METHOD FOR LOCALIZING A VEHICLE

Description

[0001] The invention relates to a method for locating a vehicle, which makes it possible to determine the position of the vehicle in the vehicle itself.

[0002] US 2019 / 056737 A1 describes a vehicle, a system and a method for operating the vehicle.

[0003] The article by ALEMPIJEVIC A ET AL: entitled "An efficient algorithm for line extraction from laser scans", published in ROBOTICS, AUTOMATION AND MECHATRONICS, 2004 IEEE, CONFERENCE ON SINGAPORE 1-3 DEC. 2004, PISCATAWAY, NJ, USA,IEEE, US, describes an algorithm for line extraction from laser scans.

[0004] US 2013 / 131985 A1 describes a wearable electronic image capture and processing system (or vision enhancement system) that guides visually impaired people through their environment, providing the user with information about nearby objects of interest, potentially dangerous obstacles, their location, and possible routes to their destination.

[0005] WO 2018 / 191818 A1 describes systems and methods for an autonomous, self-driving material transport vehicle with mapping of an industrial facility.

[0006] Precise vehicle positioning is of great importance in many modern telematics applications, for example, in autonomous driving or navigation, and in the use of complex driver assistance systems. Autonomous vehicle localization within a vehicle itself requires the actual vehicle position and direction of movement to be specified or estimated with high accuracy. Autonomous driving, in particular, places high demands on the robustness of a localization method, in addition to the high accuracy requirements. Localization accuracy is typically required in the centimeter range.

[0007] There are currently several computer-based image processing methods, such as the perspective-n-point method, which can be used to implement computer-based vision with high accuracy. However, strict boundary conditions must be met. For example, a highly accurate database whose data represents a model of the vehicle's surroundings must be available. Furthermore, most localization methods and systems require that clear structures / textures are visible in the vehicle's surroundings. Furthermore, data mapping must be sufficiently good, whereby the data mapping specifies the relationship between actually visually observed objects / features of the environment and objects / features represented by the database.

[0008] In particular, in some critical cases, known localization systems and methods exhibit weaknesses when using a pre-created database. The performance of known localization systems and methods depends on the quality of a pre-created database containing a model of a vehicle's surroundings. If the database data contains even minor errors, known localization systems and methods are susceptible to incorrect positioning, resulting in the loss of a vehicle's actual lane. Furthermore, known localization systems and methods fail, particularly when the actual surrounding scenery has changed since the database containing the model of the surroundings was created.

[0009] An object of the present invention is to provide a method for localizing a vehicle which provides reliable positioning results even when a database provided by a server contains inaccuracies in the description of the model of the vehicle's surroundings.

[0010] An embodiment of a method which enables precise localization of a vehicle even in the case of an erroneous database containing data of an environmental model of the vehicle is specified in patent claim 1.

[0011] According to the method for locating a vehicle, it is first checked whether an external database created by an external server, whose data represents a model of the vehicle's surroundings, is available in the vehicle for locating the vehicle. If it is determined that the external database is actually available in the vehicle, the next step is to check whether the external database can be used to locate the vehicle.

[0012] If it is determined that the external database can be used to locate the vehicle, the vehicle is located using a position of the vehicle determined using the external database. If, however, it is determined that the external database cannot be used to locate the vehicle, the vehicle is located by determining the vehicle's position using a map of the vehicle's surroundings, which map was created in the vehicle while the vehicle was traveling.

[0013] According to the proposed method, the localization of the vehicle in the vehicle while driving can thus be carried out in two possible ways.

[0014] If an external database with data representing an exact model of the vehicle's surroundings is provided by an external server, the vehicle's control computer can use the database to localize the vehicle. To localize the vehicle, an optical method can be used in the vehicle. This method first detects objects that stand out particularly well from the scene. A camera in the vehicle first takes an image of the vehicle's surroundings. During the subsequent evaluation of the image in the vehicle, feature points and / or objects are extracted and compared with the environment model provided by the external database. The feature points can, for example, describe edges or corners of objects in the scene or other image areas that stand out in the scene.In particular, the feature points and / or objects identified in the captured image are compared with feature points and / or objects in the environment model of the external database. By comparing the feature points and / or objects captured in the captured image of the vehicle's current surroundings with the feature points and / or objects contained in the environment model of the database, the vehicle's position can ultimately be determined.

[0015] However, the specified method variant only provides correct position data if the external database provided by the server is highly accurate with respect to the actual surroundings of the vehicle. Therefore, if a vehicle control computer determines that the data in the external database provided by the server does not match, or only inadequately matches, the actual surroundings of the vehicle, the method for locating the vehicle is modified according to the invention.

[0016] In such a case, it is proposed that a control computer within the vehicle itself create a (dynamic) map of the vehicle's surroundings and determine the vehicle's position using the map created in the vehicle itself. For this purpose, images of the vehicle's surroundings are captured using a vehicle camera and evaluated by the control computer. During image analysis, feature points and / or objects are extracted from the captured images. These feature points and / or objects are detected for all captured images while the vehicle is moving.

[0017] Using a matching process, the control computer then determines which feature points and / or objects in the consecutively recorded images correspond to each other. Based on the corresponding feature points and / or objects in the recorded images, the control computer can generate and evaluate a dynamic map of the vehicle's surroundings while the vehicle is traveling in order to determine the vehicle's position. The (dynamic) map is created and compared by the control computer during the same journey and evaluated to locate the vehicle's position.

[0018] The invention is explained in more detail below with reference to figures which illustrate a process sequence of the method for locating a vehicle.

[0019] They show: Figure 1Aa flowchart for explaining a first embodiment of a method for locating a vehicle; Figure 1B a flowchart for explaining a second embodiment of a method for locating a vehicle; and Figure 2 different steps of the vehicle location process.

[0020] The Figure 1A and 1B Each of these describes an embodiment of a method V1 for locating a vehicle with high accuracy. Using the two specified embodiments of the method, the position of the vehicle can be determined very accurately even if the external database of the vehicle's environmental model provided by a server is faulty. The individual method steps are executed in the vehicle using a control computer, for example, an ADAS control unit.

[0021] First, a first embodiment of the method is described using the Figure 1A explained. In a method step S11, a check is performed to determine whether an external database created by an external server, whose data represents a model of the vehicle's surroundings, is available in the vehicle for localizing the vehicle. If it is determined that the external database is available in the vehicle, a subsequent method step S12 checks whether the external database can be used to localize the vehicle.

[0022] The vehicle is then located either in a step S13 by using the position of the vehicle, which was determined with the aid of a map of the vehicle's surroundings that was created in the vehicle while the vehicle was traveling, if it was previously determined in method step S13 that the external database cannot be used to locate the vehicle. Alternatively, the vehicle can be located in a method step S14 by using a position of the vehicle that was determined with the aid of the external database, if it was previously determined in method step S12 that the external database can be used to locate the vehicle.

[0023] The external database provided by the server cannot be used for positioning, especially if a data set in the database is incomplete or of poor quality. Furthermore, the database provided by the server cannot be used if the current scene in which the vehicle is located does not match the environment model in the database provided by the server. Furthermore, the external database provided by the server cannot be used if the vehicle's current environment has few structures or textures.

[0024] If it has been determined in method step S11 that an external database containing data describing the vehicle's surroundings is not available, in method step S13, a vehicle's control computer creates a (dynamic) map of the vehicle's surroundings while the vehicle is traveling. The vehicle's position is then determined based on this map created in the vehicle. At the same time, the (dynamic) map is updated during the journey. The creation of the (dynamic) map and the localization of the vehicle thus occur almost simultaneously during the same journey.

[0025] If it has been determined in method step S11 that an external database created by an external server and transmitted to the vehicle is available in the vehicle, but it is determined in method step S12 that the external database cannot be used to locate the vehicle, a (dynamic) map of the surroundings of the vehicle is also created in method step S13 while the vehicle is traveling, and the position of the vehicle is determined using this map.

[0026] If, however, it is determined in method step S11 that an external database containing an environmental model of the vehicle is present in the vehicle and it is determined in method step S12 that this external database can actually be used to localize the vehicle, the position of the vehicle is determined in method step S14 using the external database.

[0027] If the vehicle's position is determined using the external database, in method step S14, an image of the vehicle's surroundings is first captured by a vehicle camera. The image data contained in the image are evaluated in the vehicle using the vehicle's control computer and compared with the data from the external database. In particular, in method step S14, feature points and / or objects are detected and extracted from the captured image. The feature points and / or objects extracted from the captured image are compared with feature points and / or objects determined from the external database in order to find corresponding feature points and / or objects.

[0028] The corresponding feature points and / or objects are found using a matching procedure. Non-corresponding feature points and / or objects, or outliers, are excluded from the vehicle localization procedure. A chi-square test, for example, can be used for this purpose.

[0029] Once the position of the vehicle is determined using the (dynamic) map created in the vehicle in method step S13, successive images of the surrounding scenery are recorded by a vehicle camera, for example a mono camera, while the vehicle is traveling. The image data contained in the images are then evaluated in the vehicle using the vehicle's control computer. In particular, feature points and / or objects are extracted from the recorded images and compared with each other in order to find corresponding feature points and / or objects in the various images. The feature points can be, for example, edges or corners of an object in the vehicle's surroundings or protruding image areas.

[0030] The corresponding feature points and / or objects are thus also determined in method step S13 using a matching process between the images taken at different times during the journey. In method step S13, the vehicle's control computer creates the (dynamic) map in the vehicle by determining a three-dimensional position in the vehicle's surroundings for each of the corresponding feature points and / or for each of the corresponding objects.

[0031] By means of the detected feature points and / or objects, the position of the vehicle can be determined with high precision both in method step S13 and in method step S14.

[0032] The Figure 1B The second embodiment of the method shown differs from that shown in Figure 1Ashown first embodiment of the method in that method step S14 is carried out after method step S11. In method step S11, it is first checked whether an external database, which has been created by an external server and whose data represent a model of the vehicle's surroundings, is available in the vehicle for localizing the vehicle. If it is determined that no external database is present in the vehicle, the vehicle is localized according to method step S13. In this case, the position of the vehicle is determined using the (dynamic) map that was created in the vehicle during the journey.

[0033] If, however, it has been determined in method step S11 that the external database is available in the vehicle, the vehicle's position is first determined using the external database in method step S14. Using the determined position, a check is then carried out in method step S12 to determine whether the external database can be used to locate the vehicle. If the check in method step S12 determines that the external database can be used to locate the vehicle, the vehicle is located using the vehicle's position, which was previously determined in method step S14 using the external database.

[0034] The position of the vehicle is determined in particular by creating the map in the vehicle if it is determined in method step S12 that the position previously determined in method step S14 with the aid of the external database has insufficient accuracy.

[0035] If, however, the check in method step S12 has determined that the external database cannot be used to locate the vehicle, the vehicle is located in method step S13 by using the position of the vehicle that was determined with the help of the (dynamic) map of the vehicle's surroundings that was created in the vehicle while the vehicle was traveling.

[0036] Determining the position of the vehicle using process steps S13 or S14 corresponds to the process sequence as described in the Figure 1A for process steps S13 and S14.

[0037] It should be noted that in both the first embodiment of the method and the second embodiment of the method, both method step S13 and method step S14 have a relatively long computing time. Assuming that the vehicle has continued to move during the evaluation and calculations of the individual method steps of method sequence V1, it is not possible to determine exactly where the vehicle actually is at the current location. It is only possible to determine where the vehicle was a certain time ago, i.e., at the beginning of the computing time for method sequence V1.

[0038] The one in the Figure 1A and 1B The process sequence V1 shown can therefore be extended by the Figure 2 The procedures V2 and V3 described in Figure 2 The procedure extended by the procedures V2 and V3 corresponds to a dead reckoning.

[0039] In process sequence V2, a movement or a movement state of the vehicle is determined by evaluating sensor data provided by vehicle sensors in the vehicle. For this purpose, data from vehicle sensors, for example, steering wheel sensors, wheel sensors, or an inertial measurement unit, are evaluated. Starting from a known fixed point, which is connected to the Figure 1 Once the position has been determined according to the specified procedure V1, a further, current position of the vehicle can be determined or estimated using dead reckoning based on the detected movement of the vehicle. Using the vehicle's movement, a further, current position of the vehicle, including altitude information, can thus be approximately determined or estimated.

[0040] By comparing the position estimated using procedure V2 with the position precisely determined using procedure V1, an error in the position estimated using dead reckoning can be determined. The further position approximately determined using procedure V2 can be corrected using the vehicle's position, which was determined using the external database or the map created in the vehicle in procedure V1. Kalman filtering can be used for the correction.

Claims

1. A method for locating a vehicle, comprising: - checking (S11) whether an external database which has been created by an external server and whose data represent a model of an environment of the vehicle is available in the vehicle for locating the vehicle, - checking (S12) whether the external database can be used to locate the vehicle if it has previously been determined that the external database is available in the vehicle, - capturing an image of an environment of the vehicle by a camera in the vehicle, - locating (S14) the vehicle by using a position of the vehicle determined by means of the captured image and the external database, if it has been previously determined that the external database can be used for locating the vehicle, and locating (S13) the vehicle by using the position of the vehicle which by means of the captured image and a map of the environment of the vehicle created in the vehicle while the vehicle is moving, if it has been previously determined that the external database cannot be used for locating the vehicle.

2. The method according to claim 1, wherein the position of the vehicle is determined by creating the map in the vehicle when it is determined that the position determined by means of the external database has insufficient accuracy.

3. The method according to any one of claims 1 or 2, wherein the external database provided by the server cannot be used to determine the position if a data set of the database is not complete.

4. The method according to any one of claims 1 to 3, wherein the position of the vehicle is determined by creating the map in the vehicle when it has been determined that the environment of the vehicle does not match the model of the environment represented by the data of the external database.

5. The method according to any one of claims 1 to 4, wherein the position of the vehicle is determined by creating the map in the vehicle when it has been determined that the environment of the vehicle has no or less than a predefined number of textures.

6. A method according to any one of claims 1 to 5, wherein the position of the vehicle is determined by means of the external database by evaluating image data present in the image and comparing it with the data in the external database.

7. The method according to claim 6, wherein feature points and / or objects are extracted in the captured image and compared with feature points and / or objects determined from the external database in order to find feature points and / or objects corresponding to one another in the captured image and in the external database.

8. The method according to claim 7, wherein feature points and / or objects which do not correspond to each other are excluded from the method for locating the vehicle.

9. The method according to any one of claims 1 to 8, wherein the map is created in the vehicle by capturing successive images from a camera of the vehicle while the vehicle is moving and evaluating image data present in the images.

10. The method according to claim 9, wherein feature points and / or objects are extracted from the captured images and compared with each other in order to find feature points and / or objects in the captured images which correspond to one another.

11. The method according to claim 10, wherein the map is created in the vehicle by determining a three-dimensional position in the environment of the vehicle for each of the feature points corresponding to one another and / or for each of the objects corresponding to one another.

12. The method according to any one of claims 1 to 11, comprising: - determining a movement of the vehicle by evaluating sensor data provided by vehicle sensors of the vehicle in the vehicle, - determining a further position of the vehicle depending on the determined movement of the vehicle.

13. The method of claim 12, further comprising: - correcting the determined further position of the vehicle using the position of the vehicle determined by means of the external database or the map created in the vehicle.