Method and system for generating a localization map containing at least one unique feature combination

Abstract

The invention relates to a method for generating a localization map containing at least one unique feature combination for operating an automated and / or assisted vehicle system (2) of a vehicle (3a, 3b), wherein fleet data are collected by driving fleet vehicles (3b) of a vehicle fleet along a traffic route section assigned to the localization map, the localization map is generated from the collected fleet data, and at least one unique feature combination is determined from the collected fleet data, wherein the result of a comparison of the unique feature combination is determined either with fleet data for each individual driving of the traffic route section or with a modeling of a sensor detection area of ​​a non-fleet vehicle (3a) along the traffic route section,whether the unique feature combination is detectable by an environmental sensor system (4) of the fleet vehicles (3b) or the vehicle (3a) or not, and the unique feature combination is stored in the provisional localization map if the unique feature combination is detectable by the environmental sensor system (4).

Description

[0001] The invention relates to a method for generating a localization map containing at least one unique combination of features for operation of an automated and / or assisted vehicle system of a vehicle, wherein fleet data are collected by driving fleet vehicles of a vehicle fleet along a traffic route section assigned to the localization map, the localization map is generated from the collected fleet data, and at least one unique combination of features is determined from the collected fleet data.The invention also relates to a system for generating a localization map containing at least one unique combination of features for operating an automated and / or assisted vehicle system of a vehicle, comprising a plurality of fleet vehicles, each of which has at least one environmental sensor system, at least one central data processing unit which can be connected to the fleet vehicles for collecting fleet data by driving along a traffic route section assigned to the localization map with the fleet vehicles and which is configured to generate the localization map from the collected fleet data and to determine at least one unique combination of features from the collected fleet data. State of the art

[0002] Detecting a vehicle's surroundings using an on-board perception system based on environmental sensors, which may include, for example, a radar sensor and / or a video camera, is a fundamental component of modern automated vehicle systems (AD systems) and assisted vehicle systems (DA systems). The environment models used in this context are often supplemented by additional map information, such as lane layouts. The map data used for this purpose, for example in the form of map products such as planning maps, behavior maps, and / or localization maps, can be made available to the respective vehicle system via a map service.

[0003] Determining a vehicle's position relative to the map in use is a basis for using map data. Feature-based localization, particularly self-localization, of the vehicle enables an estimation of a map-relative vehicle position and orientation. This estimation is based on a localization map and on-board radar feature measurements. Ambiguities in a localization map, in particular, pose a challenge for current feature-based localization approaches.

[0004] Crowdsourcing of fleet data for the purpose of map creation is already known. For this purpose, fleet data for traffic route areas to be mapped and / or monitored is requested and transmitted without any targeted selection. In addition, a variety of algorithms for creating, expanding, and updating localization maps are known. Furthermore, approaches are known that identify unique local feature combinations in localization maps.

[0005] DE 10 2018 008 988 A1 relates to a method for updating map data used for the navigation of an autonomously operated motor vehicle. According to the method, a route section in the surroundings of the motor vehicle is detected by an environmental sensor system of the motor vehicle, in which the environmental sensor system generates detection data that characterizes the route section. In addition, a deviation data set characterizing a difference between the detection data and the map data with regard to the route section is determined if the detection data differ from the map data. Furthermore, the deviation data set is transmitted to an external computing device. Furthermore, the map data is updated based on the deviation data set using the external computing device. Disclosure of the invention

[0006] The subject matter of the invention is a method for generating a localization map containing at least one unique feature combination for operation of an automated and / or assisted vehicle system of a vehicle. The method can be used to improve the reliability of feature-based vehicle localization systems by processing unique feature combinations in a new way and storing them in a localization map. In particular, the inventive comparison of the respective unique feature combination either with the fleet data for each individual travel along the traffic route section or with the modeling of the sensor detection range of the vehicle not belonging to the fleet along the traffic route section ensures that only those unique feature combinations are selected and stored in the localization map that are detected by the environmental sensors of the fleet vehicle ora vehicle not belonging to the fleet is also detectable. Only such detectable, unique feature combinations can be used for localization. Thus, vehicle self-localization can then be performed using reliably detectable, unique feature combinations, which significantly improves the reliability and accuracy of this self-localization. In contrast, conventional self-localization does not take into account the detectability of the respective unique feature combinations by the vehicle's environmental sensors.

[0007] According to the method according to the invention, the respective unique feature combination can be compared with the fleet data for each individual travel along the traffic route section, so that for each travel, it can be determined whether or not the unique feature combination can be detected by the environmental sensors of the fleet vehicle. In this alternative, it is assumed that the vehicles providing the fleet data (fleet vehicles) are also the vehicles using the localization map with the at least one unique feature combination. Alternatively, the vehicles using the localization map with the at least one unique feature combination can differ from the vehicles providing the fleet data, in particular with regard to the sensor detection range of their respective environmental sensors.In this case, the at least one unique feature combination is then compared with the model of the sensor detection range of the non-fleet vehicle along the traffic route section, for example, along a lane of the traffic route section. For example, a simple model of the non-fleet vehicle can be used. This vehicle can be a vehicle from a vehicle fleet that is differently configured with regard to the respective sensor detection range, i.e., a vehicle fleet whose vehicles differ from the fleet vehicles that provide the fleet data with regard to the respective sensor detection range.Then, the result of the comparison of the at least one unique feature combination with the modeling of the sensor detection area of ​​the non-fleet vehicle along the traffic route section can be used to generate a localization map that is valid for all vehicles of the otherwise configured vehicle fleet.

[0008] According to the method according to the invention, the fleet data is initially collected by driving fleet vehicles of a vehicle fleet along a traffic route section assigned to the localization map. The respective fleet vehicle drives along the traffic route section in order to capture data on the surroundings of the fleet vehicle using the environmental sensors. The environmental sensors can, for example, have at least one camera and / or at least one radar sensor and / or at least one Li-DAR sensor. In particular, the fleet vehicles can be designed identically with regard to their respective environmental sensors. Furthermore, the fleet vehicles can also be otherwise essentially identical. The collected fleet data can, for example, be transmitted to a cloud system, i.e., to a fleet cloud.From the fleet cloud, the map data can, for example, be transferred to a fleet data mapping cloud and processed there, in particular in the conventional way, to create the localization map.

[0009] At least one unique feature combination can then be determined from the fleet data, in particular the fleet data assigned to the localization map or the traffic route section. This is described, for example, in the article "Comparison of 3D interest point detectors and descriptors for point cloud fusion," R. Hänsch et al., ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2014, volume II-3, pages 57-64 (https: / / isprs-annals.copernicus.org / articles / II-3 / 57 / 2014 / ).

[0010] The localization map that can be generated using the method according to the invention and contains at least one unique combination of features can be used in particular for operating an automated vehicle system (AD system) and / or assisting vehicle system (DA system) of a vehicle.

[0011] According to an advantageous embodiment, the comparison results are used to determine from which lane(s) of the traffic route section the unique feature combination can be detected by the environmental sensors. Information from a lane data layer of the generated localization map can be used for this purpose. If the individual travels of the traffic route section were already assigned to individual lanes to generate this localization map, this information is already available and can be used to determine from which lane(s) of the traffic route section the unique feature combination can be detected by the environmental sensors.

[0012] According to a further advantageous embodiment, at least one quality value is determined for the unique feature combination using a detection rate specific to the unique feature combination determined from the result of the comparison and / or a visibility of the unique feature combination from at least two different areas of the traffic route section determined from the result of the comparison and / or a timeliness of the fleet data, and the quality value is stored in the localization map. The detection rate can indicate, for example, in percent or as a ratio, the number of times the fleet vehicles travel the traffic route section during which the unique feature combination was actually detected. The visibility of the unique feature combination can be given, for example, by the number of lanes from which the unique feature combination can be detected.The quality value is preferably continuous.

[0013] A further subject matter of the invention is a system for generating a localization map containing at least one unique combination of features for operation of an automated and / or assisted vehicle system of a vehicle, comprising a plurality of fleet vehicles, each of which has at least one environmental sensor system, at least one central data processing unit which can be connected to the fleet vehicles for collecting fleet data by driving along a traffic route section assigned to the localization map with the fleet vehicles and which is configured to generate the localization map from the collected fleet data and to determine at least one unique combination of features from the collected fleet data.The central data processing unit is configured to determine, based on the result of a comparison of the unique feature combination either with fleet data for each individual travel along the traffic route section or with a model of a sensor detection area of ​​a vehicle not belonging to the fleet along the traffic route section, whether or not the unique feature combination is detectable by the environmental sensors of the fleet vehicles or the vehicle, and to store the unique feature combination in the provisional localization map if the unique feature combination is detectable by the environmental sensors.

[0014] The system offers the advantages mentioned above with regard to the method. Advantageous embodiments of the method can correspond to advantageous embodiments of the system. The central data processing unit can, for example, be cloud-based.

[0015] According to an advantageous embodiment, the central data processing unit is configured to determine, based on the result of the comparison, from which lane(s) of the traffic route section the unique feature combination can be detected by the environmental sensors. This embodiment provides the advantages mentioned above with reference to the corresponding embodiment of the method.

[0016] According to a further advantageous embodiment, the central data processing unit is configured to determine at least one quality value for the unique feature combination using a detection rate specific to the unique feature combination determined from the result of the comparison and / or a visibility of the unique feature combination from at least two different areas of the traffic route section determined from the result of the comparison and / or a timeliness of the fleet data, and to store the quality value in the localization map. This embodiment provides the advantages mentioned above with reference to the corresponding embodiment of the method.

[0017] According to a further advantageous embodiment, the central data processing unit is configured to generate a separate acquisition query for each unique feature combination and send it to the fleet vehicles in order to preferentially or exclusively acquire the respective unique feature combination while driving along the traffic route section using the environmental sensors of the respective fleet vehicle. The acquisition queries, also called job requests, can be used to specifically request the acquisition of unique feature combinations. This ensures a high level of timeliness and accuracy of the unique feature combinations, particularly through the significance of the feature statistics, while simultaneously keeping the data rate low. This enables efficient crowdsourcing of the unique feature combinations.

[0018] According to a further advantageous embodiment, the central data processing unit is configured to determine a route and trajectory plan for the operation of the automated and / or assisted vehicle system, with which unique feature combinations can be detected. This ensures that the respective unique feature combination can be reliably detected by the environmental sensors, in particular enabling precise self-localization.

[0019] According to a further advantageous embodiment, the automated and / or assisting vehicle system of at least one fleet vehicle or at least one non-fleet vehicle of the system is configured to carry out self-localization using the localization map containing the at least one unique feature combination, and in doing so to only consider those unique feature combinations whose quality value stored in the localization map lies above a predetermined lower limit. The vehicle system can therefore use the quality value to ensure a minimum quality of unique feature combinations for a given application, for example the initialization of self-localization. The quality value can therefore be used during activation orThe initialization of self-localization can be taken into account, so that, for example, in applications with high error rate requirements, only high-quality, unique feature combinations can be used. This enables a further increase in the efficiency and reliability of self-localization. This allows the system to be used as part of a safety concept for feature-based vehicle self-localization.

[0020] In the following, the invention is explained by way of example with reference to the attached figures using preferred embodiments, wherein the features explained below can represent an advantageous and / or further developing aspect of the invention both individually and in different combinations of at least two of these features with one another. Short description of the characters

[0021] It shows: Fig. 1 a schematic representation of an embodiment of a system according to the invention; and Fig. 2 a flowchart of an embodiment of a method according to the invention. Detailed description of the characters

[0022] In the figures, identical or functionally identical components are provided with the same reference numerals. A repeated description of such components may be omitted in detail to avoid unnecessary repetition.

[0023] Fig. 1 shows a schematic representation of an embodiment of a system 1 according to the invention for generating a localization map containing at least one unique feature combination for operation of an automated and / or assistive vehicle system 2 of a vehicle 3a or 3b.

[0024] The system 1 comprises a plurality of fleet vehicles 3b, each of which has at least one environmental sensor system 4. The system 1 also comprises a central data processing unit 5, which can be connected to the fleet vehicles 3b for collecting fleet data by driving along a traffic route section assigned to the localization map. The central data processing unit 5 is configured to generate the localization map from the collected fleet data and to determine at least one unique feature combination from the collected fleet data.

[0025] The central data processing unit 5 is also configured to determine, based on a result of a comparison of the unique feature combination either with fleet data for each individual travel along the traffic route section or with a modeling of a sensor detection area of ​​a non-fleet vehicle 3a along the traffic route section, whether the unique feature combination is detectable by the environmental sensor system 4 of the fleet vehicles 3b or of the vehicle 3a or not, and to store the unique feature combination in the provisional localization map if the unique feature combination is detectable by the environmental sensor system 4.

[0026] Furthermore, the central data processing unit 5 is configured to determine, on the basis of the result of the comparison, from which lane(s) of the traffic route section the unique feature combination can be detected by the environmental sensor system 4.

[0027] In addition, the central data processing unit 5 is configured to determine at least one quality value for the unique feature combination using a detection rate specific to the unique feature combination determined from the result of the comparison and / or a visibility of the unique feature combination from at least two different areas of the traffic route section determined from the result of the comparison and / or an up-to-dateness of the fleet data and to store the quality value in the localization map.The automated and / or assisting vehicle system 2 of at least one fleet vehicle 3b or at least one non-fleet vehicle 3a of the system 1 is configured to carry out self-localization using the localization map containing the at least one unique feature combination and to take into account only those unique feature combinations whose quality value stored in the localization map is above a predetermined lower limit.

[0028] In addition, the central data processing unit 5 is configured to generate a separate detection query for each unique feature combination and to send it to the fleet vehicles 3b in order to detect preferentially or exclusively the respective unique feature combination during travel along the traffic route section by means of the environmental sensors 4 of the respective fleet vehicle 3b.

[0029] Furthermore, the central data processing unit 5 can be configured to determine a route and trajectory planning for the operation of the automated and / or assisting vehicle system 2, with which unique feature combinations can be detected.

[0030] Fig. 2 shows a flowchart of an embodiment of a method according to the invention for generating a localization map containing at least one unique feature combination for operation of an automated and / or assistive vehicle system 2 of a vehicle (not shown).

[0031] In method step 100, fleet vehicles of a vehicle fleet drive along a traffic route section assigned to the localization map, during which the traffic route section is recorded using the environmental sensors of the respective fleet vehicle. As a result, fleet data assigned to the traffic route section is generated in method step 100. The fleet data generated in method step 100 is collected centrally in method step 101.

[0032] In method step 102, the localization map is generated from the collected fleet data and stored in method step 103. In method step 104, at least one unique feature combination is determined from the collected fleet data.

[0033] In method step 105, based on a result of a comparison of the unique feature combination either with collected fleet data for each individual travel of the traffic route section or with a modeling of a sensor detection area of ​​a non-fleet vehicle along the traffic route section, it is determined whether or not the unique feature combination is detectable by an environmental sensor system of the fleet vehicles or the vehicle, wherein the unique feature combination is stored in the provisional localization map if the unique feature combination is detectable by the environmental sensor system.

[0034] Furthermore, in method step 105, based on the result of the comparison, it can be determined from which lane(s) of the traffic route section the unique feature combination can be detected by the environmental sensors.

[0035] Furthermore, in method step 105, at least one quality value is determined for the unique feature combination using a detection rate specific to the unique feature combination determined from the result of the comparison and / or a visibility of the unique feature combination from at least two different areas of the traffic route section determined from the result of the comparison and / or a timeliness of the fleet data. The quality value is stored in the localization map in method step 103. QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] DE 10 2018 008 988 A1

[0005] Zitierte Nicht-Patentliteratur

[0000] Comparison of 3D interest point detectors and descriptors for point cloud fusion“, R. Hänsch et al., ISPRS Annals of the Photogrammetry, Remote Sensing and Spatial Information Sciences, 2014, volume II-3, pages 57-64

[0009]

Claims

[1] Method for generating a localization map containing at least one unique combination of features for operation of an automated and / or assisted vehicle system (2) of a vehicle (3a, 3b), wherein fleet data are collected by driving fleet vehicles (3b) of a vehicle fleet along a traffic route section assigned to the localization map, the localization map is generated from the collected fleet data, and at least one unique combination of features is determined from the collected fleet data, characterized bythat on the basis of a result of a comparison of the unique feature combination either with fleet data for each individual travel along the traffic route section or with a model of a sensor detection area of ​​a vehicle (3a) not belonging to the fleet along the traffic route section, it is determined whether the unique feature combination can be detected by an environmental sensor system (4) of the fleet vehicles (3b) or of the vehicle (3a) or not, and the unique feature combination is stored in the provisional localization map if the unique feature combination can be detected by the environmental sensor system (4). [2] Method according to claim 1, characterized by that, on the basis of the result of the comparison, it is determined from which lane(s) of the traffic route section the unique feature combination can be detected by the environmental sensor system (4). [3] Method according to claim 1 or 2, characterized bythat at least one quality value is determined for the unique feature combination using a detection rate specific to the unique feature combination determined from the result of the comparison and / or a visibility of the unique feature combination from at least two different areas of the traffic route section determined from the result of the comparison and / or an up-to-dateness of the fleet data, and the quality value is stored in the localization map. [4] System (1) for generating a localization map containing at least one unique combination of features for operation of an automated and / or assisted vehicle system (2) of a vehicle (3a, 3b), comprising a plurality of fleet vehicles (3b), each having at least one environmental sensor system (4), at least one central data processing unit (5) which can be connected to the fleet vehicles (3b) for collecting fleet data by driving along a traffic route section associated with the localization map, and which is configured to generate the localization map from the collected fleet data and to determine at least one unique combination of features from the collected fleet data, characterized byin that the central data processing unit (5) is configured to determine, on the basis of a result of a comparison of the unique feature combination either with fleet data for each individual travel along the traffic route section or with a model of a sensor detection area of ​​a vehicle (3a) not belonging to the fleet along the traffic route section, whether or not the unique feature combination can be detected by the environmental sensor system (4) of the fleet vehicles (3b) or of the vehicle (3a), and to store the unique feature combination in the provisional localization map if the unique feature combination can be detected by the environmental sensor system (4). [5] System (1) according to claim 4, characterized bythat the central data processing unit (5) is configured to determine, on the basis of the result of the comparison, from which lane(s) of the traffic route section the unique feature combination can be detected by the environmental sensor system (4). [6] System (1) according to claim 4 or 5, characterized by that the central data processing unit (5) is configured to determine at least one quality value for the unique feature combination using a detection rate specific to the unique feature combination determined from the result of the comparison and / or a visibility of the unique feature combination from at least two different areas of the traffic route section determined from the result of the comparison and / or an up-to-dateness of the fleet data and to store the quality value in the localization map. [7] System (1) according to one of claims 4 to 6, characterized bythat the central data processing unit (5) is configured to generate a separate detection query for each unique feature combination and to send it to the fleet vehicles (3b) in order to detect preferably or exclusively the respective unique feature combination during travel along the traffic route section by means of the environmental sensors (4) of the respective fleet vehicle (3b). [8] System (1) according to one of claims 4 to 7, characterized by that the central data processing unit (5) is configured to determine a route and trajectory planning for the operation of the automated and / or assisting vehicle system (2), with which unique feature combinations can be detected. [9] System (1) according to one of claims 6 to 8, characterized bythat the automated and / or assisting vehicle system (2) of at least one fleet vehicle (3b) or of at least one non-fleet vehicle (3a) of the system (1) is set up to carry out self-localization using the localization map containing the at least one unique feature combination and to take into account only those unique feature combinations whose quality value stored in the localization map lies above a predetermined lower limit value.

Images