Method for positioning a vehicle based on GNSS

Abstract

The invention relates to a method for positioning a vehicle (1) based on GNSS, comprising at least the following steps: a) receiving GNSS satellite signals (2) from GNSS satellites (3) and determining at least one distance information about a distance (4) between the vehicle (1) and a GNSS satellite (3) transmitting the associated GNSS satellite signal (2), b) determining at least one environment information about the surroundings of the vehicle (1) by using image information determined with at least one environment sensor (5) of the vehicle (1), which is capable of capturing images of at least one part of the surroundings of the vehicle (1) from different perspectives, c) determining at least one correction information by using the at least one environment information, d) correcting the at least one distance information by means of the at least one correction information.

Description

Technical Field

[0001] This invention relates to a method for GNSS-based vehicle positioning. Furthermore, a computer program for executing the method, a machine-readable storage medium, and a positioning device for the vehicle are proposed. This method can be used in conjunction with at least partially autonomous or fully autonomous driving. Background Technology

[0002] Providing accurate and reliable positioning is crucial for highly automated driving, especially through GNSS systems. However, in urban environments, particularly in narrow streets and canyons, GNSS reception can be affected. Here, satellite signals may reflect off building walls, causing a time delay in reaching the receiver. This typically leads to an overestimation of the associated pseudorange, resulting in an incorrect position calculation. In these cases of so-called multipath effects, it is particularly important to distinguish between two distinct scenarios. In the first scenario, the receiver still has a line-of-sight (LOS) and receives both direct satellite signals and reflected signals. In the second scenario, the receiver does not have a line-of-sight (NLOS) because it is obstructed, for example, by buildings, and receives only reflected signals. Since reflected signals typically have attenuated amplitude and / or delayed propagation time, in the LOS scenario, reflected signals can be distinguished from directly received satellite signals, and errors can be compensated for (to some extent) by appropriate correlation within the receiver. Conversely, the multipath effects caused in the NLOS scenario are generally more difficult to detect, leading to significantly larger positioning errors.

[0003] Against this backdrop, efforts are being made to improve the compensation or correction of multipath effects, especially in the case of NLOS. Summary of the Invention

[0004] Here, the present invention proposes a method for GNSS-based vehicle positioning, which includes at least the following steps:

[0005] a) Receive GNSS satellite signals from a GNSS satellite and determine at least one distance information regarding the distance between the vehicle and the GNSS satellite transmitting the relevant GNSS satellite signals.

[0006] b) Determine at least one piece of environmental information about the vehicle's surroundings by using image information determined by at least one environmental sensor of the vehicle, which can acquire images of at least a portion of the vehicle's surroundings from different perspectives.

[0007] c) Determine at least one correction information by using at least one environmental piece of information.

[0008] d) Correct at least one distance information by means of at least one correction information.

[0009] In this context, GNSS stands for Global Navigation Satellite System, such as GPS (Global Positioning System) or Galileo. The given order of steps a), b), c), and d) is exemplary and can therefore occur in the regular flow of this method, or at least be run once in the given order. Additionally, at least steps a) and b) can also be performed at least partially in parallel or simultaneously. This method can be performed, for example, by means of a positioning device for a vehicle (also described herein).

[0010] Using the methods described herein, it is advantageous to correct erroneous pseudoranges of reflected GNSS satellite signals from GNSS satellites (NLOS satellites) that do not have a direct line of sight by using additional values ​​determined by one or more cameras. In this regard, it is particularly advantageous to use the structure-of-motion (COMO) method to evaluate the acquired image information. The vehicle may be, for example, a motor vehicle, such as a car. Furthermore, the vehicle may be configured for at least partially automated driving and / or autonomous driving.

[0011] In step a), satellite signals are received from a GNSS satellite, and at least one distance information regarding the distance between the vehicle and the GNSS satellite transmitting the relevant GNSS satellite signals is determined. This reception can be performed via at least one GNSS antenna on the vehicle, which can transmit the received signal, either raw or pre-processed, to the vehicle's positioning equipment. This at least one distance information can be determined by measuring the signal propagation time. In addition to the actual distance between the vehicle and the GNSS satellite, the distance information may, if necessary, include portions attributable to at least one reflection of the relevant GNSS satellite signal. In other words, the distance information may involve or describe pseudoranges that may be excessively long and / or uncorrected.

[0012] In step b), at least one environmental information about the vehicle's surroundings is determined using image information obtained from at least one environmental sensor of the vehicle, which can acquire images of at least a portion of the vehicle's surroundings from different perspectives. The at least one environmental sensor may, for example, include at least one optical sensor and / or at least one acoustic sensor. For example, the at least one environmental sensor may include at least one camera, radar sensor, lidar sensor, ultrasonic sensor, etc. Preferably, the at least one environmental sensor includes at least one camera. The at least one environmental sensor may be (controlled) movably positioned at the vehicle so as to acquire at least a portion of the surroundings from different perspectives. The at least one environmental sensor may preferably be mounted at the vehicle in such a way that it can detect at least a portion of the surroundings from different perspectives due to the movement of the vehicle relative to its surroundings. Furthermore, at least two environmental sensors may be present, and they may be arranged at the vehicle and / or oriented relative to each other such that they can collectively acquire at least a portion of the surroundings from different perspectives.

[0013] In step c), at least one correction information is determined using at least one piece of environmental information. Environmental information may describe, for example, the spatial distance to an object, such as the distance to a building or building wall in the vehicle's surrounding environment. In this regard, correction information may also be determined using multiple, for example, at least two, spatial distances to two objects, such as to buildings or building walls (opposite to each other) in the vehicle's surrounding environment. The correction information may be provided, for example, in the form of a scaling factor or a correction value. The correction value preferably describes an overestimated distance.

[0014] In step d), at least one distance information is corrected using at least one correction information. For example, the correction information can be used to scale the distance information or to reduce the distance information by a correction value to obtain the actual distance between the vehicle and the GNSS satellite transmitting the relevant GNSS satellite signal.

[0015] According to an advantageous design, it is also verified whether one or more GNSS satellite signals are reflected GNSS satellite signals from GNSS satellites without a direct line of sight (NLOS satellites). In this regard, reflected satellite signals can be identified at low carrier-to-noise ratios (C / N0) and / or significant pseudorange residuals. These values ​​can also be observed over time, especially when the vehicle is moving. Thus, jumps in C / N0 and / or residuals (e.g., C / N0 decreases, residuals increase) can allow the inference of reflected signals. Alternatively or additionally, GNSS satellites without a direct line of sight can be identified by means of at least one camera on the vehicle. For this purpose, surrounding buildings in the camera image can be detected by image processing methods such as semantic segmentation. Ray tracing can be performed, for example, with the known azimuth and elevation angles of the satellite to identify whether there is a direct line of sight to the corresponding satellite. Ray tracing methods can be advantageously used to calculate reflections. It is particularly advantageous here that it is sufficient to verify the existence of a direct path, especially without calculating reflections. This can be advantageously implemented with very little computation.

[0016] According to another advantageous design, the environmental sensor includes at least one camera. For example, the environmental sensor may include two or more cameras. The cameras may be arranged at the vehicle and / or oriented relative to each other such that they can capture at least a portion of the environment from different perspectives. The environmental sensor particularly preferably includes a camera system with omnidirectional visibility.

[0017] According to another advantageous design proposal, at least one piece of environmental information is determined by using a motion reconstruction structure (SfM) method. Motion reconstruction structure (SfM) generally refers to a method of obtaining 3D information through the overlay of time-shifted images, particularly using so-called parallax. A particular advantage of the motion reconstruction structure method is that, compared to conventional photogrammetry, it does not require prior specification of a target object with a known 3D position. In other words, motion reconstruction structure describes a method for calculating 3D surfaces using 2D image information from different viewpoints. Thus, distances to pixels and / or objects, for example, can be determined by time-lapse observation of them by a camera from different viewpoints. Therefore, a vehicle equipped with one or more cameras can advantageously determine its distances to surrounding objects, such as, in particular, buildings. With the aid of a camera system with omnidirectional visibility, a 3D model of the surrounding environment can even be advantageously created in this way. This model can be particularly advantageous in determining at least one piece of environmental information and / or at least one correction information.

[0018] According to another advantageous design, at least one piece of environmental information includes at least one spatial distance to an object in the area surrounding the vehicle. This spatial distance may, for example, involve a horizontal distance. The object may, for example, be a building or a building wall.

[0019] According to another advantageous design, at least one correction information describes a measure of the portion of the range information attributable to at least one reflection of the relevant GNSS satellite signal. In other words, this can also be referred to as so-called overestimation of range.

[0020] According to a particularly advantageous design, GNSS pseudorange correction is performed using a motion-restoring structure method. This method can thus provide a particularly advantageous approach for highly automated driving, thereby significantly improving the accuracy of GNSS-based positioning in urban environments. Specifically, the pseudorange of the reflected satellite signal measured from the receiver to the satellite can be corrected using an additional value determined by means of a 3D reconstruction method using a camera and a motion-restoring structure.

[0021] According to another aspect, a computer program for performing the methods described herein is also proposed. In other words, it particularly relates to a computer program (product) comprising instructions that, when run by a computer, cause the computer to perform the methods described herein.

[0022] In another aspect, a machine-readable storage medium is proposed, on which a computer program is stored. Machine-readable storage media are typically computer-readable data carriers.

[0023] According to another aspect, a positioning device for a vehicle is described, configured to perform the methods described herein. In other words, this relates to a positioning device for a vehicle, configured to perform the methods described herein. For example, the aforementioned storage medium may be a component of or connected to the positioning device. The positioning device is preferably disposed in or on a (motorized) vehicle, or provided and configured for installation in or on a vehicle. The positioning device preferably forms a GNSS sensor or includes at least one GNSS sensor. Furthermore, the positioning device may also preferably be provided and configured for autonomous operation of the vehicle. Additionally, the positioning device may be a combination of motion sensors and position sensors. This is particularly advantageous for autonomous vehicles. The positioning device or its processing unit (processor) may, for example, access the computer program described herein to perform the methods described herein.

[0024] The details, features, and advantageous design solutions discussed in conjunction with this method can also be found in the computer programs and / or storage media and / or positioning devices described herein, and vice versa. For this purpose, full reference can be made to the statements herein used to characterize the features in detail. Attached Figure Description

[0025] The solutions and their technical environment described herein are explained in more detail below with reference to the accompanying drawings. It should be noted that the invention should not be limited to the embodiments shown. In particular, unless otherwise expressly stated, certain aspects of the facts explained in the drawings may be extracted and combined with other elements and / or understandings from other drawings and / or this specification.

[0026] Figure 1 A flowchart of the method described herein is shown as an example and illustration.

[0027] Figure 2 The application possibilities of the method are illustrated exemplarily and schematically in a view along the street canyon direction.

[0028] Figure 3 An illustrative and schematic view of the street canyon is shown below. Figure 2 The possibility of its application

[0029] Figure 4 The application possibilities of the method in cases of multiple reflections are illustrated exemplaryly and schematically.

[0030] Figure 5 A vehicle having the positioning device described herein is shown as an example and schematic illustration. Detailed Implementation

[0031] Figure 1 A flowchart of the method described herein is shown illustratively and schematically. This method is used for vehicle 1 (see...). Figure 2 GNSS-based positioning. The order of steps a), b), and c) represented by boxes 110, 120, and 130 is exemplary and can be set up according to the regular operating procedure.

[0032] In block 110, according to step a), GNSS satellite signal 2 is received from GNSS satellite 3, and at least one distance information 4 regarding the distance 4 between vehicle 1 and GNSS satellite 3 transmitting the relevant GNSS satellite signal 2 is determined. In block 120, according to step b), at least one environmental information regarding the surrounding environment of vehicle 1 is determined using image information determined by at least one environmental sensor 5 of vehicle 1, which can acquire images of at least a portion of the surrounding environment of vehicle 1 from different perspectives. In block 130, according to step c), at least one correction information is determined using at least one environmental information. In block 140, according to step d), at least one distance information is corrected by means of the at least one correction information.

[0033] A Global Navigation Satellite System, or GNSS, is a system used for positioning and navigation. This is a collective term for different satellite systems, including GPS, GLONASS, Galileo, and BeiDou. For positioning, satellites communicate their exact positions (elevation and azimuth) and clocks via radio codes. The pseudorange, determined in the receiver, is the distance between the satellite and the receiver based on signal propagation time, including clock errors between the two systems. If four or more satellites are received simultaneously, clock errors can be compensated for, and the receiver's current position can be determined.

[0034] At least one environmental sensor 5 may include at least one camera. Furthermore, at least one piece of environmental information can be determined using a so-called motion-reconstruction method. Thus, a particularly advantageous method can be specified for preferred highly automated driving to improve the accuracy of GNSS-based positioning in urban environments. The pseudorange of the reflected satellite signal measured from the receiver to the satellite can be advantageously corrected by an additional value determined using a 3D reconstruction method involving the camera and motion-reconstruction.

[0035] Figure 2 The application possibilities of the method are illustrated exemplarily and schematically using a view along the street canyon direction. Here, the street canyon is illustrated by two objects 7 in the form of buildings or skyscrapers.

[0036] In this method, it can be (firstly) examined whether one or more GNSS satellite signals 2 are reflected GNSS satellite signals 2 from GNSS satellites 3 that do not have a direct line of sight. In other words, this can also be described in particular as determining (firstly) which pseudorange corrections are applied. Here, in particular, it is examined which(s) of the received satellite signals are actually reflected signals from satellites that do not have a direct line of sight. For this purpose, different methods can be used in principle. For example, reflected satellite signals can be identified by a low carrier-to-noise ratio (C / N0) and / or significant pseudorange residuals. Especially in the case of vehicle movement, these values ​​can also be observed over time. Thus, jumps in C / N0 and / or residuals (e.g., C / N0 decreases, residuals increase) can also indicate new reflected signals.

[0037] Alternatively or additionally, NLOS satellites (i.e., GNSS satellites without a direct line of sight) can also be identified (directly) using a (vehicle's) camera. For this purpose, surrounding buildings in the camera image can be detected using image processing methods such as semantic segmentation. Ray tracing can be performed, for example, with the satellite's known azimuth and elevation angles to determine whether there is a direct line of sight to the corresponding satellite.

[0038] In particular, if the NLOS satellite whose pseudorange should be corrected is known, it is advantageous to calculate the overestimated distance (due to reflection) as a correction value in the next step. Figure 2 An illustrative cross-section perpendicular to the street canyon direction is shown. Here, vehicle 1 receives a reflected signal 2 from NLOS satellite 3 at an elevation angle θ. Due to the reflection, the position of vehicle 1 is incorrectly assumed to be... Figure 2 To the left of building 7 shown on the left side of the middle.

[0039] Here, ε v This is the overestimation of the pseudorange. Based on a right triangle, this value can be calculated as ε. v = 2d*cos(θ), where d is the unknown distance to the building wall. If, for example, a camera is installed in vehicle 1 as an environmental sensor 5 to detect the building wall, this distance can be advantageously determined using a motion-reconstruction 3D reconstruction method. For this purpose, pixels and / or objects are observed from different angles in a time-shifting manner from the moving vehicle 1. The distance d to object 7, in this case, is the distance to the building wall, can then be determined using triangulation.

[0040] In this regard, distance d represents, for example, that at least one piece of environmental information may include, and if necessary, how to include, at least one spatial distance 6 to an object 7 in the area surrounding vehicle 1. Furthermore, the pseudorange overestimates the distance ε. v This is exemplified by the following example, where at least one correction information (e.g., ε) is used. vIt describes, and, where necessary, how to describe, a measure of at least one reflection (in this case, pseudorange) of the NLOS GNSS satellite signal 2 attributable to range information.

[0041] Figure 3 An illustrative and schematic view of the street canyon is shown below. Figure 2 The possibilities for its application. Based on... Figure 3 The diagram exemplarily explains the combination Figure 2 The explanation specifically refers to the correction value ε determined for the two-dimensional case. v How can missing components be supplemented in a three-dimensional context? To this end, in Figure 3 From China Figure 2 The observation is shown from the viewpoint indicated by arrow 9 in the upper left corner (a top view through a plane tilted at an elevation angle θ from satellite signal 2). Here, the (new) correction value ε can also be calculated using a right triangle; this time, the already determined value ε is used. v As the hypotenuse. In the three-dimensional case, the final correction value is ε = ε v *|sin(β)|=2d*cos(θ)*|sin(β)|. Here, angle β is the difference between the azimuth of satellite 3 and the direction of the street canyon. The correspondingly determined correction value for ε represents the correction information particularly preferred in the sense of the method described herein, which is determined by means of the collected environmental information d (spatial distance 6).

[0042] To finally obtain the new pseudorange (the actual distance 4 between vehicle 1 and GNSS satellite 3), the correction value ε is subtracted from the initially measured pseudorange (distance information from step a). This represents an example of correcting at least one distance piece of information using at least one correction value.

[0043] Figure 4 The application possibilities of the method in cases of multiple reflections are illustrated exemplarily and schematically. In other words, in Figure 4 The diagram illustrates how the method can also, and where necessary, be advantageously used for the correction of multiple reflections. Thus, for example in the case of double reflections, the error can also be additionally considered based on the distance d2 to the building 7 (shown on the right here) where the second reflection occurs.

[0044] This leads to the advantageous conclusion that the correction value is ε. m= ε1 + ε2 = 2d1 * cos(θ) * |sin(β)| + 2d2 * cos(θ) * |sin(β)| = 2(d1 + d2) * cos(θ) * |sin(β)|. Here, ε1 represents the correction value associated with the first reflection, ε2 represents the correction value associated with the second reflection, d1 represents the distance from the building wall associated with the first reflection, and d2 represents the distance from the building wall associated with the second reflection. The correspondingly determined correction value ε m represents correction information that is particularly preferred in the sense of the method described herein and is determined by means of the acquired environmental information d1 and d2 (spatial distance 6).

[0045] To identify the presence of double reflections, it is advantageously (also) possible to determine the height of the surrounding buildings by means of structure from motion. Here, especially if the building wall associated with the first reflection is higher than h min,1 = (d1 + 2d2) * tan(θ) / sin(β), and (simultaneously) the height h2 of the building wall associated with the second reflection has a minimum height of h min,2 = d2 * tan(θ) / |sin(β)| and a maximum height of h max,2 = (2d1 + 3d2) * tan(θ) / |sin(β)| (or h min,2 < h2 < h max,2 ), then a double reflection is identified.

[0046] However, double reflections or multiple reflections do not necessarily have to be corrected in this way (i.e., using the correction value ε m ). Because in these cases, the method for correcting single reflections described more generally herein can already achieve an improvement in the pseudorange, and thus an improvement in the position accuracy can be achieved.

[0047] Figure 5 A vehicle 1 with a positioning device 8 described herein is shown示例性 and schematically. The positioning device 8 is set up to perform the method described herein. The positioning device 8 can receive data from the environmental sensors 5 of the vehicle 1.

[0048] Using the method, it is advantageously possible to better compensate or correct the multipath effect, especially in NLOS situations.

Claims

1. A method for GNSS-based vehicle positioning (1), comprising at least the following steps: Receive GNSS satellite signals (2) from GNSS satellite (3) and determine at least one distance information (4) regarding the distance between the vehicle (1) and the GNSS satellite (3) that transmitted the relevant GNSS satellite signals (2). At least one piece of environmental information about the surrounding environment of the vehicle (1) is determined by using image information determined by at least one environmental sensor (5) of the vehicle (1), wherein the environmental sensor is capable of acquiring images of at least a portion of the surrounding environment of the vehicle (1) from different perspectives. At least one correction information is determined by using the at least one environmental information. The at least one distance information is corrected using the at least one correction information. The following method is used to verify whether one or more of the GNSS satellite signals (2) are reflected GNSS satellite signals (2) from GNSS satellites (3) that do not have a direct line of sight: detect surrounding buildings in the acquired image, determine the azimuth and elevation angles of the GNSS satellite (3), and perform ray tracing based on the detected surrounding buildings and the determined azimuth and elevation angles to identify when the GNSS satellite signal (2) is received from a GNSS satellite with a direct line of sight and when the GNSS satellite signal (2) is received from a GNSS satellite without a direct line of sight.

2. The method according to claim 1, wherein the at least one environmental sensor (5) includes at least one camera.

3. The method according to claim 1 or 2, wherein the at least one environmental information is determined by using a motion recovery structure method.

4. The method according to claim 1 or 2, wherein the at least one environmental information includes at least one spatial distance (6) to an object (7) in the surrounding area of ​​the vehicle (1).

5. The method according to claim 1 or 2, wherein the at least one correction information describes a measure of at least one reflection of the distance information attributable to the associated GNSS satellite signal (2).

6. A computer program product having instructions, which, when executed by a computer, cause the computer to perform the method according to any one of claims 1 to 5.

7. A machine-readable storage medium having a computer program stored thereon, the computer program being used, when run by a computer, to cause the computer to perform the method according to any one of claims 1 to 5.

8. A positioning device (8) for a vehicle (1), configured to perform the method according to any one of claims 1 to 5.

Images