Generating or updating a digital representation of a trajecture, self-positioning of an

Abstract

In order to generate or update a digital representation of a trajectory (T1, TN) in a predefined spatial region, a digital representation of the trajectory (T1, TN) is generated along the trajectory (T1, TN). TN) drives each of a plurality of times at which the capturing vehicle (1 ') is driven, a camera image of the environment of the capturing vehicle (1') is captured by a camera (3) of the capturing vehicle (1 '), at least one feature descriptor of the respective camera image is determined, and a two-dimensional pose of the capturing vehicle (1') in a digitally represented coordinate system is determined from the camera image, a height (z) of the capturing vehicle (1 ') is determined, and the height (z) is determined. And generating or updating a respective data set of the digital representation, the data set comprising the respective at least one feature descriptor, the respective attitude of the captured vehicle (1 ') and the respective height (z) of the captured vehicle (1').

Description

Technical Field

[0001] The present invention relates to a method for generating or updating a digital representation of a trajectory in a predefined spatial region, a method for self-localization of a self-vehicle, and a method for at least partially automatically guiding a self-vehicle. The present invention also relates to a corresponding system for generating or updating a digital representation of a trajectory in a predefined spatial region, a system for self-localization of a self-vehicle, an electronic vehicle guidance system, and a computer program product. Background Art

[0002] For autonomous or semi-autonomous driving functions of a vehicle, such as autonomous or semi-autonomous parking, the vehicle's system must be able to determine the vehicle's position in its environment. Global Navigation Satellite Systems GNSS can in principle be used for this purpose. However, they may have limitations in terms of accuracy, especially when the vehicle is located in a building such as a parking facility, for example, a parking garage or an underground parking garage.

[0003] It is also well known to use methods for Simultaneous Localization and Mapping SLAM to construct a local map and / or a trajectory in the vehicle or in the vehicle's environment based on sensor data from one or more sensor systems of the vehicle. In the case where a camera is used as a sensor, this method is denoted as Visual Simultaneous Localization and Mapping VSLAM.

[0004] VSLAM is mainly based on a solution of feature points for constructing a local map or a trajectory. In the case where the scenario is an underground parking facility or a multi-story parking facility, considering that each floor of a multi-story parking facility or an underground parking facility may look very similar visually, self-localization may be unreliable. In this case, features obtained from one floor may match features obtained from other parking floors. This makes it difficult for VSLAM to determine the vehicle's position relative to the digital map.

[0005] Document CN 113947636 A relates to a laser SLAM positioning system and method based on deep learning. Ground points, planes, and edge features are extracted from the point cloud. For this purpose, the Random Sample Consensus RANSAC algorithm is used. Laser odometry is used to optimize the pose.

[0006] Other SLAM or VSLAM algorithms are based on filter algorithms, especially based on Kalman filtering or Extended Kalman filtering.

[0007] Document WO 2019 / 073038 A1 describes a method for automatically parking a vehicle in a parking space. In a training step, the vehicle is manually driven into the parking space, and then in a replay step, the vehicle is automatically driven into the parking space. For the autonomous driving of the vehicle, characteristic information corresponding to the driving trajectory in the vehicle's environment is detected, the characteristic descriptors of the detected environmental characteristics are matched with the characteristic descriptors stored in a digital map, and the vehicle is repositioned with reference to the trajectory stored in the digital map so as to navigate the vehicle into the parking space along the stored trajectory. Among them, the characteristic descriptor involves the algorithmic representation of how the corresponding characteristic looks. For example, the characteristic descriptor can involve a part of a two-dimensional visual image of the environment. Summary of the Invention

[0008] The object of the present invention is to improve the reliability of self-positioning and / or at least partial automatic guidance of a vehicle in a spatial area based on camera images.

[0009] This object is achieved by the subject matter of the independent claims. Further embodiments and preferred embodiments are the subject matter of the dependent claims.

[0010] The present invention is based on the idea that when generating or updating the digital representation of a trajectory, the height of the capturing vehicle is explicitly tracked, which can be used for the self-positioning or at least partial automatic guidance of the vehicle along a pre-recorded trajectory.

[0011] According to one aspect of the present invention, there is provided a method for generating or updating a digital representation of a trajectory in a predefined spatial area. A capturing vehicle, in particular a motor vehicle, is driven along a trajectory in the spatial area, and for each of a plurality of moments when driving the capturing vehicle along the trajectory, a corresponding camera image of the environment of the capturing vehicle is captured by a camera of the capturing vehicle. For each of the plurality of moments, at least one characteristic descriptor of the corresponding camera image is determined according to the corresponding camera image, in particular determined by at least one computing unit of the capturing vehicle. For each of the plurality of moments, according to the corresponding camera image, the two-dimensional pose of the capturing vehicle in the coordinate system of the digital representation, in particular a two-dimensional coordinate system, is determined, in particular determined by at least one computing unit. For each of the plurality of moments, the height of the capturing vehicle at the corresponding moment is determined. For each of the plurality of moments, a corresponding data set of the digital representation is generated or updated and in particular stored, which includes the corresponding at least one characteristic descriptor, the corresponding pose of the capturing vehicle, and the corresponding height of the capturing vehicle.

[0012] The spatial area is in particular a three-dimensional spatial area and includes, for example, a multi-storey facility having two or more floors on which the capturing vehicle can travel. Two or more floors can be above ground and / or underground. The spatial area can, for example, include a multi-storey underground and / or above-ground parking facility.

[0013] The digital representation comprises or consists of all data sets generated or updated for a plurality of moments. The digital representation and the correspondingly generated or updated data sets are in particular stored on a storage device of the capturing vehicle or on a storage device external to the capturing vehicle, such as a server computer or a cloud computer.

[0014] A camera is mounted to the capturing vehicle such that the field of view of the camera covers a corresponding part of the external environment of the capturing vehicle. For example, the camera can be a front camera of the capturing vehicle.

[0015] A feature descriptor can be understood as an algorithm or a computer-readable expression that describes a corresponding feature in a corresponding camera image. Such features can be extracted and determined from the camera image in a known manner, for example by using object detection algorithms, corner detection algorithms, edge detection algorithms, etc. In this case, corners, edges or another type of object or structure can correspond to the features. The feature descriptor can for example contain or depend on pixel intensity values or pixel intensity gradient values in the camera image corresponding to a given feature. However, there can also be other ways to characterize certain features in the camera image by suitable feature descriptors, and these can also be used in the present invention. In particular, the features and feature descriptors commonly used in VLSAM can also be used in the present invention.

[0016] The features or feature descriptors respectively have the purpose of later identifying a certain part of the environment. If the capturing vehicle or another vehicle travels in a predefined spatial region and captures additional camera images, certain features can be searched for in the additional camera images and an attempt can be made to match them with previously stored features by using their feature descriptors. If a match is found, this information can be used to reposition and / or guide the capturing vehicle or the other vehicle accordingly.

[0017] The coordinate system of the digital representation can be understood as the map coordinate system of a digital map, in which trajectories and optionally other information, such as information about the environment or spatial region, are stored. The two-dimensional pose corresponds to a two-dimensional position in a corresponding coordinate plane and a corresponding orientation angle in this plane relative to a reference direction and a predefined origin of the coordinate system. The height of the capturing vehicle can be understood as the height of the capturing vehicle above or below a predefined reference plane (such as the ground plane).

[0018] Multiple moments and corresponding data sets (which can also be represented as key frames) define a trajectory. However, additional images may be captured between the multiple moments that do not correspond to such key frames. When the method is executed, the trajectory does not have to be predefined. In this case, the data set is initially generated during the course of the method, in particular using VSLAM. However, the digital representation can also exist before the method is executed, and the data set of the trajectory and the trajectory itself can thus be predefined. In this case, the data set is updated such that when the corresponding data set is updated, in particular using VSLAM, the two-dimensional pose is specifically refined or its accuracy is increased.

[0019] The two-dimensional pose is determined based on the corresponding camera image captured at the corresponding moment. However, camera images captured before this specific moment can also be used to determine the two-dimensional pose, for example based on visual odometry.

[0020] The computing unit can in particular be understood as a data processing device including a processing circuit. The computing unit can thus particularly process data to perform computational operations. This can also include operations of performing index access on data structures, such as lookup tables LUT.

[0021] In particular, the computing unit can include one or more computers, one or more microcontrollers, and / or one or more integrated circuits, such as one or more application-specific integrated circuits ASIC, one or more field-programmable gate arrays FPGA, and / or one or more systems-on-chip SoC. The computing unit can also include one or more processors, such as one or more microprocessors, one or more central processing units CPU, one or more graphics processing units GPU, and / or one or more signal processors, in particular one or more digital signal processors DSP. The computing unit can also include a physical or virtual cluster of computers or other said units.

[0022] In various embodiments, the computing unit includes one or more hardware and / or software interfaces and / or one or more memory units.

[0023] The memory unit can be implemented as a volatile data memory, such as dynamic random access memory DRAM or static random access memory SRAM, or as a non-volatile data memory, such as read-only memory ROM, programmable read-only memory PROM, erasable programmable read-only memory EPROM, electrically erasable programmable read-only memory EEPROM, flash memory or flash EEPROM, ferroelectric random access memory FRAM, magnetoresistive random access memory MRAM, or phase change random access memory PCRAM.

[0024] According to the present invention, in addition to the two-dimensional pose and at least one feature descriptor, the corresponding height of the capturing vehicle is determined and stored for each key frame. In this way, the key frames can be used later to reposition or guide the capturing vehicle or another vehicle with higher reliability, because in addition to the feature descriptors used to identify the correct key frames, the height can be used as additional information. This is particularly beneficial in cases where different key frames have the same or approximately the same or similar feature descriptors, especially when the underlying camera images respectively depict parts of visually similar environments or spatial regions. For example, in a multi-level parking facility, the environment may look similar on different levels or floors where the vehicle can drive. Therefore, at least one feature descriptor of different key frames can also be similar. Thus, the correct key frame can be identified by means of the height, or in other words, the available key frames or trajectories can be filtered with respect to the stored height to improve the reliability of repositioning or guidance.

[0025] Hereinafter, a vehicle represented by a later-generated or updated trajectory digital representation is referred to as a ego vehicle. The ego vehicle can be the same as the capturing vehicle, but can also be a different vehicle that has received the generated or updated digital representation for repositioning purposes and / or for the automatic or semi-automatic guidance of the vehicle.

[0026] The height can be determined in different ways known per se. For example, an odometer, especially a visual odometer, can be used to determine the height. Alternatively, a differential global navigation satellite system dGNSS, also known as a differential global positioning system dGPS, can be used. Thus, for example, at least one computing unit of the capturing vehicle can determine the height based on camera images captured at multiple moments and / or additional camera images captured by a camera of the vehicle or another camera. In the case of dGNSS, at least one computing unit can determine the height based on the corresponding position and correction information obtained by the dGNSS receiver system of the vehicle, especially the position and correction information obtained from navigation satellites and base stations installed in or near a predefined spatial region that has an accurately known position in a coordinate system.

[0027] According to multiple embodiments, while driving the capturing vehicle along a trajectory, the motion data of the capturing vehicle is determined, and the two-dimensional pose is determined using an odometer based on the motion data.

[0028] The motion data can include, for example, the number of wheel rotations of the respective wheels of the capturing vehicle when driving along the trajectory and / or the steering angle when the capturing vehicle drives along the trajectory, and can thus be determined by corresponding wheel rotation sensors and / or steering angle sensors, etc.

[0029] In this case, the two-dimensional pose of the capturing vehicle is determined based on the corresponding one or more camera images and motion data. Thus, the accuracy of determining the two-dimensional pose can be improved.

[0030] According to multiple embodiments, an algorithm for VSLAM is used to determine the two-dimensional pose of the capturing vehicle based on camera images and, in the corresponding embodiments if applicable, based on motion data. For example, a filter-based VSLAM algorithm can be used.

[0031] Thus, the two-dimensional pose can be determined with high accuracy, and this is also the case when the capturing vehicle follows a trajectory for the first time or when the capturing vehicle travels in a predefined spatial region for the first time.

[0032] According to multiple embodiments, when the capturing vehicle travels along a trajectory, visual odometry is used to determine the height based on the corresponding camera images captured at multiple moments and / or based on additional camera images captured by the camera of the capturing vehicle or another camera.

[0033] In particular, all camera images available when determining the height or a part of all available camera images can be used to calculate the height.

[0034] Visual odometry can determine the height with high accuracy, especially with improved accuracy compared to other methods (such as mechanical odometry). In addition, compared to methods based on GNSS, for example, visual odometry does not depend on the availability of the corresponding radio signals or the good reception of such signals. This makes visual odometry particularly suitable for scenarios inside buildings, such as multi-story parking facilities.

[0035] According to multiple alternative embodiments, dGNSS is used to determine the height.

[0036] With dGNSS, higher accuracy of the height can be achieved compared to methods using conventional GNSS. For dGNSS, the satellite receiver of the capturing vehicle can receive the corresponding positioning signals from two or more navigation satellites and a correction signal from a base station, which is arranged in the environment of the capturing vehicle, especially in a spatial region. The base station also receives the position signals from the navigation satellites. Contrary to the position of the capturing vehicle, since its position is pre-determined and known with high accuracy, the base station or another computing unit can determine the correction signal based on the deviation of the base station position, which is calculated based on the position signals received from the navigation satellites and the predetermined position of the base station.

[0037] According to multiple embodiments, the spatial region includes a multi-story parking facility, such as a multi-story parking garage or an underground parking garage with multiple floors or a parking facility with underground and above-ground floors.

[0038] In this case, the present invention is particularly advantageous because it may often occur here that due to the similar visual appearance of the environment and the different floors of the parking facility, the feature descriptors of different key frames are approximately the same.

[0039] According to another aspect of the present invention, there is provided a method for self-localization of a self-vehicle. Wherein, a digital representation of a trajectory is generated or updated by using the method for generating or updating a digital representation of a trajectory according to the present invention. Additional camera images of the environment of the self-vehicle are captured by a camera of the self-vehicle, especially when the self-vehicle is located in a predefined spatial region. In particular, at least one additional feature descriptor of the additional camera image is determined by at least one computing unit of the self-vehicle, and the height of the self-vehicle is determined by at least one computing unit of the self-vehicle. One of the data sets of the digital representation is selected according to the determined height of the self-vehicle and the at least one additional feature descriptor, especially by at least one computing unit of the self-vehicle. The two-dimensional pose of the self-vehicle in the coordinate system of the digital representation is determined from the selected data set.

[0040] In particular, the pose of the self-vehicle determined from the selected data set corresponds to the two-dimensional pose of the captured vehicle included in the corresponding selected data set.

[0041] Since the selected data set is selected based on the determined height and the at least one additional feature descriptor, the reliability of the determined pose of the self-vehicle is improved, as described above.

[0042] According to various embodiments of the method for self-localization of a self-vehicle, selecting the data set includes determining a subset of the data sets of the digital representation of the trajectory, wherein for each data set of the subset, the corresponding height of the captured vehicle matches the height of the self-vehicle within a predefined tolerance range. One of the data sets is selected from the subset of the data sets according to the at least one additional feature descriptor.

[0043] In other words, in the first step, the available data sets of the digital representation are filtered according to the corresponding heights of the captured vehicles they store. In the second step, only such data sets are used to attempt to match the at least one additional feature descriptor, and the data sets contain the height of the captured vehicle, which is equal to the determined height of the self-vehicle up to a predefined tolerance range. In this way, the risk of false positive matches of the feature descriptors is reduced.

[0044] According to another aspect of the present invention, there is provided a method for at least partially automatically guiding a self-vehicle. For each of a plurality of reference trajectories, a digital representation of the reference trajectory is generated or updated by using the method according to the present invention. Additional camera images of the self-vehicle environment are captured by a camera of the self-vehicle, particularly when the self-vehicle is located in a predefined spatial region. At least one additional feature descriptor of the additional camera image is determined, in particular, by at least one computing unit of the self-vehicle, and the height of the self-vehicle is determined. One of the plurality of reference trajectories is selected, in particular, by at least one computing unit of the self-vehicle, according to the determined height of the self-vehicle and according to at least one additional feature descriptor. The self-vehicle is at least partially automatically guided along the selected reference trajectory.

[0045] For example, at least one computing unit may generate a control signal for guiding the self-vehicle along the selected reference trajectory and provide the control signal to a corresponding actuator of the self-vehicle, which affects the lateral and / or longitudinal control of the self-vehicle to guide it along the selected reference trajectory.

[0046] For example, by at least partially automatically guiding the self-vehicle along the selected reference trajectory, the self-vehicle can be parked in a parking space or a parking lot.

[0047] According to multiple embodiments of the method for at least partially automatically guiding a self-vehicle, selecting a reference trajectory includes determining a subset of the plurality of reference trajectories, wherein for each reference trajectory of the subset, the height of the capturing vehicle according to at least one data set of the corresponding digital representation matches the height of the self-vehicle within a predefined tolerance range. Then, one of the reference trajectories in the subset is selected according to at least one additional feature descriptor.

[0048] In particular, as described above for the self-localization method, in the first step, a plurality of reference trajectories are filtered according to the height of the self-vehicle, and in the second step, one of the remaining reference trajectories in the subset is selected by matching at least one additional feature descriptor with the corresponding feature descriptor of the reference trajectories in the subset.

[0049] According to another aspect of the present invention, there is provided a system for generating or updating a digital representation of a trajectory in a predefined spatial region. The system is adapted to execute the method according to the present invention for generating or updating a digital representation of a trajectory. To this end, the system may particularly include a camera for the capturing vehicle, at least one computing unit for the capturing vehicle, and corresponding means for determining the height of the capturing vehicle.

[0050] According to another aspect of the present invention, there is provided a system for self-localization of a self-vehicle. The system is adapted to perform a method for self-localization of a self-vehicle according to the present invention. To this end, the system may in particular include a camera for the self-vehicle, at least one computing unit for the self-vehicle, and corresponding means for determining the height of the self-vehicle.

[0051] According to another aspect of the present invention, there is provided an electronic vehicle guidance system for a self-vehicle. According to the present invention, the electronic vehicle guidance system is adapted to at least partially automatically perform a method for guiding a self-vehicle. To this end, the system may in particular include a camera for the self-vehicle, at least one computing unit for the self-vehicle, and corresponding means for determining the height of the self-vehicle.

[0052] The electronic vehicle guidance system can be understood as an electronic system configured to guide a vehicle in a fully automatic or fully autonomous manner, and in particular, without manual intervention or control by the driver or user of the vehicle. The vehicle performs all required functions, such as steering maneuvers, deceleration maneuvers, and / or acceleration maneuvers, as well as automatically monitoring and recording road traffic and corresponding responses. In particular, according to Level 5 of the SAE J3016 classification, the electronic vehicle guidance system can achieve a fully automatic or fully autonomous driving mode. The electronic vehicle guidance system can also be implemented as an Advanced Driver Assistance System (ADAS) to assist the driver in performing partial automatic or partial autonomous driving. In particular, according to Levels 1 to 4 of the SAE J3016 classification, the electronic vehicle guidance system can achieve a partial automatic or partial autonomous driving mode. Herein and hereinafter, SAE J3016 refers to the corresponding standard dated June 2018.

[0053] Therefore, at least partially automatically guiding a vehicle may include guiding the vehicle in a fully automatic or fully autonomous driving mode according to Level 5 of the SAE J3016 classification. At least partially automatically guiding a vehicle may also include guiding the vehicle in a partial automatic or partial autonomous driving mode according to Levels 1 to 4 of the SAE J3016 classification.

[0054] According to another aspect of the present invention, there is provided a first computer program including first instructions. When the first instructions are executed by a system according to the present invention for generating or updating a digital representation of a trajectory, the first instructions cause the system to perform a method for generating or updating a digital representation of a trajectory according to the present invention.

[0055] According to another aspect of the present invention, there is provided a second computer program including second instructions. When the second instructions are executed by a system according to the present invention for self-localization of a self-vehicle, the second instructions cause the system to perform a method for self-localization of a self-vehicle according to the present invention.

[0056] According to another aspect of the present invention, there is provided a third computer program including a third instruction. When the third instruction is executed by an electronic vehicle guidance system according to the present invention, the third instruction causes the electronic vehicle guidance system to execute a method according to the present invention for at least partially automatically guiding a self-vehicle.

[0057] According to another aspect of the present invention, there is provided a computer-readable storage medium storing a first computer program according to the present invention, a second computer program according to the present invention, and / or a third computer program according to the present invention.

[0058] The computer program according to the present invention and the computer-readable storage medium according to the present invention may be represented as corresponding computer program products respectively including first, second, and / or third instructions.

[0059] The first, second, and / or third instructions may be provided as program code. The program code may be provided as binary code or assembly code and / or as source code of a programming language (such as C) and / or as a program script (such as Python). BRIEF DESCRIPTION OF THE DRAWINGS

[0060] In the drawings:

[0061] Figure 1 A vehicle having an exemplary embodiment of an electronic vehicle guidance system according to the present invention is schematically shown;

[0062] Figure 2 A multi-story parking facility is schematically shown; and

[0063] Figure 3 A schematic flow chart showing an exemplary embodiment of a method according to the present invention for at least partially automatically guiding a self-vehicle is shown. DETAILED DESCRIPTION

[0064] Figure 1 A self-vehicle 1 is schematically shown, which includes an exemplary embodiment of an electronic vehicle guidance system 2 according to the present invention. The electronic vehicle guidance system 2 is adapted to execute a method according to the present invention for at least partially automatically guiding the self-vehicle 1 according to the present invention. The self-vehicle 1 may be the same as the capture vehicle 1'. In this case, the electronic vehicle guidance system 2 is also adapted to execute a method according to the present invention for generating or updating a digital representation or a trajectory. The self-vehicle 1 may also be different from the capture vehicle 1'. In this case, according to the present invention, the capture vehicle 1' includes a system 2' for generating or updating a digital representation of a trajectory. Hereinafter, it is assumed that the capture vehicle 1' is the same as the self-vehicle 1 unless otherwise specified. In particular, the self-vehicle 1 also acts as the capture vehicle 1'. For a skilled reader, these explanations also apply to the opposite case.

[0065] The electronic vehicle guidance system 2 includes at least one computing unit, which is represented by a single computing unit 4 in Figure 1 and a camera 3, such as a front camera. Optionally, the electronic vehicle guidance system 2 may include a GNSS receiver, which is particularly adapted to operate according to dGNSS technology.

[0066] Figure 2 The ego vehicle 1 or the capture vehicle 1' in a multi-story parking facility 6 is shown respectively. In Figure 2 a non-limiting example, the parking facility 6 has a ground floor 6a, two floors 6b, 6c above the ground, and two floors 6d, 6e below the ground. Without loss of generality, the height z of the ego vehicle 1 or the capture vehicle 1' can be defined relative to the ground floor 6a.

[0067] Figure 3 A flowchart of an exemplary embodiment of a method for at least partially automatically guiding the ego vehicle 1 according to the present invention is shown. Among them, in steps 310 to 340, a method for generating or updating a digital representation of a trajectory is executed for each of a plurality of reference trajectories T1, TN. To this end, the capture vehicle 1' is driven along the corresponding reference trajectories T1, TN in a corresponding predefined spatial region, such as in the parking facility 6. For a plurality of moments, steps 310 to 340 are repeated multiple times. In step 310, for each of the plurality of moments, the camera 3 of the capture vehicle 1' captures an environmental image of the capture vehicle 1'. In step 320, the computing unit 4 of the capture vehicle 1' determines at least one feature descriptor of the corresponding camera image and determines the two-dimensional pose of the capture vehicle 1' in a predefined coordinate system based on the camera image, particularly using VSLAM. The height z of the capture vehicle 1' is determined in step 330. To this end, the computing unit 4 of the capture vehicle 1' can, for example, apply a visual odometry method to the captured image and / or an additional captured image, or can use the corresponding satellite signals and correction signals received by the optional dGNSS receiver 5 to determine the height z.

[0068] In step 340, the computing unit 4 of the capture vehicle 1' generates or updates a corresponding data set of the digital representation of the corresponding reference trajectories T1, TN. The data set includes the corresponding at least one feature descriptor, the corresponding pose of the capture vehicle 1', and the corresponding height z of the capture vehicle 1'.

[0069] From step 350 onwards, the capture vehicle 1' is now represented as the ego vehicle 1. In an alternative embodiment where the capture vehicle 1' and the ego vehicle 1 are not the same vehicle, the generated or updated representation of the reference trajectories T1, TN is transmitted to the computing unit 4 of the ego vehicle 1.

[0070] In step 350, the ego vehicle 1 is located in a spatial environment, such as a parking facility 6, and a camera 3 of the ego vehicle 1 captures additional camera images of the environment of the ego vehicle 1. In step 360, a computing unit 4 of the ego vehicle 1 uses VSLAM to determine at least one additional feature descriptor of the additional camera images and a two-dimensional pose of the ego vehicle 1. In step 370, the height z of the ego vehicle 1 is determined, for example, by visual odometry or using dGNSS.

[0071] In step 380, the reference trajectories T1, TN are filtered according to the determined height z of the ego vehicle 1. Among them, all reference trajectories T1, TN that do not contain any data sets corresponding to key frames are ignored, where the captured height z of vehicle 1' is equal to the height z of the ego vehicle 1 up to a predefined tolerance range. The computing unit 4 attempts to match at least one additional feature descriptor with a corresponding at least one feature descriptor of the data sets of the remaining reference trajectories T1, TN. If a match is found, the corresponding trajectory is selected. If more than one trajectory match is found, one of them is selected according to a predefined rule, such as based on user input or randomly or based on the history of the ego vehicle 1 or the user of the ego vehicle 1.

[0072] Then, in step 390, the ego vehicle 1 is at least partially automatically guided along the selected reference trajectories T1, TN to a parking space in, for example, the parking facility 6.

[0073] For example, in a multi-level parking facility, considering that each floor of the multi-level parking facility looks almost the same except for some specific markings / details, it may be difficult for the VSLAM algorithm to accurately identify the position of the vehicle. In this case, the features obtained from the camera images captured on one floor may match the corresponding features on other floors. Features obtained from other parking floors. Therefore, the performance of the VSLAM algorithm may be reduced.

[0074] According to multiple embodiments of the present invention, height measurements are recorded together with two-dimensional poses and feature descriptors. Similar scenes around the vehicle on multiple parking floors generate similar data, which may be difficult to distinguish. Therefore, the height information of each key frame helps to identify the most relevant key frames, thereby eliminating false positive key frames recorded on different parking floors. The height can be measured relative to the ground. This measurement can be achieved in different ways, such as using visual odometry or dGNSS.

[0075] This method allows for very reliable filtering and only matches feature points with valid key frames. Based on the current position of the ego vehicle, key frames can be filtered using height information, and the matching between the current frame key points and these filtered frames provides a more reliable repositioning.

Claims

1. A method for generating or updating a digital representation of trajectories (T1, TN) in a predefined spatial region, wherein, For each of a plurality of moments of driving a capture vehicle (1’) along a trajectory (T1, TN), - a camera image of the environment of the capture vehicle (1’) is captured by a camera (3) of the capture vehicle (1’); - at least one feature descriptor of the corresponding camera image is determined, and a two-dimensional pose of the capture vehicle (1’) in a numerically represented coordinate system is determined based on the camera image; - the height (z) of the capture vehicle (1’) is determined; and - a corresponding data set in numerical representation is generated or updated, which includes the corresponding at least one feature descriptor, the corresponding pose of the capture vehicle (1’), and the corresponding height (z) of the capture vehicle (1’).

2. The method according to claim 1, wherein, - when driving the capture vehicle (1’) along the trajectory (T1, TN), motion data of the capture vehicle (1’) is determined; and - the two-dimensional pose is determined based on the motion data using an odometer.

3. The method according to any one of the preceding claims, wherein, an algorithm for visual simultaneous localization and mapping is used to determine the two-dimensional pose of the capture vehicle (1’) based on the camera image.

4. The method according to any one of the preceding claims, wherein, when the capture vehicle (1’) travels along the trajectory (T1, TN), visual odometry is used to determine the height (z) based on the corresponding camera images captured for the plurality of moments and / or based on other camera images captured by the camera (3).

5. The method according to any one of claims 1 to 3, wherein, a differential global navigation satellite system is used to determine the height (z).

6. The method according to any one of the preceding claims, wherein, the spatial region is part of a multi-level parking facility (6).

7. A method for self-localization of a self-vehicle (1), wherein, - a digital representation of a trajectory (T1, TN) is generated or updated by using the method according to any one of the preceding claims; - additional camera images of the environment of the self-vehicle (1) are captured by a camera (3) of the self-vehicle (1); - at least one additional feature descriptor of the additional camera images is determined, and the height (z) of the self-vehicle (1) is determined; - one of the data sets in numerical representation is selected based on the determined height (z) of the self-vehicle (1) and the at least one additional feature descriptor; and - the two-dimensional pose of the self-vehicle (1) in a numerically represented coordinate system is determined from the selected data set.

8. The method according to claim 7, wherein, selecting the data set includes, - determining a subset of the data sets in numerical representation, wherein for each data set of the subset, the corresponding height (z) of the capture vehicle (1’) matches the height (z) of the self-vehicle (1) within a predefined tolerance; and - selecting one of the data sets from the subset of data sets based on the at least one additional feature descriptor.

9. A method for at least partially automatically guiding a self-vehicle (1), wherein, - For each of the plurality of reference trajectories (T1, TN), the digital representation of the reference trajectory (T1, TN) is generated or updated by using the method according to any one of claims 1 to 6; - An additional camera image of the environment of the ego vehicle (1) is captured by a camera (3) of the ego vehicle (1); - Determine at least one additional feature descriptor of the additional camera image and determine the height (z) of the ego vehicle (1); - Select one of the plurality of reference trajectories (T1, TN) based on the determined height (z) of the ego vehicle (1) and the at least one additional feature descriptor; And - The ego vehicle (1) is at least partially automatically guided along the selected reference trajectory (T1, TN).

10. The method according to claim 9, wherein selecting a reference trajectory (T1, TN) comprises: - Determining a subset of the plurality of reference trajectories (T1, TN), wherein for each reference trajectory (T1, TN) of the subset, the height (z) of a vehicle (1’) captured according to at least one data set of the corresponding digital representation matches the height (z) of the ego vehicle (1) within a predefined tolerance; and - Selecting one of the reference trajectories (T1, TN) in the subset based on the at least one additional feature descriptor.

11. The method according to any one of claims 9 or 10, wherein the ego vehicle (1) is parked in a parking space by at least partially automatically guiding the ego vehicle (1) along the selected reference trajectory (T1, TN).

12. A system (2’) for generating or updating a digital representation of trajectories (T1, TN) in a predefined spatial region, wherein, The system (2’) is adapted to perform the method according to any one of claims 1 to 6.

13. A system for self-localization of an ego vehicle (1), wherein the system is adapted to perform the method according to any one of claims 7 or 8.

14. An electronic vehicle guidance system (2) adapted to perform the method according to any one of claims 9 to 11.

15. A computer program product comprising, - Instructions which, when executed by the system (2’) according to claim 12, cause the system (2’) to perform the method according to any one of claims 1 to 6; or - Instructions which, when executed by the system according to claim 13, cause the system to perform the method according to any one of claims 7 or 8; or - Instructions which, when executed by the electronic vehicle guidance system (2) according to claim 14, cause the electronic vehicle guidance system (2) to perform the method according to any one of claims 9 to 11.

Images