Method and system for creating a positioning map for a vehicle

Abstract

A method for creating a radar localization map (K) comprises the following steps: a) detecting a defined area (10) by means of environmental sensors of a mapping vehicle; b) providing recorded satellite data (SD) of the defined area (10) by means of a satellite; c) determining identically detected objects of the area (10) in the environmental sensor data and in the recorded satellite data (SD); d) creating a transformation model (M) from the identically detected objects, wherein the recorded satellite data (SD) can be converted alternately into environmental sensor data by means of the transformation model (M); e) creating a radar localization map (K) by means of the recorded satellite data (SD) using the transformation model (M), wherein the recorded satellite data (SD) are converted into corresponding data of the radar localization map (K).

Description

Technical Field

[0001] The present invention relates to a method for creating a positioning map for a vehicle. The present invention also relates to a system for creating a positioning map for a vehicle. The present invention also relates to a computer program product having program code means for implementing the method. Background Art

[0002] Methods for fleet mapping and the creation of radar position maps are already known.

[0003] It is already known to determine the precise vehicle position by identifying landmarks and comparing them with a map. Here, the vehicle uses radar sensors to determine a single, unambiguous landmark in its surroundings and compares it with a map stored on the vehicle. Thus, the precise vehicle position can be determined by combining multiple landmarks.

[0004] Known by A method is performed in which a three-dimensional building model is created from satellite information.

[0005] DE 10 2016 210 495 A1 discloses a method and a device for creating an optimized positioning map, as well as a method for creating a positioning map for a vehicle, in which method data from radar satellites are used.

[0006] WO 2017 / 215964 A1 discloses the creation of a radar map using satellites, which is suitable for positioning a vehicle. A method for creating an optimized positioning map for a vehicle is provided. The method includes the step of providing at least one positioning map, which represents at least one position of a landmark read by a vehicle reading unit. The method also includes the step of reading a radar map via an interface, wherein the radar map includes or depicts at least one further position of the landmark in the radar map, provided by radar measurements using a satellite. Finally, the method includes the step of generating and storing an optimized positioning map by combining the positioning map with the radar map. Summary of the Invention

[0007] The object of the present invention is to provide an improved method for creating a localization map for a vehicle.

[0008] According to a first aspect, the object is achieved by means of a method for creating a localization map for a vehicle, the method comprising the following steps:

[0009] a) Detecting a defined area using environmental sensors of a mapping vehicle;

[0010] b) providing satellite data recorded over a defined area using satellites;

[0011] c) ascertaining consistently detected objects of the region in the surrounding sensor data and in the recorded satellite data;

[0012] d) creating a transformation model from the consistently detected objects, wherein the recorded satellite data can be converted alternately into environmental sensor data using the transformation model;

[0013] e) Creating a positioning map using the acquired satellite data using a transformation model, wherein the acquired satellite data are converted into corresponding data of the positioning map.

[0014] In this way, a radar map is created with a training system in the form of a so-called "conversion model," which converts recorded satellite data into environmental sensor data and vice versa. Advantageously, this allows the creation of a positioning map solely using satellite data, taking advantage of the fact that the update rate of recorded satellite data is generally very high and thus supports a high degree of currency in the positioning map.

[0015] Advantageously, the proposed method does not require a comprehensive tour of the area to be mapped using a mapping vehicle. Advantageously, in this way, a positioning map can be provided solely using recorded satellite data and a transformation model.

[0016] Advantageously, this allows the fleet of mapping vehicles to be significantly reduced. Using the mapping vehicles, only a statistically reliable image of a defined area needs to be created. Advantageously, the radar map can then also be "expanded" to areas that have not yet been traveled.

[0017] According to a second aspect, the object is achieved by means of a system for creating a localization map for a vehicle, the system comprising:

[0018] a model device having a transformation model, wherein the environmental sensor data and the recorded satellite data can be converted alternately into one another by means of the transformation model;

[0019] A mapping device, which is functionally connected to the model device, is designed to provide a position map using the transformation model and the provided recorded satellite data.

[0020] An advantageous embodiment of the method provides that steps a) to d) are carried out once or multiple times. In this way, the provision of the transformation model can be carried out with any desired accuracy.

[0021] Another advantageous embodiment of the method provides that step e) is performed once or multiple times, in which way a positioning map can be created in each case using the latest recorded satellite data. This advantageously supports a high accuracy of the positioning map.

[0022] Another advantageous embodiment of the method provides for performing manual classification in step d) to alternately convert the recorded image data into environmental sensor data. This allows for classification of relevant and unimportant objects, allowing them to be distinguished from one another. For example, houses, traffic signs, natural objects, and the like can be distinguished from signal noise, which can then be ignored and not incorporated into the creation of the positioning map. Consequently, real objects and satellite images can be associated with one another.

[0023] Another advantageous embodiment of the method provides that step e) is performed using currently recorded satellite data of the mapping area. This allows an up-to-date positioning map to be created using the latest recorded satellite data, advantageously without having to travel the mapping area with a mapping vehicle.

[0024] Another advantageous embodiment of the method provides for using a radar sensor as the surroundings sensor of the mapping vehicle. This allows the surroundings sensors that are usually already present in the vehicle to be used to cost-effectively provide the surroundings sensor data.

[0025] According to a third aspect, the object is achieved by means of a computer program product having program code means that runs on the proposed system or is stored on a computer-readable storage medium. Advantageously, the method can be configured as software in this way, thereby making it easy and efficient to modify and adapt the method.

[0026] Other measures for improving the present invention will be further illustrated below with reference to the accompanying drawings and in combination with the description of preferred embodiments of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0027] In the accompanying drawings:

[0028] Figure 1 A system diagram with a diagram of the proposed method for creating a positioning map for a vehicle is shown;

[0029] Figure 2 A block diagram illustrating a system for creating a positioning map for a vehicle;

[0030] Figure 3 The basic sequence of the proposed method for creating a localization map for a vehicle is shown.

[0031] The drawings are only schematic and to scale. In the drawings, identical, similar or similar elements are always provided with the same reference numerals. DETAILED DESCRIPTION

[0032] The present invention proposes a system similar to the one mentioned above. The system uses captured satellite images to create a 3D model in order to create a map of radar landmarks (i.e., a radar map or radar positioning map). To do this, reference data must first be input using a mapping vehicle or a fleet of mapping vehicles that use high-precision GPS to create a radar image of the area. The landmarks thus determined are recorded from the perspective of the mapping vehicle and processed into a reference positioning map or transformation model M. This processing can be performed, for example, on a central processing unit (e.g., cloud-based).

[0033] Figure 1 A system diagram with a schematic representation of an embodiment of the proposed method is shown. In a first step, a defined area 10 is detected in close temporal proximity using two technically different devices.

[0034] On the one hand, in step 20, area 10 is recorded by satellite, and recorded satellite data SD are provided. In step 30 (temporally adjacent to step 20), a mapping vehicle (not shown) is driven through essentially the same defined area 10 and records it using environmental sensors (preferably radar sensors), thereby detecting landmarks specific to area 10 (e.g., houses, traffic signs, buildings, natural objects, etc.).

[0035] In step 40, the recorded satellite data generated in step 20 in the form of a "satellite image" is combined with the radar data of the mapping vehicle, wherein objects that are consistently ascertained in the recorded satellite data and the radar data are ascertained. These consistently ascertained objects form a so-called transformation model M, which is preferably taught or trained offline in a computer center using further recorded satellite data and environmental sensor data, which can be done, for example, using a neural network.

[0036] Therefore, it is a prerequisite that the area 10 detected by the mapping vehicle using sensors is recorded by satellite as simultaneously as possible. The time delay between the mapping vehicle and the image recording of the defined area 10 by satellite must be at most large enough that no significant changes occur in the infrastructure of the defined area 10 between the detection by the mapping vehicle and the detection by satellite, because the recording contents of the mapping vehicle and the satellite must be very similar for the purpose of creating the transformation model M.

[0037] The defined area 10 under consideration should preferably include as many, more preferably all, important and relevant landmark types as possible (e.g. street signs, trees, houses, natural objects, etc.). It is also advantageous here to record the captured satellite images from a plurality of angles in order to provide as many different views of the landmarks as possible.

[0038] From now on, the transformation model M created in the above-described manner is available in step 50. Phase A of creating the positioning map is thus completed.

[0039] Therefore, stage A of the proposed method consists in providing and teaching or training a transformation model M that can identify important, relevant landmarks from satellite images. First, all objects that are subsequently expected in the positioning map are classified using the satellite images. This classification can be performed manually or using common classification methods. A set of training data is generated, with which a machine learning algorithm (e.g., a neural network in the form of a convolutional network) can be trained.

[0040] Ideally, the defined area 10 is therefore detected only once from two different angles or viewing angles using the proposed method in phase A. Using the resulting transformation model M, the satellite data or satellite images can be converted into landmarks for the vehicle's positioning map in a subsequent phase B, and vice versa.

[0041] Next, from now on, in phase B, a positioning map is created using the transformation model created in phase A, wherein now only the recorded satellite data is taken into account for the creation of the positioning map.

[0042] For this purpose, the area to be mapped is detected by means of a satellite, preferably high-resolution, photo camera (fotokamera) in step 60. In step 70, the area to be mapped is detected from another perspective, position data (e.g., GPS position data) also being detected and stored together with the data detected by the camera.

[0043] In step 80 , the satellite data are converted into data of a positioning map K using the conversion model M created in phase A, as a result of which a positioning map K is created.

[0044] The positioning map K is thus made available in electronic form and can be used by the user vehicle in a manner known per se to determine a high-precision position of the user vehicle from environmental sensor data detected by sensors (e.g., radar data, lidar data, ultrasound data, camera data, etc.) in conjunction with the positioning map K. In this case, highly accurate knowledge of object positions from satellite data is utilized, which can be used to locate objects of the positioning map K.

[0045] Advantageously, the transformation model M determined in phase A can be used for areas not traveled by the mapping vehicle in order to create there an incomplete positioning map K or even a complete positioning map K. Advantageously, the proposed method can be used to generate positioning maps for areas in which no mapping vehicle or specific survey vehicle has yet traveled, but in which recorded data from specific satellites are present.

[0046] Advantageously, the steps of phase B can be carried out multiple times in a simple manner using new or newly acquired satellite data, whereby the position map K can always be further improved and kept up to date.

[0047] As a result, the proposed method advantageously eliminates the need for large fleets of mapping vehicles, as is done in the prior art, to travel and detect the area to be mapped using sensor technology. This advantageously minimizes the logistical and technical costs of creating a positioning map.

[0048] Advantageously, the radar positioning map K can therefore be created much more formally and cost-effectively much faster.

[0049] Figure 2 A block diagram of a proposed system 200 for creating a positioning map for a vehicle is shown.

[0050] A modeling device 100 is shown, which is used to create a transformation model M and transmit it to a mapping device 110. Mapping device 110 is functionally connected to modeling device 100. Mapping device 110 is used to create a positioning map K for the vehicle from the provided recorded satellite data SD in the above-mentioned manner, using transformation model M.

[0051] Figure 3 The basic sequence of a method for creating a localization map K for a vehicle is shown.

[0052] In step 300 , defined area 10 is detected using environmental sensors of the mapping vehicle.

[0053] In step 310 , the recorded satellite data SD of the defined area 10 are provided by means of satellites.

[0054] In step 320 , identically detected objects of region 10 are ascertained in the surroundings sensor data and in the acquired satellite data SD.

[0055] In step 330 , a transformation model M is created from the consistently detected objects, wherein the recorded satellite data SD can be converted alternately into surroundings sensor data using the transformation model M.

[0056] In step 340 , a positioning map K is created with the aid of the recorded satellite data SD using a transformation model M, wherein the recorded satellite data SD are converted into corresponding data of the positioning map K.

[0057] Advantageously, all mentioned components of the system 200 can be implemented in software, thereby enabling efficient and easy adaptability of the method.

[0058] Those skilled in the art will also realize implementation methods not described above during the practice of the present invention.

Claims

1. A method for creating a positioning map (K) for a vehicle, comprising the following steps: a) detecting a defined area (10) by means of environmental sensors of a mapping vehicle; b) providing recorded satellite data (SD) of the defined area (10) by means of a satellite; c) determining identically detected objects in the region (10) in the surrounding sensor data and in the recorded satellite data (SD); d) creating a transformation model (M) from said consistently detected objects, wherein By means of the conversion model (M), the recorded satellite data (SD) can be converted into the environmental sensor data and / or the environmental sensor data can be converted into the recorded satellite data (SD); e) creating the positioning map (K) using the conversion model (M) solely with the recorded satellite data (SD), wherein the recorded satellite data (SD) are converted into corresponding data of the positioning map (K), The area detected by the mapping vehicle with the aid of the surroundings sensor is recorded by the satellite as simultaneously as possible, wherein the time delay between the mapping of the vehicle and the detection of the defined area with the aid of the satellite is at most so great that no significant changes occur in the infrastructure of the defined area between the detection with the aid of the mapping vehicle and the detection with the aid of the satellite. The conversion model can be used for areas not traveled by the mapping vehicle in order to generate a positioning map for areas in which the mapping vehicle has not yet traveled, but in which satellite data recorded by the satellites are present. In this case, step e) is performed once or multiple times, wherein the positioning map (K) is created in each case using the latest recorded satellite data.

2. The method according to claim 1, wherein Steps a) to d) are performed one or more times.

3. The method according to claim 1 or 2, wherein: In step d), manual classification is performed in order to convert the recorded satellite data (SD) alternately into the environmental sensor data, wherein relevant and unrelated objects are classified and thus differentiated from one another.

4. The method according to claim 1 or 2, wherein: Step e) is carried out using currently acquired satellite data of the mapping area.

5. The method according to claim 1 or 2, wherein: A radar sensor is used as an environmental sensor of the mapping vehicle.

6. A system (200) for creating a positioning map (K) for a vehicle, the system being configured to implement the method according to any one of the preceding claims, the system comprising: A model device (100) having a transformation model (M), wherein: By means of the transformation model (M), environmental sensor data and recorded satellite data can be converted alternately into one another; A mapping device (110) is functionally connected to the model device (100) and is designed to provide the positioning map (K) using the transformation model (M) and the provided recorded satellite data (SD).

7. A computer program product having program code means for carrying out the method according to any one of claims 1 to 5 when the program code means is run on a system (200) for creating a positioning map (K) for a vehicle or when the program code means is stored on a computer-readable storage medium.

Images