Method for GNSS-Based Localization of a Vehicle

Abstract

The disclosure relates to a method for GNSS-based localization of a vehicle, comprising at least the following steps: a) receiving GNSS-satellite signals from GNSS satellites and determining at least one item of distance information about the distance between the vehicle and the GNSS satellite emitting the relevant GNSS-satellite signal, b) determining at least one item of environmental information about the environment around the vehicle using image information determined using at least one environment sensor of the vehicle, which is capable of capturing images of at least part of the environment around the vehicle from different perspectives, c) determining at least one item of correction information using the at least one environmental information item, and d) correcting the at least one distance information item by means of the at least one correction information item.

Claims

1. A method for GNSS-based localization of a vehicle, the method comprising: a) receiving GNSS satellite signals from GNSS satellites and determining at least one distance information item about a distance between the vehicle and a GNSS satellite of the GNSS satellites that emits a respective GNSS satellite signal of the GNSS satellite signals; b) determining at least one environmental information item about an environment around the vehicle using image information determined using at least one environment sensor of the vehicle that is configured to capture images of at least part of the environment around the vehicle from different perspectives; c) determining at least one correction information item using the at least one environmental information item; and d) correcting the at least one distance information item using the at least one correction information item.

2. The method according to claim 1 further comprising: carrying out a test to determine whether at least one of the GNSS satellite signals are reflected GNSS satellite signals of the GNSS satellites to which no direct line of sight exists.

3. The method according to claim 1, wherein the at least one environment sensor includes at least one camera.

4. The method according to claim 1, the determining at least one environmental information item further comprising: determining the at least one environmental information item using a structure-from-motion method.

5. The method according to claim 1, wherein the at least one environment information item includes at least one spatial distance to an object in the environment around the vehicle.

6. The method according to claim 1, wherein the at least one correction information item describes a measure for a component of the at least one distance information item that is attributable to at least one reflection of the respective GNSS satellite signal.

7. The method according to claim 1, wherein the method is carried out by a computer program.

8. A non-transitory machine-readable storage medium that stores a computer program for GNSS-based localization of a vehicle, the computer program, when executed by a computer, causing the computer to: a) receive GNSS satellite signals from GNSS satellites and determine at least one distance information item about a distance between the vehicle and a GNSS satellite of the GNSS satellites that emits a respective GNSS satellite signal of the GNSS satellite signals; b) determine at least one environmental information item about an environment around the vehicle using image information determined using at least one environment sensor of the vehicle that is configured to capture images of at least part of the environment around the vehicle from different perspectives; c) determine at least one correction information item using the at least one environmental information item; and d) correct the at least one distance information item using the at least one correction information item.

9. A localization device for a vehicle, the localization device being configured to: a) receive GNSS satellite signals from GNSS satellites and determine at least one distance information item about a distance between the vehicle and a GNSS satellite of the GNSS satellites that emits a respective GNSS satellite signal of the GNSS satellite signals; b) determine at least one environmental information item about an environment around the vehicle using image information determined using at least one environment sensor of the vehicle that is configured to capture images of at least part of the environment around the vehicle from different perspectives; c) determine at least one correction information item using the at least one environmental information item; and d) correct the at least one distance information item using the at least one correction information item.

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0026] The solution presented here as well as its technical background will be explained in more detail below on the basis of the figures. It should be noted that the disclosure is not intended to be limited by the exemplary embodiments. In particular, unless explicitly indicated otherwise, it is also possible to extract partial aspects of the facts explained in the figures and to combine them with other components and/or information from other figures and/or the present description. In the schematic and exemplary drawings:

[0027] FIG. 1: shows a flowchart of the method described here,

[0028] FIG. 2: shows one possible application of the described method, viewed in the direction of a street canyon,

[0029] FIG. 3: shows the possible application from FIG. 2 with a view of the street canyon,

[0030] FIG. 4: shows a possible application of the described method with multiple reflections, and

[0031] FIG. 5: shows a vehicle with a localization device as described here.

DETAILED DESCRIPTION

[0032] FIG. 1 shows a schematic and exemplary flowchart of the method described here. The method is used for GNSS-based localization of a vehicle 1 (cf. FIG. 2). The sequence of steps a), b) and c) shown by the blocks 110, 120 and 130 is exemplary and can thus be used in a normal operating sequence.

[0033] In block 110, according to step a), GNSS-satellite signals 2 are received from GNSS satellites 3 and at least one item of distance information is obtained about the distance 4 between the vehicle 1 and the GNSS satellite 3 transmitting the relevant GNSS satellite signal 2. In block 120, according to step b) at least one item of environmental information about the environment around the vehicle 1 is determined by using image information determined using at least one environment sensor 5 of the vehicle 1, which is capable of capturing images of at least part of the environment around the vehicle 1 from different perspectives. In block 130, according to step c), at least one item of correction information is determined using the at least one environment information item. In block 140, according to step d) the at least one distance information item is corrected by means of the at least one correction information item.

[0034] The global navigation satellite system, or GNSS, is a system for positioning and navigation. GNSS is a collective term for various satellite systems, including NAVSTAR GPS, GLONASS, Galileo and Beidou. To determine position, the satellites communicate their exact position (elevation and azimuth) and clock time via radio codes. In the receiver, the pseudo-ranges are obtained, i.e. the distances between satellite and receiver, determined from the signal propagation times inclusive of clock errors between satellite and receiver. If four or more satellites receive at the same time, the clock error can be compensated and the current position of the receiver determined.

[0035] The at least one environment sensor 5 can comprise at least one camera. In addition, the at least one item of environment information can be determined using a so-called structure-from-motion method. It is thus possible to specify a particularly advantageous approach for preferably highly automated driving that will advantageously increase the accuracy of GNSS-based localization in urban environments. The pseudo-range of reflected satellite signals measured from the receiver to the satellite can advantageously be corrected by an additional value which is determined using the at least one camera and the structure-from-motion 3D reconstruction method.

[0036] FIG. 2 shows a schematic and exemplary drawing of a possible application of the described method, viewed in the direction of a street canyon. The street canyon is illustrated here in an exemplary way by two objects 7 in the form of buildings or high-rise buildings.

[0037] In the method a test can (firstly) be carried out as to whether one or more of the GNSS-satellite signals 2 are reflected GNSS-satellite signals 2 from GNSS satellites 3 to which no direct line of sight exists. In other words, this can also be described in particular by stating that a decision is (firstly) made as to which pseudo-ranges the correction will be applied to. In particular, it is checked which of the received satellite signal or signals is/are actually reflected signals from satellites to which there is no direct line of sight. In principle, different approaches can be applied to this problem. For example, reflected satellite signals can be detected at a low carrier-to-noise ratio (C/NO) and/or a significant pseudo-range residual. In particular in the case of a moving vehicle, these values can also be observed over time. Thus, jumps in the C/NO and/or residuals (e.g. C/NO becomes smaller, residual becomes larger) can also indicate newly reflected signals.

[0038] Alternatively or in addition, NLOS satellites (i.e. GNSS satellites to which there is no direct line of sight) can also be identified (directly) by means of cameras (on the vehicle). For this purpose, surrounding buildings can be detected in the camera images, for example using the image processing method of semantic segmentation. Using known azimuth and elevation angles of the satellites, for example, ray tracing can be performed to detect whether a direct line of sight to the corresponding satellite is available.

[0039] In particular if the NLOS satellites with pseudo-ranges to be corrected are known, the overestimated distance (due to the reflection) can be advantageously calculated as a correction value in a following step. For this purpose, FIG. 2 shows an example of a section perpendicular to the direction of a street canyon. Here, the vehicle 1 receives a reflected signal 2 of an NLOS satellite 3, which is under the elevation angle θ. Due to the reflection, the position of the vehicle 1 is wrongly assumed to be on the left-hand side of the building 7 shown on the left in FIG. 2.

[0040] Here, ε.sub.v is the overestimated distance of the pseudo-range. Due to the right-angled triangle, this value can be calculated to ε.sub.v=2d*cos(θ), with d being the unknown distance to the building wall. If, for example, the vehicle 1 is equipped with cameras as environment sensors 5 which detect the building wall, this distance can be determined in a particularly advantageous way using the structure-from-motion 3D reconstruction method. For this purpose, pixels and/or objects from the moving vehicle 1 are viewed from different angles with time offsets. The distance d to the objects 7, in this case to the building wall, can then be determined using triangulation.

[0041] In this context, the distance d is an example of the fact that, and possibly how, the at least one environmental information item can comprise at least one spatial distance 6 to an object 7 in the vicinity of the vehicle 1. Furthermore, the overestimated distance ε.sub.v of the pseudo-range is an example of the fact that, and possibly how, the at least one correction information item (here, for example, ε.sub.v) describes a measure for a component of the distance information (here pseudo-range) that can be attributed to at least one reflection of the relevant (NLOS) GNSS-satellite signal 2.

[0042] FIG. 3 shows a schematic and exemplary drawing of the possible application of FIG. 2, with a view of the street canyon. The illustration according to FIG. 3 explains an example of how the correction value ε.sub.v explained in connection with FIG. 2, in particular for the 2-dimensional case, can be supplemented by the missing component for the 3-dimensional case. For this purpose, in FIG. 3 the situation is considered from a perspective which is indicated in FIG. 2 by the arrow 9 at the top left (plan view of a plane inclined with the elevation angle θ through the satellite signal 2). Here also, the (new) correction value ε can be calculated by means of a right-angled triangle, this time with the already determined value ε.sub.v as the hypotenuse. For the 3-dimensional case, the final correction value evaluates to ε=ε.sub.v*|sin(β)|=2d*cos(θ)*|sin(β)|. The angle θ here is the difference between the azimuth angle of the satellite 3 and the direction of the street canyon. The correction value for ε thus determined represents a particularly preferred correction information item in the sense of the method described here, which is determined by means of the captured environment information d (spatial distance 6).

[0043] Finally, in order to obtain the new pseudo-range (actual distance 4 between the vehicle 1 and the GNSS satellite 3), here the correction value ε is subtracted from the originally measured pseudo-range (distance information from step a)). This is an example of the correction of the at least one distance information item by means of the at least one correction information item.

[0044] FIG. 4 shows a schematic and exemplary drawing of a possible application of the described method in the case of multiple reflection. In other words, FIG. 4 illustrates that, and possibly how, the described method can also be advantageously used for the correction of multiple reflections. For example, in the case of a double reflection, for example, the error based on the distance d.sub.2 to the building 7 at which the second reflection occurs (here the building 7 shown on the right) can also be taken into account.

[0045] The correction value therefore evaluates advantageously to ε.sub.m=ε.sub.1+ε.sub.2=2d.sub.1*cos(θ)*|sin(β)|+2d.sub.2*cos(θ)*|sin(β)|=2(d.sub.1+d.sub.2)*cos(θ)*|sin(β)|, where ε.sub.1 denotes the correction value associated with the first reflection, ε.sub.2 the correction value associated with the second reflection, d.sub.1 the distance to the building wall associated with the first reflection, and d.sub.2 the distance to the building wall associated with the second reflection. The correction value for ε.sub.m thus determined represents a particularly preferred correction information item in the sense of the method described here, which is determined by means of the captured environment information d.sub.1 and d.sub.2 (spatial distances 6).

[0046] To identify the presence of a double reflection, surrounding building heights can (also) be determined in an advantageous way by structure-from-motion. A double reflection is identified in particular when the building wall associated with the first reflection is higher than h.sub.min,1=(d.sub.1+2d.sub.2)*tan(θ)/sin(β) and (simultaneously) the height h.sub.2 of the building wall associated with the second reflection has a minimum height of h.sub.min,2=d.sub.2*tan(θ)/|sin(β)| and a maximum height of h.sub.max,2=(2d.sub.1+3d.sub.2)*tan(θ)/|sin(β)|(or h.sub.min,2<h.sub.2<h.sub.max,2).

[0047] However, double or multiple reflections do not necessarily need to be corrected in this way (i.e. with the correction value ε.sub.m). This is because even with the more general procedure described here for correcting single reflections, improvements in the pseudo-range and thus also in the positional accuracy can already be achieved.

[0048] FIG. 5 shows a schematic and exemplary drawing of a vehicle 1 having a localization device 8 described here. The localization device 8 is configured to carry out the method described here. The localization device 8 can obtain data from an environment sensor 5 of the vehicle 1.

[0049] The described method allows the compensation or correction of multipath effects to be improved in an advantageous way, in particular in the NLOS case.