image sequence of the environment, when using a radar sensor, the environment representation is for example an object list.

[0014] Map data may include, for example, a height profile of the surrounding environment as well as road Transition and number of vehicles. These map data may for example be present in a database internal or external to the vehicle and may be provided via this database. It is also conceivable to use navigation data. If the database is arranged external to the vehicle, the data is transmitted by means of a vehicle-to-X communication device (V2X communication device), in which case the vehicle is equipped with a corresponding V2X communication device.

[0015] In the context of the present invention, dynamically reducing means that the relevant roads or road sections change continuously with the movement of the vehicle, so that at each detection cycle of the sensor and / or at each position estimation cycle of the ego-vehicle in the 3D map, the relevance of the roads or road sections is checked and the 3D map is adapted, for example, to the changed position of the ego-vehicle, in this way the 3D map is dynamically reduced.

[0016] The reduction algorithm handles a 3D map that contains potentially overlapping roads with different height information, and outputs a reduced road map that does not contain overlapping roads based on different height information. To ensure minimal preview or review, the reduced map is built road by road, starting from the ego vehicle's current road, and is expanded in all directions until an end point is reached. Such an end point is reached when no more roads or road sections can be added, either because the initial 3D map ends or one or more roads (or road sections) cannot be added, or because one or more roads would result in overlaps with different height information. The end point is also reached when all roads that follow from this point on have already been added to the reduced road model. To control the expansion, the algorithm assigns different priorities to the roads (sections), so that roads with relevance are prioritized.

[0017] Particularly preferably, the reduction of the three-dimensional map is performed such that the three-dimensional map does not contain overlapping roads or road sections at different heights.

[0018] In a preferred configuration, the reduction algorithm is a graph search algorithm. Particularly preferably, the graph search algorithm may be a Dijkstra algorithm or a breadth search. Dijkstra's algorithm generally always follows the edge with the highest priority in the graph starting from the current source node. In a breadth search or breadth-first search (BFS), starting from the source, first all nodes that are directly reachable from the source via edges are examined. Then, starting from these nodes, further nodes that are directly reachable from these nodes are examined.

[0019] In a further preferred embodiment, the relevance of a road is determined based on its proximity to the ego-vehicle, its detection range, its level relative to the ego-vehicle and / or the road's drivability. In general, the relevance of a road or road section decreases the further the road or road section is from the ego-vehicle in relation to the travel distance. Nearby roads located directly around the ego-vehicle at the same level have high relevance and therefore have high priority for addition to the reduced map. Also, roads that are characterized as drivable and / or that are located on the planned route have high relevance. This may be, for example, roads, lanes or road sections with a traffic direction in the driving direction of the ego-vehicle.

[0020] Furthermore, the roads present in the 3D map are preferably divided into road sections, which is advantageous because in this way future road changes can be better accommodated and the 3D map can be reduced with high accuracy.

[0021] Particularly preferably, the road sections are divided into two groups according to their respective levels and / or Transition If the road ahead is, for example, an uphill Transition In the case where the road has a slope, the road may be divided into a section before the uphill slope and a section after the uphill slope. Also, the sections may be introduced at, for example, intersections or branches of the road.

[0022] Furthermore, according to the invention, a system for reducing the complexity of a 3D map in an ego-vehicle is proposed, which comprises at least one surrounding environment detection sensor for recording the surrounding environment of the ego-vehicle, an evaluation unit for evaluating the recording of the surrounding environment detection sensor, a device for providing a 3D map, and a calculation unit for applying a reduction algorithm for estimating the location of the ego-vehicle and generating a reduced 3D map. The evaluation unit may be configured as a separate element or as a component of the surrounding environment detection sensor. The calculation unit is preferably an ECU (electronic control unit) or an ADCU (autonomous driving control unit). The device for providing the 3D map may be, for example, a database inside or outside the vehicle. Alternatively, the device may be configured to fuse the surrounding environment representation of the sensor data of the surrounding environment detection sensor with map data retrieved or provided via a database. In this case, the device may be configured as a separate calculation unit. It is also conceivable to perform the fusion in an already provided calculation unit, for example an ADCU.

[0023] Further advantageous configurations are the subject of the drawings. [Brief description of the drawings]

[0024] [Figure 1] FIG. 1 shows a schematic flow diagram of a method according to an embodiment of the present invention. [Diagram 2] FIG. 2 illustrates a schematic diagram of a system according to an embodiment of the present invention. [Diagram 3] FIG. 3 shows a schematic diagram of a three-dimensional map according to the prior art. [Figure 4] FIG. 4 shows a schematic diagram of a three-dimensional map according to the prior art. [Diagram 5] FIG. 5 illustrates a schematic representation of a three-dimensional map according to an embodiment of the present invention. [Figure 6] FIG. 6 illustrates a schematic representation of a three-dimensional map according to an embodiment of the present invention. [Figure 7] FIG. 7 illustrates a schematic representation of a three-dimensional map according to an embodiment of the present invention. DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0025] Fig. 1 shows a schematic flow diagram of a method according to an embodiment of the invention. In step S1, the surroundings of the ego-vehicle are recorded by means of at least one surroundings detection sensor. In a further step S2, a 3D map with height information is provided. Then, in step S3, the ego-vehicle E is localized in the 3D map. In step S4, a reduced 3D map is generated by reducing the provided 3D map using a reduction algorithm, the reduced 3D map preferably comprising roads which are currently associated to the ego-vehicle E or further traffic participants which in turn are associated to the ego-vehicle.

[0026] FIG. 2 shows a schematic diagram of a system according to an embodiment of the invention. A system 1 for reducing the complexity of a 3D map in a vehicle E comprises at least one surrounding environment detection sensor 2 for recording the surrounding environment of the vehicle E, an evaluation unit 3 for evaluating the recording of the surrounding environment detection sensor 2, a device 5 for providing a 3D map, and a calculation unit for applying a reduction algorithm for estimating the location of the vehicle E and generating a reduced 3D map. Here, these elements are connected to each other via a data connection D. This data connection D may be configured in a wired or wireless manner. Here, the database 5 may be provided either inside the vehicle or outside the vehicle. The evaluation unit 3 may be configured as a separate element or as a component of the surrounding environment detection sensor 2. The calculation unit 4 is preferably an ECU or an ADCU. In this preferred configuration, the system comprises a database as device 5 for providing the 3D map. It is also conceivable to create the 3D map using a fusion of sensor data and map data.

[0027] FIG. 3 shows a schematic diagram of a three-dimensional map according to the prior art. In this diagram, the road of the three-dimensional map is divided into different road sections R1-R4. Also, in this diagram, a pedestrian P is detected and input into the three-dimensional map. Furthermore, the trajectory T of the pedestrian P is predicted. Here, the road sections R1 and R4 are located at different height levels. Based on this diagram and the current state of the art, the system assumes that the pedestrian P is present at a level that is higher than the host vehicle E because, in the bird's-eye view of the three-dimensional map, even if the pedestrian is normally detected at the same level as the host vehicle E, the pedestrian also crosses the road section R4 that is higher than the host vehicle's localization R1. Thus, in this case, a misinterpretation occurs, which may result in a wrong or no reaction of the AD system.

[0028] Figure 4 shows a schematic representation of a 3D map according to the prior art. It corresponds to the representation of Figure 3. Here, R2 is a ramp leading from the level of road section R1 to road sections R3, R4 at higher levels. This is a representation of a 3D map transmitted according to the prior art to a downstream AD or ADAS system.

[0029] Fig. 5 shows a schematic diagram of a 3D map according to an embodiment of the invention. In this figure, the ego-vehicle E is located on road section R1. By pruning the 3D map, only road sections R1, R2 and R3 are included in the 3D map at this point. Road section R4 has been removed from the 3D map by pruning, since R4 in this case intersects with R1, but at a higher level. Therefore, in this situation, R4 is not relevant for the ego-vehicle E.

[0030] Fig. 6 shows a schematic diagram of a 3D map according to an embodiment of the present invention. In this figure, the ego-vehicle E has moved from road section R1 to road section R2. R1 is no longer output in the reduced 3D map because it is not relevant to the vehicle. R1 and R4 obviously overlap at different levels, as in Fig. 2 or 3. In this regard, R4 is now output together, because R4 and R3 are located at the same level and are potentially relevant for the ego-vehicle's further travel.

[0031] In Fig. 7 a schematic diagram of a 3D map according to an embodiment of the invention is again shown, where the ego-vehicle E is now located on road sections R3, R4, and in the reduced 3D map R1 has been discarded again since R4 has been output.

[0032] In general, Figures 5 to 7 illustrate a dynamic reduction of the 3D map described at the beginning, since in each figure the host vehicle E moves further and accordingly different road sections R1 to R4 become associated with the host vehicle E and the 3D map is reduced accordingly. This application relates to the invention described in the claims, but also includes the following as other aspects. 1. A method for dynamically reducing the complexity of a 3D map in an ego-vehicle (E), comprising: - a step (S1) of recording the surrounding environment of the vehicle (E) using at least one surrounding environment detection sensor (2); - providing a three-dimensional map with height information (S2); - estimating (S3) the position of the vehicle (E) in the three-dimensional map; - generating a reduced 3D map by reducing the supplied 3D map using a reduction algorithm, said reduced 3D map preferably comprising roads, said roads currently associated with said ego vehicle (E) or, alternatively, associated with a traffic participant associated with said ego vehicle (E). 2. 2. The method according to claim 1, characterized in that the reduction of the 3D map is performed such that the 3D map does not contain overlapping roads at different heights. 3. 2. The method according to claim 1, wherein the reduction algorithm is based on a graph search algorithm. 4. The method according to claim 1, wherein the relevance of a road is determined based on the proximity to the host vehicle (E), the detection range of the host vehicle (E), the height relative to the host vehicle (E), the drivability of the road and / or the planned route. 5. 5. The method according to any one of 1 to 4 above, characterized in that roads present in the three-dimensional map are divided into road sections (R1 to R4). 6. 5. The method according to claim 4, characterized in that the road sections (R1 to R4) are divided based on their respective height information and / or transitions. 7. A system (1) for dynamically reducing the complexity of a three-dimensional map in a host vehicle (E), comprising at least one surrounding environment detection sensor (2) for recording the surrounding environment of the host vehicle (E), an evaluation unit (3) for evaluating the recording of the surrounding environment detection sensor (2), a device (5) for providing a three-dimensional map with height information, and a calculation unit (4) for applying a reduction algorithm to estimate the position of the host vehicle (E) and generate a reduced three-dimensional map. [Explanation of symbols]

[0033] 1 System 2. Surrounding environment detection sensor 3 Evaluation Unit 4 Computing Unit 5 Equipment D Data Connection E Vehicle P Pedestrian R1~R4 Road Section S1~S7 Method steps T Pedestrian trajectory

Claims

1. A method for removing overlapping roads at different heights from a three-dimensional map of a vehicle (E) as the vehicle (E) moves, comprising: - a step (S1) of recording the surrounding environment of the vehicle (E) using at least one surrounding environment detection sensor (2); - providing a 3D map with height information (S2); - a step (S3) of estimating the position of said vehicle (E) in said three-dimensional map; - generating a reduced 3D map by using a reduction algorithm to reduce overlapping roads at different heights from the provided 3D map, the reduced 3D map including roads on which the ego-vehicle (E) is traveling or on which traffic participants are located around the ego-vehicle (E).

2. 2. The method of claim 1, wherein the reduction of the three-dimensional map is performed such that the three-dimensional map does not contain overlapping roads at different heights.

3. The method of claim 1 , wherein the reduction algorithm is based on a graph search algorithm.

4. The method according to claim 1, wherein the relevance of a road is determined based on its proximity to the host vehicle (E), its detection range to the host vehicle (E), its height relative to the host vehicle (E), its drivability and / or the planned route.

5. Method according to any one of claims 1 to 4, characterized in that the roads present in the 3D map are divided into road sections (R1 to R4).

6. Method according to claim 4, characterised in that the road sections (R1-R4) are divided based on their respective height information and / or transitions.

7. A system (1) for reducing overlapping roads at different heights from a three-dimensional map of a host vehicle (E) as the host vehicle (E) moves, the system (1) comprising at least one surrounding environment detection sensor (2) for recording the surrounding environment of the host vehicle (E), an evaluation unit (3) for evaluating the record of the surrounding environment detection sensor (2), a device (5) for supplying a three-dimensional map having height information, and a calculation unit (4) for estimating the position of the host vehicle (E) and applying a reduction algorithm to generate a reduced three-dimensional map.

Images