Method for locating a vehicle and localization module

Abstract

Method (400) for locating a vehicle (100), wherein the vehicle (100) has a localization module (102) for performing at least one localization procedure for locating the vehicle (100), wherein the method (400) comprises the following steps: Reading (410) a landmark density value (108) representing a density of landmarks (106) in the area of ​​a track section (104) assigned to the vehicle (100); Assigning (420) a first part of a computing capacity of the localization module (102) to an odometry-based method as a first localization method and a second part of the computing capacity to a map-matching-based method as a second localization method different from the first localization method using the landmark density value (108); and Executing (430) the first localization procedure using the first part of the computing capacity and / or the second localization procedure using the second part of the computing capacity to locate the vehicle (100).

Description

State of the art

[0001] The invention relates to a device or a method according to the preamble of the independent claims. The present invention also relates to a computer program.

[0002] For example, a motor vehicle can be equipped with a system for scenario-dependent switching between different driver assistance functions.

[0003] Furthermore, the methods described in the paper "M. Hernandez, B. Ristic, Alfonso Farina, L. Timmoneri, "A comparison of two Crame'r-Rao bounds for nonlinear filtering with Pd<1," Signal Processing, IEEE Transactions on, vol. 52, no. 9, pp. 2361-2370, 2004" and the paper "D. Nister, O. Naroditsky, J. Bergen, "Visual odometry," in Computer Vision and Pattern Recognition, 2004. CVPR 2004. Proceedings of the 2004 IEEE Computer Society Conference on, vol.1, pp. 652-659, 2004" can be used.

[0004] From DE 10 2014 002 821 A1 a method and a system for localizing a mobile device are known, wherein the device has several sensors for detecting the environment of the device using different localization methods.

[0005] US Patent 2011 / 0054791A1 discloses a landmark-based method for determining the position of a vehicle, wherein the vehicle has an in-vehicle database in which the landmarks and their respective locations are stored, and wherein, upon detection of such a landmark, the corresponding location replaces a location estimated by in-vehicle sensors. Disclosure of the invention

[0006] Against this background, the approach presented here introduces a method for locating a vehicle, a localization module that uses this method, and finally a corresponding computer program according to the main claims. Advantageous further developments and improvements of the device specified in the independent claim are possible through the measures listed in the dependent claims.

[0007] A method for locating a vehicle is presented, wherein the vehicle has a localization module for executing at least one localization procedure for locating the vehicle, the method comprising the following steps: Reading in a landmark density value, which represents the density of landmarks in the area of ​​a route segment assigned to the vehicle; Allocating a first portion of the localization module's computing capacity to a first localization method and / or a second portion of the computing capacity to a second localization method that differs from the first localization method, using the landmark density value; and Executing the first localization procedure using the first part and / or the second localization procedure using the second part to locate the vehicle.

[0008] A vehicle can be understood as a motor vehicle such as a car, truck, or motorcycle. Landmarks can be static objects that can be repeatedly detected. For example, a landmark can be a distinctive point within the vehicle's assigned section of road or within the vehicle's surroundings, such as a traffic light, a road sign, or a bollard. The vehicle might, for example, be equipped with an environmental sensor to detect such landmarks. Landmark density can be understood as the number of landmarks per unit area. For example, the landmark density value can represent a current or expected landmark density. The landmark density value can, for example, be read from a central server via an interface.The section of track in question could be, for example, a section of track currently being travelled by the vehicle or a section of track that the vehicle is expected to travel on.

[0009] The first and second parts of the computing capacity can be the same or different. In extreme cases, either the first or the second part can represent the entire computing capacity available for vehicle localization. For example, the first localization method might be based primarily on visual odometry, while the second localization method might be landmark-based, i.e., map-matching-based.

[0010] The approach presented here is based on the understanding that sufficient computing power can be managed to execute at least one localization procedure for determining a vehicle's geographic position based on the density of landmarks in the vehicle's vicinity. This allows for the implementation of a method and system for continuous vehicle localization with optimized utilization of computing power.

[0011] The approach described here enables, for example, the targeted dynamic allocation of computing capacity to an odometry- or landmark-based map-matching localization module, depending on the expected landmark density. This allows for demand-driven vehicle localization with continuous pose estimation. Furthermore, the overall system's resource consumption is significantly reduced, as multiple localization systems do not need to run in parallel at full capacity. Another advantage lies in the ability to use information about expected landmark densities for the proactive planning of resource allocation to visual odometry and landmark-based localization.

[0012] According to one embodiment, the method can include a step of comparing the landmark density value with a reference value to determine any deviation between the landmark density value and the reference value. In this allocation step, a larger portion of the computing resources can be allocated to the first localization method than to the second localization method if the comparison step reveals that the landmark density value is smaller than the reference value, or a larger portion of the computing resources can be allocated to the second localization method than to the first localization method if the comparison step reveals that the landmark density value is larger than the reference value. A reference value can, for example, be a landmark density value that just barely allows for sufficiently accurate localization using landmarks.This design allows resource consumption for the precise localization of the vehicle to be reduced to a minimum.

[0013] According to a further embodiment, in the assignment step, the first part can be assigned to an odometry-based method as the first localization method. Additionally or alternatively, in the assignment step, the second part can be assigned to a map-matching-based method as the second localization method. An odometry-based method can be understood as a method that enables the localization of the vehicle using vehicle data such as wheel speed, steering angle, or yaw rate. Map matching can be understood as a comparison of landmarks detected by the vehicle with landmarks stored in a digital map. Such map matching allows the vehicle to be located on the digital map. This embodiment enables the most accurate possible localization of the vehicle.

[0014] Furthermore, it is advantageous if, in the assignment step, at least one of the two parts is assigned using a fuzzy logic algorithm and, additionally or alternatively, a stochastic algorithm. This allows the computing power to be distributed between the two localization methods with the lowest possible resource consumption.

[0015] Furthermore, in the input step, a localization map encompassing the route segment, a detection signal representing at least one landmark detected using at least one of the vehicle's environmental sensors, or a sensor signal provided by at least one of the vehicle's sensors representing wheel speed, yaw rate, or steering angle can be read. Correspondingly, in the execution step, the first localization method, the second localization method, or both localization methods can be executed using the localization map, the detection signal, or the sensor signal. A localization map can be understood as a digital map. The localization map can, for example, be stored on a central server or transmitted to the vehicle via a suitable interface.The detection signal can be generated, for example, by the environmental sensor itself or by an evaluation unit coupled with the environmental sensor to process data from the sensor. For instance, the localization map and the detection signal can be used for localization via map matching when performing the second localization method. Similarly, the sensor signal can be used for localization via odometry when performing the first localization method. This allows the vehicle's geographic position to be determined with high accuracy and efficiency.

[0016] According to another embodiment, in the reading step, a value provided by a central server, such as the landmark density value, can be read in via an interface to the central server. The central server could, for example, be a backend server wirelessly connected to the vehicle. This allows the landmark density value to be provided centrally.

[0017] Furthermore, during the allocation step, at least a further portion of the computing capacity can be allocated to at least one additional localization method that differs from at least one of the two existing localization methods, using the landmark density value. In the execution step, this additional localization method can then be executed using this additional capacity to locate the vehicle. This further increases the efficiency and accuracy of vehicle localization.

[0018] This process can be implemented, for example, in software or hardware, or in a hybrid form of software and hardware, such as in a control unit.

[0019] The approach presented here further creates a localization module designed to execute, control, and implement the steps of a variant of the method presented here in appropriate devices. This embodiment of the invention, in the form of a localization module, also allows the underlying problem to be solved quickly and efficiently.

[0020] For this purpose, the localization module can have at least one processing unit for processing signals or data, at least one storage unit for storing signals or data, at least one interface to a sensor or actuator for reading sensor signals from the sensor or for outputting data or control signals to the actuator, and / or at least one communication interface for reading or outputting data embedded in a communication protocol. The processing unit can be, for example, a signal processor, a microcontroller, or the like, while the storage unit can be flash memory, an EPROM, or a magnetic storage device.The communication interface can be configured to read or output data wirelessly and / or via wired connections, whereby a communication interface that can read or output wired data can, for example, read this data electrically or optically from or output it into a corresponding data transmission line.

[0021] In this context, a localization module can be understood as an electrical device that processes sensor signals and outputs control and / or data signals accordingly. The localization module can have an interface, which may be implemented in hardware and / or software. In the case of a hardware-based interface, the interfaces can, for example, be part of a so-called system ASIC, which incorporates various functions of the device. However, it is also possible that the interfaces are separate integrated circuits or at least partially comprised of discrete components. In the case of a software-based interface, the interfaces can be software modules, which, for example, are located on a microcontroller alongside other software modules.

[0022] Also advantageous is a computer program product or computer program with program code that can be stored on a machine-readable carrier or storage medium such as a semiconductor memory, a hard disk memory or an optical memory and is used to carry out, implement and / or control the steps of the method according to one of the embodiments described above, in particular if the program product or program is executed on a computer or device.

[0023] Exemplary embodiments of the invention are shown in the drawings and explained in more detail in the following description. It shows: Fig. 1 a schematic representation of a vehicle with a localization module according to an exemplary embodiment; Fig. 2 a schematic representation of a vehicle made of Fig. 1; Fig. 3 a localization module according to an exemplary embodiment; and Fig. 4 a flowchart of a procedure according to an exemplary embodiment.

[0024] In the following description of favorable embodiments of the present invention, the same or similar reference numerals are used for the elements shown in the various figures and acting similarly, without repeating these elements.

[0025] Fig. Figure 1 shows a schematic representation of a vehicle 100 with a localization module 102 according to an exemplary embodiment. The vehicle 100 travels on a track section 104, here exemplified as a straight roadway. The track section 104 is lined with a plurality of landmarks 106. The localization module 102 is configured to receive a landmark density value 108 regarding the density of the landmarks 106 in the track section 104, here from a central server 110, for example, a backend server. The localization module 102 uses the landmark density value 108 to allocate the computing capacity available to the localization module 102 between two different localization methods for locating the vehicle 100.Here, the localization module 102 allocates a first portion of its computing capacity to the first of the two localization methods and a second portion to the second of the two localization methods, depending on the landmark density. The two localization methods are different. According to one embodiment, the first localization method is based on odometry, and the second localization method is based on landmarks, i.e., map matching.

[0026] To manage computing capacity particularly efficiently, the localization module 102 is designed, according to one embodiment, to compare the landmark density value 108 with a reference value and to allocate the computing capacity depending on the result of the comparison. Specifically, the localization module 102 allocates more computing capacity to the first localization method, here the odometry-based method, than to the second localization method if the landmark density value 108 is below the reference value. Conversely, it is possible for the localization module 102 to allocate more computing capacity to the second localization method, here the map-matching-based method, than to the first localization method if the landmark density value 108 is above or identical to the reference value.

[0027] Furthermore, the localization module 102 is designed to locate the vehicle 100 by executing at least one of the two localization methods using the portion of the computing capacity allocated to the respective localization method.

[0028] According to a further embodiment, the localization module 102 is configured to locate the vehicle 100 by executing at least one further localization method, which may differ from the two aforementioned localization methods. For this purpose, the localization module 102 allocates a corresponding additional portion of its computing capacity to the additional localization method using the landmark density value 108. The allocation of this additional portion can be carried out analogously to the allocation of the two portions for the first and second localization methods.

[0029] According to the in Fig. In the embodiment shown in Figure 1, the vehicle 100 is configured to detect the landmarks 106 using environmental sensors and to send corresponding data to the localization module 102. The localization module 102 is configured to locate the vehicle 100 using the data provided by the environmental sensors, as described in more detail below.

[0030] According to one embodiment, the landmark density value 108 represents information about upcoming landmark densities. The localization module 102 uses this information to allocate computing resources to odometry, landmark detection, or a global localization module. For example, the localization module 102 receives a message from server 110 stating "low landmark density approaching." In response, the localization module 102 increases the accuracy and computing resources for visual odometry while simultaneously reducing the computing resources for landmark detection.

[0031] Information from digital maps, provided via a backend server, enables semi- or highly automated vehicle systems to extend their planning horizon. Using this map information requires sufficiently accurate vehicle localization relative to the digital maps. Approaches commonly used for continuous localization, such as Kalman filtering, utilize information from map matching and odometry, i.e., the estimation of a vehicle pose relative to a previous pose, also known as state prediction. The process of detecting landmarks using vehicle sensors and comparing them with landmarks from a localization map to obtain a global pose estimate can also be referred to as pose fixation.Increasing the accuracy of either method can improve the overall result of continuous vehicle localization. Determining a pose fix requires that landmarks be detected by vehicle sensors and compared with landmarks from the localization map provided by the backend server. This map comparison yields a pose estimate at a specific point in time. Odometry is used to predict the vehicle's global pose at the next time step or to estimate it relative to a previous pose. Odometry is particularly useful for bridging areas where there are no or insufficient landmarks to determine a sufficiently accurate pose fix.However, the use of odometry may be limited to relatively short sections, for example, sections between 10 m and 100 m in length, depending on the required accuracy, because pose estimation can drift over time, thus increasing errors in pose estimation. Such drift can be reduced by more accurate odometry.

[0032] Both determining a pose fixation through landmark detection and map comparison, and odometry, can be computationally intensive. Furthermore, the computational effort can scale with the accuracy achievable by the respective method. The approach presented here enables optimized use of the available computing power.

[0033] For example, information about the expected landmark density is sent from a backend server to a vehicle system. The vehicle system can then be configured to optimally distribute computing power between odometry, such as that based on wheel speed or visual data, and landmark-based global localization. For instance, more computing power is allocated to landmark-based localization when the landmark density is high. Conversely, more computing power is allocated to odometry when the landmark density is low.

[0034] This ensures optimized use of computing capacity and positively influences the localization result.

[0035] A sufficiently accurate digital localization map is required for the global localization of vehicle systems. The localization map contains landmarks that can be used for localization.

[0036] In this case, vehicle 100 is equipped with an environmental perception module trained to detect landmarks 106 in the vicinity of vehicle 100. The detected landmarks 106 are then compared with the landmarks stored in the localization map. This process is called map matching.

[0037] In some urban or rural areas, there may not be enough landmarks to locate the vehicle, or their number may be too low for sufficiently accurate localization. In such areas with very low landmark density, sufficiently accurate localization can only be achieved using precise odometry. Odometry can generally be understood as a method for estimating relative poses from time step to time step. This can be done using wheel speed and steering angle measurements, as well as visual data, such as from a monocular camera installed in the vehicle. Visual odometry, in particular, can incur a significant computational load. For this reason, an odometry-based localization method should ideally only be performed with full accuracy in areas with very low landmark density.Information about the expected landmark density is transmitted from the backend map server to the vehicle system. The vehicle system then allocates computing resources based on the landmark density.

[0038] Depending on the specific implementation, computing capacity is allocated in different ways.

[0039] For example, computing power is allocated based on a fuzzy logic algorithm. In this case, odometry is assigned "high" computing power when landmark density is "low". The values ​​for "low" or "high", or for other such levels, can be determined empirically.

[0040] Another possibility is an assignment based on a stochastic approach. Here, a stochastic algorithm is used to calculate the expected accuracy of the pose fixation based on the anticipated landmark density. The computing power required for landmark detection scales linearly with the expected accuracy of the pose fixation. The advantage of such a stochastic algorithm lies in the low computational effort required. This allows for optimized use of computing power in vehicle localization, so that when using a suitable computing unit with consistent processing power, the localization accuracy is favorably affected.

[0041] An example of the applicability of the system according to the invention is described below.

[0042] For example, if a vehicle is located in a narrow residential street with parked cars along the sides, few or no landmarks are likely to be visible in such an environment. Therefore, precise odometry is required to locate the vehicle in this setting. If the residential street merges into an intersection with a main road, many landmarks such as lampposts or street signs are usually visible, enabling global, landmark-based localization. In both situations, this information is transmitted to the vehicle from a backend server and results in settings such as the following.

[0043] In the area of ​​the narrow residential street, little to no computing capacity is allocated to global, landmark-based localization, while odometry is allocated a large amount or all of the available computing capacity.

[0044] In contrast, in the intersection area of ​​global, landmark-based localization, a large or all of the available computing capacity is allocated, while odometry is allocated little to no computing capacity.

[0045] Fig. Figure 2 shows a schematic representation of a vehicle 100 made of Fig. 1. The vehicle 100 is equipped with an environment sensor 200, which is configured to detect the vehicle 100's surroundings. The environment sensor 200 is implemented, for example, as a camera, ultrasonic, radar, or lidar sensor. According to this embodiment, the environment sensor 200 is configured to send a corresponding detection signal 202 to the localization module 102. The localization module 102 is configured to locate the vehicle 100 using the detection signal 202, for example, in the context of a map-matching-based localization method. For this purpose, the localization module 102 receives, via an interface 204 to the central server 110, in addition to the landmark density value 108, for example, additional map information 206 from a digital localization map 207 stored on the central server 110, which contains the landmarks in the section of the route traveled by the vehicle 100.The landmarks 106 detected by the environmental sensor 200 are now compared with each other by the localization module 102 using the detection signal 202 and the map information 206. Based on this comparison, also called map matching, the vehicle 100 can be geographically located.

[0046] The localization module 102 is according to Fig. 2 optionally coupled with a sensor 208 of the vehicle 100, which is configured to acquire odometry data such as wheel speed, yaw rate, or steering angle of the vehicle 100 and to transmit a corresponding sensor signal 210 to the localization module 102. Depending on the embodiment, the sensor 208 can comprise a plurality of different sensor units for acquiring different odometry data. Accordingly, the localization module 102 is configured to locate the vehicle 100 using the sensor signal 210, for example, in the context of an odometry-based localization method.

[0047] Fig. Figure 3 shows a schematic representation of a localization module 102 according to an exemplary embodiment. The localization module 102 is, for example, a localization module as described above. Fig. 1 and Fig. As described in Figure 2, the localization module 102 comprises a read unit 310 for reading the landmark density value 108. The read unit 310 is coupled to an allocation unit 320, which is configured to divide the computing capacity of the localization module 102 between the first and second localization methods using the landmark density value 108, i.e., depending on the landmark density represented by the landmark density value 108. According to this embodiment, the allocation unit 320 is configured to send an execution signal 325 to an execution unit 330 of the localization module 102 in response to the allocation of computing capacity. The execution unit 330 is configured to execute the first localization method or the second localization method, respectively, using the portion of the computing capacity allocated to the respective localization method, based on the execution signal 325.Optionally, the execution unit 330 is configured to output a position signal 335, which represents a geographical position of the vehicle determined during the execution of the first or second localization procedure, for example to an interface to a control unit for controlling the vehicle.

[0048] Fig. Figure 4 shows a flowchart of an embodiment of method 400 for locating a vehicle. Method 400 can, for example, be used in conjunction with a previously performed procedure based on the Fig.The localization module described in sections 1 to 3 is executed or controlled. Procedure 400 comprises a step 410 in which the landmark density value is read. In a further step 420, using the landmark density value, a first portion of the computing capacity is allocated to the first localization procedure and a second portion to the second localization procedure. Finally, in a step 430, the first localization procedure is executed using the first portion of the computing capacity, and the second localization procedure is executed using the second portion of the computing capacity, in order to locate the vehicle.

[0049] Steps 410, 420, and 430 can be performed continuously to enable continuous vehicle localization.

[0050] If an embodiment includes an “and / or” connection between a first feature and a second feature, this is to be read as meaning that the embodiment according to one embodiment has both the first feature and the second feature, and according to another embodiment either only the first feature or only the second feature.

Claims

[1] Method (400) for locating a vehicle (100), wherein the vehicle (100) has a localization module (102) for performing at least one localization procedure for locating the vehicle (100), wherein the method (400) comprises the following steps: Reading (410) a landmark density value (108) representing a density of landmarks (106) in the area of ​​a track section (104) assigned to the vehicle (100); Assigning (420) a first part of a computing capacity of the localization module (102) to an odometry-based method as a first localization method and a second part of the computing capacity to a map-matching-based method as a second localization method different from the first localization method using the landmark density value (108); and Executing (430) the first localization procedure using the first part of the computing capacity and / or the second localization procedure using the second part of the computing capacity to locate the vehicle (100). [2] Method (400) according to claim 1, comprising a step of comparing the landmark density value (108) with a reference value to determine a deviation between the landmark density value (108) and the reference value, wherein in the allocation step (420) a larger portion of the computing capacity is allocated to the first localization method than to the second localization method if the comparison step reveals that the landmark density value (108) is smaller than the reference value, or a larger portion of the computing capacity is allocated to the second localization method than to the first localization method if the comparison step reveals that the landmark density value (108) is larger than the reference value. [3] Method (400) according to any of the preceding claims, wherein in the assignment step (420) the first part and / or the second part is assigned using a fuzzy logic algorithm and / or a stochastic algorithm. [4] Method (400) according to one of the preceding claims, wherein in the reading step (410) a localization map (207) encompassing the track section (104) and / or a detection signal (202) representing at least one landmark (106) detected using at least one environment sensor (200) of the vehicle (100), and / or a sensor signal (210) provided by at least one sensor (208) of the vehicle (100) representing a wheel speed and / or a yaw rate and / or a steering angle of the vehicle (100) is read in, wherein in the execution step (430) the first localization method and / or the second localization method is executed using the localization map (207) and / or the detection signal (202) and / or the sensor signal (210). [5] Method (400) according to one of the preceding claims, wherein in the reading step (410) a value provided by a central server (110) as the landmark density value (108) is read in via an interface (204) to the central server (110). [6] Method (400) according to one of the preceding claims, wherein in the allocation step (420) using the landmark density value (108) at least a further part of the computing capacity is allocated to at least one further localization method different from the first and / or second localization method, wherein in the execution step (430) the further localization method is executed using the further part to locate the vehicle (100). [7] Localization module (102) with units (310, 320, 330) configured to execute and / or control the method (400) according to any of the preceding claims. [8] Computer program configured to execute and / or control the method (400) according to any one of claims 1 to 6. [9] Machine-readable storage medium on which the computer program according to claim 8 is stored.

Images