Method for constructing a positioning layer of a digital positioning map for automated driving

Abstract

A method for constructing a positioning layer (12) of a digital positioning map for automated driving, the method comprising the following steps: providing the positioning layer (12) for a defined area; providing a planning layer (11) for the area; and extracting alignment features (1a...1n) from the positioning layer (12), the alignment features being configured for alignment with the planning layer (11), wherein the alignment features (1a...1n) are extracted from the positioning layer (12) in such a way that in the alignment of the planning layer (11) with the positioning layer (12), an unacceptable deformation of the positioning layer (12) can be identified.

Description

Technical Field

[0001] The present invention relates to a method for constructing a positioning layer of a digital positioning map for automated driving. The present invention also relates to a device for constructing a positioning layer of a digital positioning map for automated driving. The present invention also relates to a computer program product. Background Art

[0002] Map creation techniques are known and used to create a map of the surrounding environment from sensor observations of a mapping vehicle from a fleet of mapping vehicles. Such surrounding environment maps are known to have applications in, for example, robotics, logistics, automotive, aerospace, consumer products, etc.

[0003] Very high expectations and requirements are generally placed on the technologies that enable automated driving. One of the requirements is the availability of a detailed, accurate, up-to-date and complete digital map of the surrounding environment ("digital positioning map") that is sufficiently accurate to ensure robust, accurate and reliable positioning, planning and navigation of AD (English: automatic driving) vehicles or ADAS (English: advanced driver assistance system) vehicles.

[0004] Today, automated vehicles rely on map-based information that supports them in route planning, object recognition and situation assessment. The maps are usually denoted as HD maps as opposed to SD maps used for conventional navigation tasks.

[0005] The following three layers are usually distinguished in such high-resolution digital positioning maps:

[0006] a. The (lowest) planning layer, which contains, for example, precise lane geometry and connectivity information. This planning layer is primarily used to keep the automated vehicle within the lane and to support the automated vehicle in maneuver planning.

[0007] b. The (middle) positioning layer, which contains objects ("landmarks") detected by different types of sensors (e.g. radar sensors, lidar sensors, cameras, etc.). The latest information detected by the corresponding sensors is compared with the positioning layer in order to determine the vehicle's position in the digital positioning map or in the surrounding environment.

[0008] c. (Topmost) dynamic layer, which contains dynamically changing information, such as lane status, traffic conditions, weather, parking conditions, etc. This layer is mainly used for comfort functions in the digital positioning map.

[0009] In addition to the increased accuracy of digital localization maps for automated driving compared to conventional localization maps (eg for navigation systems), digital localization maps for automated driving should also be highly up-to-date in order to provide a high localization accuracy in time and space.

[0010] When using a positioning layer, feature-based positioning systems can make it possible to locate a mobile system (e.g., an automated vehicle) that senses the surroundings, in particular by decoding location coordinates from observable landmarks. In such systems, location-specific landmark elements (e.g., landmarks in the form of traffic signs, lanes, lane markings) are identified in the surroundings. In this case, the visible angle of the mobile device relative to the landmark element is measured, making it possible to determine the own position relative to the landmark.

[0011] WO 2017 / 174229 A1 discloses a method for determining the posture of a vehicle driven at least partially automatically in an environment by means of landmarks. In this case, the vehicle is moved in the environment, thereby generating a sequence of positioning scenarios, and landmark data are digitally processed at least by a vehicle control system to determine the posture of the vehicle. It is proposed to increase or decrease the amount of the landmark data as required depending on the positioning scenario.

[0012] DE 10 2011 119 762 A1 discloses a system and a method for determining the position of a motor vehicle, wherein a system for determining the position of a motor vehicle and a corresponding method are proposed. The system comprises a digital map in which data on location-specific features are recorded in a localized manner, at least one surroundings recognition device for detecting location-specific features in the vehicle's surroundings, and a localization module coupled to the digital map and the surroundings recognition device. The localization module has a processing unit for comparing the detected data with the data on location-specific features recorded in the digital map, and for localizing the vehicle position based on the location-specific features recorded in the digital map in a localized manner. Summary of the invention

[0013] One object of the present invention is to provide an improved method for creating a positioning layer of a digital positioning map.

[0014] According to a first aspect, the object is achieved by a method for constructing a localization layer of a digital localization map for automated driving, the method comprising the following steps:

[0015] Provide positioning layers for limited areas;

[0016] Provide planning layers for the area; and

[0017] Alignment features are extracted from the positioning layer, which are provided for alignment with the planning layer, wherein the alignment features are extracted from the positioning layer in such a way that an inadmissible deformation of the positioning layer can be detected when aligning the planning layer with the positioning layer.

[0018] Using the positioning layer thus generated, it is advantageously possible to perform the alignment of the planning map and the positioning map with each other by means of the alignment features of the positioning map, so that errors or impermissible deviations between the positioning layer and the planning layer can be simply detected. In this way, it is advantageously possible to produce a correct and usable digital positioning map.

[0019] The proposed method advantageously eliminates the need to test the complete data set for the accuracy of the alignment of the positioning layer with the planning layer. In this way, it is particularly possible to efficiently use planning layers and positioning layers from different manufacturers in the creation of digital positioning maps.

[0020] According to a second aspect, the object is achieved by a device for constructing a localization layer of a digital localization map for automated driving, the device comprising:

[0021] A providing device for providing a positioning layer for a defined area and for providing a planning layer for the area; and

[0022] An extraction device for extracting alignment features from the positioning layer, wherein the alignment features are provided for alignment with the planning layer, wherein the alignment features can be extracted from the positioning layer in such a way that an inadmissible deformation of the positioning layer can be identified when aligning the planning layer with the positioning layer.

[0023] According to a third aspect, the object is achieved by a computer program product.

[0024] Advantageous developments of the method are the subject matter of the dependent claims.

[0025] An advantageous development of the method provides that the alignment feature is designed to be structurally defined. In this way, the alignment feature in the positioning layer forms a structure, the changes of which can be easily noticed when aligning the layers in question.

[0026] Another advantageous development of the method provides that the alignment feature is configured according to a defined pattern. With this measure, in the alignment of the mentioned layers, inadmissible deformations of the positioning layer can also be easily observed.

[0027] Another advantageous development of the method provides that the alignment feature is constructed according to at least one defined statistical criterion. This allows for the use of high-performance statistical tools for extracting and constructing alignment features, using which inadmissible deviations can be easily detected when aligning the layers in question.

[0028] Another advantageous embodiment of the method is characterized in that the statistical criterion comprises at least one of the following aspects: density distribution, variance. In this way, a known statistical criterion is provided for detecting deformations of the alignment features when aligning the aforementioned layers.

[0029] Another advantageous extension of the method is characterized in that the maximum permissible degree of distortion of the alignment feature when aligning the positioning layer with the planning layer is defined by a threshold value. In this way, a control variable is advantageously provided, by means of which it is possible to define to what extent the deformation of the control layer is still permissible. Thus, a threshold value is provided, which defines to what extent the alignment of the mentioned layers is still permissible for providing a usable digital positioning map for automated driving. This is particularly advantageous when the positioning layer and the planning layer are provided by different manufacturers or when the two mentioned layers originate from different sensor groups.

[0030] The invention is described in detail below with further features and advantages with the aid of a number of drawings. The technical features and advantages of the described device are derived in an analogous manner from the technical features of the described method and the device. BRIEF DESCRIPTION OF THE DRAWINGS

[0031] In the accompanying drawings:

[0032] Figure 1 : Schematic illustration of the proposed device for constructing a localization layer of a digital localization map for automated driving;

[0033] Figures 2 to 4 : A diagram for illustrating in principle the proposed arrangement of alignment features for positioning layers; and

[0034] Figure 5 : Principle illustration of a flow chart of the proposed method for constructing a localization layer of a digital localization map for automated driving. DETAILED DESCRIPTION

[0035] The core idea of ​​the present invention is in particular to enable improved alignment of the positioning layer and the planning layer of the digital positioning map. In particular, errors in the alignment of the mentioned layers should be identified and the production process of the digital positioning map can therefore be efficiently controlled.

[0036] In the alignment process, a non-linear transformation is preferably applied to the positioning layer in order to accurately align the positioning layer with respect to the planning layer. It must be ensured that the applied transformation does not destroy the digital positioning map or invalidate it (error control). Here, preferably statistical measures are performed before and after the alignment process in order to determine whether the alignment process is not allowed or whether the alignment cannot be successfully performed.

[0037] If the alignment features are derived in such a way that they have important positioning layer statistical properties or parameters (eg, density distribution, variance, etc.) from which they are derived, then the alignment features themselves can be used to estimate the error introduced due to the non-linear transformation.

[0038] In this way, the complete data material of the mentioned layers is therefore advantageously not required for checking the alignment process. This caters to data protection interests, for example, if the mentioned map layers originate from different manufacturers who have an interest in data protection regarding the map layers they provide.

[0039] The proposed method can advantageously accelerate the alignment process, since the calculation of the iterative measures based on the alignment features reduced in scope is generally much faster than using the complete data of the positioning layer.

[0040] During the alignment of the positioning layer, the positioning layer is subjected to a nonlinear transformation which seeks to determine the alignment features identified in both layers (positioning layer and planning layer) with sufficient accuracy. Since the identification of consistent alignment objects or alignment features is an automated process, errors may occur here (e.g. the positioning layer is incorrectly shifted in such a way that it is not usable for the current positioning process).

[0041] Therefore, according to the invention, for the purpose of extracting the alignment features, statistical measures, such as density distribution or variance, are determined before and after the aforementioned nonlinear transformation. The analysis of the statistical measures can be used to determine erroneous transformations (for example, when the difference between the density at the alignment features or the positioning object in the theoretical area exceeds a threshold value after the transformation).

[0042] It is proposed to calculate the alignment features in such a way that the error control can be directly performed using the alignment features 1a . . . 10 , without having to use a complete positioning layer for this purpose. For this purpose, the alignment features are selected in such a way that errors can be easily identified.

[0043] The following description illustrates the proposed principle of extracting alignment features in a qualitative way.

[0044] Figure 1A block diagram of a device 100 for constructing a positioning layer of a digital positioning map for automated driving is shown. The device 100 has a providing device 10 for providing a positioning layer 12 for a defined area and for providing a planning layer 11 for the area. The device 100 also has an extraction device 20 for extracting alignment features 1a...1n from the positioning layer 12, wherein the alignment features 1a...1n are configured to be aligned with the planning layer 11, wherein the alignment features 1a...1n can be extracted from the positioning layer 12 in such a way that an unacceptable deformation of the positioning layer 12 can be identified when the planning layer 11 is aligned with the positioning layer 12.

[0045] Figure 2 A schematic representation of a localization layer 12 is shown with landmarks detected, for example, by means of a radar sensor.

[0046] Figure 3 Shown with Figure 2 The statistical distribution of the alignment objects 1a...1d does not correspond to the conventional alignment features 1a...1d extracted from the positioning layer 12 of FIG. Figure 2 Arrangement of landmarks.

[0047] exist Figure 4 It is shown in FIG. 1 that the alignment features 1a ... 1n are now structured according to the invention in a manner such that Figure 4 In the case of , the statistical distribution of the aligned objects corresponds to Figure 2 Arrangement of landmarks.

[0048] It can be seen that with the help of Figure 4 The merely illustrative and exemplary regular arrangement of the alignment features 1a...1n of the positioning layer 12 makes it easy to identify unacceptable distortions of the alignment features 1a...1n of the positioning layer 12 relative to the planning layer 11 (not shown), and thus to well determine incorrect alignments of the positioning layer 12 with the planning layer 11. As a result, a correct digital positioning map production process is supported already at an early stage.

[0049] With the surrounding data provided by the method according to the invention, it is advantageously possible to create a digital positioning map that is as complete as possible and highly available, very accurate and robust. Advantageously, this general principle is very simple and enables many implementations and application scenarios that were previously described only by way of example.

[0050] The device 100 is preferably designed as a computer device in the cloud, on which a computer program product is executed.

[0051] Figure 5The basic flow of the proposed method for constructing a positioning layer of a digital positioning map for automated driving is shown.

[0052] In step 200, a positioning layer is provided for a defined area.

[0053] In step 210, a planning layer is provided for the area.

[0054] Finally, in step 220, alignment features 10a...10n are extracted from the positioning layer, wherein the alignment features are configured for alignment with the planning layer, wherein the alignment features 10a...10n are extracted from the positioning layer in such a way that unacceptable deformations of the positioning layer can be identified when aligning the planning layer with the positioning layer.

[0055] Advantageously, the method according to the invention can be implemented as software which runs on a server device, preferably in a cloud-based manner. The simple adaptability of the method is supported in this way.

[0056] Without departing from the core of the present invention, those skilled in the art may modify and / or combine the features of the present invention in an appropriate manner.

Claims

1. A method for constructing a positioning layer (12) of a digital positioning map for automated driving, the method comprising the following steps: A location layer (12) providing a digital location map for a defined area; Providing a planning layer (11) of the digital positioning map for the area; and Alignment features 1a to 1n are extracted from the positioning layer (12), and the alignment features 1a to 1n are configured to be aligned with the planning layer (11), wherein: The alignment features 1a to 1n are extracted from the positioning layer (12) in such a way that an unacceptable deformation of the positioning layer (12) can be identified when the planning layer (11) is aligned with the positioning layer (12). The constructed positioning layer is used so that the planning layer and the positioning layer are aligned with each other by means of the alignment features 1a to 1n of the positioning layer, so that errors or unacceptable deviations between the positioning layer and the planning layer can be detected. There is no need to verify the complete data set to verify the accuracy of the alignment between the positioning layer and the planning layer. In the alignment process, a non-linear transformation is applied to the positioning layer so as to accurately align the positioning layer with respect to the planning layer.

2. The method according to claim 1, wherein: The alignment features 1 a to 1 n are designed to be structurally defined.

3. The method according to claim 1 or 2, wherein: The alignment features 1a to 1n are configured according to a defined pattern.

4. The method according to claim 1 or 2, wherein: The alignment features 1 a to 1 n are configured according to at least one defined statistical criterion.

5. The method according to claim 4, wherein: The statistical criterion includes at least one of the following aspects: density distribution and variance.

6. The method according to claim 1 or 2, wherein: A threshold value is used to define the maximum permissible degree of distortion of the alignment features 1a to 1n when aligning the positioning layer (12) with the planning layer (11).

7. A device (100) for constructing a positioning layer (12) of a digital positioning map for automated driving, the device comprising: A providing device (10) for providing a positioning layer (12) of a digital positioning map for a defined area and for providing a planning layer (11) of a digital positioning map for the area; and An extraction device (120) is used to extract alignment features 1a to 1n from the positioning layer (12), wherein the alignment features 1a to 1n are configured to be aligned with the planning layer (11), wherein: The alignment features 1a to 1n can be extracted from the positioning layer (12) in such a way that an inadmissible deformation of the positioning layer (12) can be detected when the planning layer (11) is aligned with the positioning layer (12). The constructed positioning layer is used so that the planning layer and the positioning layer are aligned with each other by means of the alignment features 1a to 1n of the positioning layer, so that errors or unacceptable deviations between the positioning layer and the planning layer can be detected. There is no need to verify the complete data set to verify the accuracy of the alignment between the positioning layer and the planning layer. In the alignment process, a non-linear transformation is applied to the positioning layer so as to accurately align the positioning layer with respect to the planning layer.

8. A computer program product having a program code unit, wherein the program code unit is used to execute the method according to any one of claims 1 to 6 when the computer program product is run on an electronic device (100) for constructing a positioning layer of a digital positioning map for automated driving or is stored on a computer-readable data carrier.

Images