Method and system for lane detection

Abstract

Method (500) for detecting a lane (82) for lateral guidance of a vehicle (100), the lateral guidance of the vehicle (100) based on a road model (412), the method (500) comprising: Determine (504) of one or more characteristics (404), wherein one or several features (404) are suitable to influence the detection of the lane (82); and Detection (506) of a lane (82) based on a sensor system (406) of the vehicle (100); characterized in that the method further comprises: Determining a weighting of one or more elements (408; 86, 88) of the lane (82) detected by the sensors (406) of the vehicle (100); and Determining (508) the road model (412) based on the recorded lane (82), on the determined one or more features (404) and on the determined weighting.

Description

[0001] The disclosure relates to systems and methods for lane detection. In particular, the disclosure relates to systems and methods for lane detection for the purpose of lateral guidance of automated vehicles. State of the art

[0002] Lane keeping systems and procedures for vehicles, particularly automated vehicles, are known in the prior art. These systems and procedures use optical sensors, such as cameras, to detect lanes based on road markings and structural elements. They are designed to issue a warning or even actively intervene in the steering if the vehicle threatens to leave its currently occupied lane. The term "vehicle" includes cars, trucks, buses, motorhomes, motorcycles, and similar vehicles used for the transport of people, goods, etc. In particular, the term includes motor vehicles used for passenger transport.

[0003] Within the context of this document, the term "automated driving" can refer to driving with automated longitudinal or lateral control, or autonomous driving with automated longitudinal and lateral control. Automated driving can, for example, involve extended periods of driving on the highway or time-limited driving during parking or maneuvering. The term "automated driving" encompasses automated driving at any level of automation. Examples of automation levels include assisted, partially automated, highly automated, and fully automated driving. These levels of automation were defined by the Federal Highway Research Institute (BASt) (see BASt publication "Research Compact," issue 11 / 2012). In assisted driving, the driver continuously performs longitudinal or lateral control, while the system takes over the other function within certain limits.In partially automated driving (TAF), the system takes over longitudinal and lateral control for a certain period and / or in specific situations, while the driver must continuously monitor the system, as with assisted driving. In highly automated driving (HAF), the system takes over longitudinal and lateral control for a certain period without the driver needing to continuously monitor the system; however, the driver must be able to take over control of the vehicle within a certain timeframe. In fully automated driving (VAF), the system can automatically handle driving in all situations for a specific use case; no driver is required for this use case. The aforementioned levels of automation correspond to SAE Levels 1 to 4 of the SAE J3016 standard (SAE - Society of Automotive Engineering). For example, highly automated driving (HAF) corresponds to Level 3 of the SAE J3016 standard.Furthermore, SAE J3016 specifies SAE Level 5 as the highest level of automation, which is not included in the BASt definition. SAE Level 5 corresponds to driverless driving, where the system can automatically handle all situations like a human driver throughout the entire journey; a driver is generally no longer required.

[0004] Driver assistance functions typically require information about the current course of drivable lanes (e.g., the vehicle's own lane, or lanes to the left or right of it) in order to enable lateral guidance (also known as "lane guidance") of the vehicle.

[0005] Under certain circumstances, for example poor visibility or weather conditions, when lane markings overlap, are completely or partially missing or poorly visible, or in the case of special road layouts (e.g. crest, depression, tight curves), problems may occur in lane recognition, with corresponding consequences especially for the lateral guidance of a vehicle.

[0006] Typical examples of such cases are entrances and exits, or road widenings and narrowings, where additional lane markings are sometimes present and lead away from the actual driving lane (also called "ego lanes"). Depending on the quality of the detected or recorded lane markings, this can lead to a different course for the generated route guidance curve. To counteract such problems, navigation data is used, among other things, to interpolate the further course of the road.

[0007] For example, German patent application DE 10 2013 220 487 A1 discloses a lane detection system comprising an optical lane detector and a GNSS-based position detector (GNSS - Global Navigation Satellite System). The GNSS-based position detector is configured to determine lane information from digital map data based on a detected position, a known position detection accuracy, and known accuracy limits of the digital map data, and to transmit this lane information to the optical lane detector. The optical lane detector, in turn, is configured to incorporate the lane information from the position detector into its lane detection process.The system is essentially designed to determine an interpolated lane profile based on very accurate map data and corresponding position information and to transmit it to the optical lane recognition system so that it can perform lane recognition for the entire section of the route even in poor visibility conditions with only partially recognized lane markings.

[0008] US Patent 8,498,782 B2 discloses a lane-keeping system that, in addition to information from optical and distance sensors, also evaluates information from a GNSS (Global Navigation Satellite System). The GNSS information is used to determine an approximate position of the vehicle on the roadway. The precise vehicle position is then determined using further information from the optical and distance sensors. In other words, the position is first approximated and then refined.

[0009] From German patent application DE 10 2013 003 216 A1, a method for determining a lane for steering control of an automated vehicle is known. A lane is determined by a first system based on acquired environmental data, and a second lane is determined by a second system. Starting from a current position, the second lane can be determined independently of the first lane using a digital map and data obtained through odometry. Process data calculated by the first and second systems for determining the lanes are continuously compared, and if one of the two systems fails, the lane is determined solely based on the remaining, functioning system.According to the described procedure, a comparison means that, based on historically determined data, an assessment is made about the quality of the currently determined data from both systems. Based on this, new, accuracy-optimized data is determined from the current data, which then serves as the basis for further calculations for both systems. As long as both systems are functional, the process data is compared so that the signals issued by both systems from a lane controller for controlling actuators are essentially identical. This comparison ensures that, before one system fails, both systems calculate a consistent target lane position. If one system fails, lane planning can continue without any changes in the target position using either system.The procedure focuses on a continuous comparison of the two systems, so that in the event of a failure of one of the systems, the steering control can continue without jumps in the target value.

[0010] From German patent application DE 10 2013 105 046 A1, a method for selecting a target lane in a vehicle lane guidance system is known, comprising: predicting lane information for a vehicle using digital map data; predicting lane information for a vehicle using trajectory data of a preceding vehicle; obtaining at least left lane marking data and / or right lane marking data from an image system; performing a lane curvature fusion calculation using the digital map data, the trajectory data of the preceding vehicle, and the lane marking data from the image system, wherein the lane curvature fusion calculation is performed on a microprocessor in the vehicle lane guidance system; and comparing the lane marking data from the image system with the digital map data and the trajectory data of the preceding vehicle.and selecting a target lane for lane guidance, based on comparing the lane marking data from the image system with the digital map data and the trajectory data of the vehicle ahead.

[0011] From the publication DE 10 2010 062 129 A1, a method for providing a virtual lane boundary is known, wherein the lane represents an area of ​​a carriageway, and wherein the method comprises a step of reading at least two carriageway marking profiles in front of or next to the vehicle from a detection device; a step of recognizing a carriageway widening using partial areas of a distance between the two read carriageway marking profiles; and a step of determining the virtual lane boundary for the area of ​​the recognized carriageway widening, wherein the virtual lane boundary is determined for the side of the carriageway that has the more rapidly changing carriageway marking profile.

[0012] From the publication DE 10 2017 005 921 A1, a method for the autonomous detection of a lane profile of at least one lane is known, comprising at least a first method step in which at least two lane boundaries are detected by means of at least one environmental sensor, at least a second method step in which the lane boundary profiles of the detected lane boundaries are described at least approximately by means of an arc model by means of a computing unit, and at least a third method step in which the lane boundary profiles are compared by means of the computing unit, whereby a parallelism of the detected lane boundaries is determined on the basis of the curvatures of the lane boundary profiles.

[0013] Known systems and methods can have one or more of the following disadvantages. Some systems rely partially or completely on the visibility of lane markings or features characteristic of lanes. Optical sensors can only reliably detect lane markings if they are also visible to the human eye. Therefore, road surface contamination, worn road markings, poor lighting conditions, and weather conditions, in particular, can prevent the lane markings from being reliably detected and may result in the underlying system or method being limited or completely inoperable in certain situations.

[0014] One disadvantage of GNSS-supported lane keeping systems is that GNSS-based positioning is often too imprecise to provide a sufficiently accurate position for a lane keeping system. The GNSS accuracy of navigation systems currently available in production vehicles is typically insufficient for a lane keeping system, meaning that systems or methods based on it can only support optical systems to a limited extent. This is further compounded by the problem that production navigation systems often rely on map data with insufficient accuracy for a lane keeping system. In short, GNSS data often cannot determine the vehicle's position within a lane with sufficient accuracy, necessitating greater accuracy in both the positioning and the map data, for example, with a resolution in the range of a few centimeters.

[0015] Due to the aforementioned limitations and shortcomings, lane detection errors can occur, particularly in certain situations, for example, when incorrectly detected or incorrect lane markings are used for the road model and lane detection. This can happen even in good visibility conditions, especially at highway entrances and exits, widening or narrowing sections (i.e., changes in the number of available lanes), intersections with multiple lane markings for different directions of travel, underpasses or overpasses, crests or dips, and in construction zones. Construction zones, in particular, can be problematic in this regard due to the often-present altered traffic patterns and other factors.Typically, road markings in construction zones are modified in addition to existing markings, usually distinguished by color (e.g., yellow versus white). Furthermore, construction zones often feature modified signage, sometimes supplemented by existing signage that may be temporarily obscured. Special regulations must also be observed (e.g., altered lane widths, road closures, shortened entrances and exits, entrances and exits for special vehicles), and / or increased dirt and grime are to be expected.

[0016] As a result, unintended or incorrect steering inputs can be triggered, particularly if a vehicle cannot reliably detect its lane position and / or maintain accurate positioning, or if the lane has only been determined inaccurately for a short period and a correction is then required within a relatively short distance. Such steering inputs can lead to comparatively jerky or misleading changes of direction, especially to so-called "steering error." This can ultimately lead to dangerous driving situations and / or situations that negatively affect effective or perceived driving behavior. Disclosure of the invention

[0017] It is an objective of the present disclosure to provide lane detection systems and methods that avoid one or more of the aforementioned disadvantages and / or enable one or more of the described advantages.

[0018] This problem is solved by the subject matter of the independent claims. Advantageous embodiments are specified in the dependent claims.

[0019] According to the present invention, a method for detecting a lane for lateral guidance of a vehicle is provided, the lateral guidance of the vehicle being based on a road model. The method comprises determining one or more features, wherein the one or more features are suitable for influencing lane detection; detecting a lane based on vehicle sensors; determining the weighting of one or more lane elements detected by the vehicle sensors; and determining the road model based on the detected lane, the determined one or more features, and the determined weighting.

[0020] In one aspect, the procedure also includes receiving navigation data. The determination of one or more features is based on this navigation data.

[0021] In one aspect, navigation data includes one or more of the following: map data, in particular Advanced Driver Assistance Systems (ADAS) data; data based on the collective driving behavior of a large number of vehicles; and data based on the detection of traffic signs, in particular the optical detection of traffic signs.

[0022] In one aspect, if the one or more features indicate the presence of an on-ramp or off-ramp on a road segment traveled by the vehicle, determining the road model includes: weighting elements present on one side of the lane facing away from the on-ramp or off-ramp with a first factor; and weighting elements present on one side of the lane facing the on-ramp or off-ramp with a second factor; where the first factor indicates a higher weighting than the second factor.

[0023] In one aspect, if the one or more features indicate the presence of a widening or narrowing on a road segment traveled by the vehicle, and the widening or narrowing relates to a beginning or ending lane, determining the road model includes: weighting elements present on one side of the lane facing away from the beginning or ending lane with a first factor; and weighting elements present on one side of the lane facing the beginning or ending lane with a second factor; where the first factor indicates a higher weighting than the second factor.

[0024] In one aspect, the first and second factors are configured to optionally display a weighting in the range of 0% to 100%, preferably where the first factor displays a weighting of 100% and where the second factor displays a weighting of 0%.

[0025] In one aspect, the procedure includes further lateral guidance of the vehicle based on the determined road model.

[0026] In one aspect according to embodiments of the present disclosure, a system for detecting a lane for lateral guidance of a vehicle is specified. The system comprises a control unit configured to execute the method according to the embodiments described herein.

[0027] In one aspect according to embodiments of the present disclosure, a vehicle is specified. The vehicle comprises the system according to the embodiments described herein. Preferably, the vehicle comprises means for semi-autonomous or autonomous control of the vehicle, optionally wherein the means for semi-autonomous or autonomous control of the vehicle are configured to perform the lateral guidance of the vehicle.

[0028] According to embodiments of the present disclosure, lane detection is improved, particularly in the area of ​​construction sites, widenings / narrowings and entrances / exits, by intelligently fusion of additional navigation data, for example map data, to recognize the respective situation in advance and then, depending on the situation, only certain information or sensor data are used for lane detection, or a special weighting of such information or data is carried out to generate a robust model of the road.

[0029] In particular, according to embodiments of the present disclosure, steering error is effectively prevented and potentially dangerous situations are effectively avoided. Furthermore, the objective driving behavior and the subjective sense of safety of vehicle occupants are improved. Brief description of the drawings

[0030] Exemplary embodiments of the disclosure are shown in the figures and are described in more detail below. Unless otherwise noted, the same reference symbols are used for identical and equivalent elements. Fig. 1 shows a schematic top view of a first driving situation of a vehicle according to embodiments of the present disclosure; Fig. 2 shows a schematic top view of a second driving situation of a vehicle according to embodiments of the present disclosure; Fig. 3 shows a schematic top view of a third driving situation of a vehicle according to embodiments of the present disclosure; Fig. 4 shows a block diagram of a system according to embodiments of the present disclosure; and Fig. Figure 5 shows a flowchart of a method according to embodiments of the present disclosure. Implementations of the revelation

[0031] Fig. Figure 1 shows a schematic top view of a first driving situation of a vehicle 100 according to embodiments of the present disclosure. In this first exemplary driving situation, the vehicle 100 is on a road with two lanes 81, 82 in the direction of travel (e.g., a motorway or a road with structurally separated carriageways).

[0032] Typically, lateral guidance requires information about the road model of the current lane (here, lane 82) and, if available, the left (here, lane 81) and right (not available) lanes. The road model is usually derived primarily from the recorded lane markings or structural features, the determined lane centerline, and information from the navigation data. To avoid an incorrect road model, the road model or the determination of the route guidance curve must be adjusted in the situations described as problematic.

[0033] In the situation shown, vehicle 100 is on lane 82 and, starting at times t0, t1, and t2, passes through sections 66, 64, and 66. Vehicle 100 detects lane marking 88, which separates lanes 81 and 82, and lane boundaries 86, which laterally delimit the drivable road area. The drivable area may also be delimited by the shoulder or a green strip 87, whereby vehicle 100 does not usually travel in the area between lane boundary 86 and boundary 87.

[0034] In section 64, a third lane 83 appears to be added to lanes 81 and 82, separated from lane 82 by a road marking 88. However, this third lane 83 is merely an exit and not an additional lane following the further course of the road. Designated as road marking 84, this is an example of a road marking that shows signs of wear and tear and is therefore difficult or impossible for vehicle 100 to detect. In some cases, a road marking is completely missing in the area of ​​road marking 84, with a similar effect on detection.

[0035] In the exemplary first driving situation shown, an additional lane 83 is supposedly added to the right side of vehicle 100, in addition to the lane 82 already being driven on by vehicle 100. By default, vehicle 100 detects its own lane 82 based on the detection of the road marking 88 in area 88a (see section 66 below). Fig. 1) and based on the detection of the lane boundary 86 in area 86a (see also section 66 below in Fig. 1) as boundaries of its own lane 82. Based on this detection, the vehicle 100 determines the road model and a target guidance curve (or straight line) 68. This runs essentially in the middle of lane 82 as planned.

[0036] In the transition between area 66 and area 64 (in the "direction of travel" from bottom to top in Fig. 1) As already described, it is possible that the lane boundary 86 is detected as the right-hand boundary of the driver's own lane 82, for example, due to a poorly visible or non-existent road marking 84. It should be noted that in Germany, road markings at entrances and exits are usually continuous, while in other countries (e.g., the USA) it is common for the road marking to initially disappear (e.g., in the area of ​​lane marking 84) and only reappear after a certain distance. Furthermore, the road marking may also be difficult to see due to dirt or wear. In such situations, a conventional lateral guidance system will attempt to find a right-hand lane boundary and, in doing so, will detect the lane boundary 86 (here, for example, in the area 86i) as such.

[0037] As a result of detecting the lane boundary 86 as the right edge of lane 82, vehicle 100 now determines a target guidance curve 68' that deviates from the actually correct target guidance curve 68. Based on the target guidance curve 68', which, as in Fig. As shown in Figure 1, the vehicle 100 deviates significantly to the right from the correct target guidance curve 68 within a very short distance. This lateral guidance induces a steering movement towards the exit or towards the supposed lane 83, exemplified by vehicle 100'. This process is generally referred to as "steering error," whereby vehicle 100' incorrectly changes lanes or at least begins to change lanes (see time t). xBased on the recorded target guidance curve 68', the longitudinal guidance can also induce a braking torque, or further guidance or assistance functions can be triggered or influenced. As already described, this entails some disadvantages.

[0038] According to embodiments of the present disclosure, the driving behavior represented by vehicle 100' is effectively prevented.

[0039] Navigation data can be used to identify a range of road features, such as the presence of an exit. Other road features include widening or narrowing of the road, intersections, forks and mergings, overpasses and underpasses, and crests or dips. Based on these identified road features, certain situations can be anticipated and the road model generation adjusted accordingly.

[0040] Navigation data includes, for example, map data available in the vehicle or online (e.g., Advanced Driver Assistance Systems (ADAS) data), data from backend structures (e.g., backend servers), data based on collective driving behavior (e.g., evaluations of GPS data from other vehicles), or data captured by the vehicle (e.g., recognition of traffic signs).

[0041] In the Fig. In the case shown in Figure 1, the presence of an exit is determined, for example, based on map data (e.g., ADAS), and the road model is adjusted as follows. Within a predetermined area 64, which essentially comprises the exit area, possibly including transition zones before and after the exit (in some embodiments in the range of 50 m to 300 m), the determination of the navigation curve 68 is adjusted by changing the weighting of the detected road features. Initially, or as planned, the determination of the navigation curve 68 takes place in area 66 (see Figure 1). Fig. 1 below) based on the left road marking 88 (see area 88a) and the right lane boundary 86 (see area 86a). In the following area 64 in the direction of travel (see Fig. In the case of the center (1), the road model is then primarily or exclusively oriented towards the left road marking 88 (see area 88a), while the potentially problematic right lane boundary 86 (see area 86i) and / or the road marking 84 are only considered secondarily or not at all. An example weighting could prioritize only the left road marking 88 in area 88a, assigning it a weight of 100%, and completely ignore the right lane boundary 86, corresponding to a weight of 0%. Depending on the situation, other weighting models or weightings are possible, enabling a desired guidance curve 68. Additionally, further processing of the determined guidance curve can be performed, for example, including smoothing or plausibility checks, to obtain an optimized guidance curve and / or to filter out erroneous points.

[0042] From a temporal perspective, the control process essentially unfolds as follows. At time t0, the planned lateral guidance occurs without any special restrictions or adjustments. At time t1, based on navigation data (see above), an event is generated indicating the presence of an exit at a specific distance (e.g., within a range of up to 500 m, preferably within a range of up to 300 m). At or from time t1 onwards, the road model or the determination of the route guidance curve 68 can then be temporarily adjusted as described. The feature triggering the event (here: exit) can also have a range, such that a second event, occurring later in the time sequence and indicating the end of the feature's presence, can be generated at time t1. Alternatively, the second event can also be generated at a later time (e.g., time t2 or even later, depending on the feature's range).Following the second event, the planned lateral guidance can then be resumed without restrictions or adjustments (for example, at time t2).

[0043] In some embodiments, a rule-based adaptation of the road model or the determination of the navigation curve can be implemented. In the present case, the presence of an exit area can, based on rules, lead to only the elements of the road markings or lane boundaries facing away from the exit being used to determine the navigation curve (see previous paragraph).

[0044] The presence of an entrance ramp is to be treated essentially the same as the described case of an exit ramp, so that the described concepts and procedures are applicable analogously.

[0045] Fig. Figure 2 shows a schematic top view of a second driving situation of a vehicle 100 according to embodiments of the present disclosure. In this second exemplary driving situation, the vehicle 100 is on a roadway with initially two lanes 82, 83 in the direction of travel (e.g., a motorway or a road with structurally separated carriageways), which then widen to three lanes 81, 82, 83.

[0046] As in the first example driving situation, information about the road model of the current lane (here lane 82) and, if applicable, the left (later lane 81) and right lanes (here lane 83) is required for lateral guidance. The road model is typically derived primarily from the detected lane markings or structural features, the determined lane centerline, and information from the navigation data. To avoid an incorrect road model, the road model or the determination of the target guidance curve must also be adjusted in this driving situation.

[0047] Widenings and narrowings behave similarly to on-ramps and off-ramps. However, one difference is that, for example, when widening from two to three lanes, it is not defined on which side of the roadway the additional lane will be created. To determine which road markings, or whether and, if so, which side of the lane, should be prioritized, heuristics can be used to evaluate and compare the quality of the lane markings. Various influencing factors can be considered, such as the stability of a marking (e.g., similarity to previously observed markings, taking odometry data into account), parallelism to other markings, marking type (e.g., white, yellow), and so on. Each quality characteristic can be assigned a weight, and, based on the heuristic used, the marking with a predetermined weight (e.g.,the highest weight) or exceeds it (e.g., is greater than a minimum weight).

[0048] In the situation shown, vehicle 100 is on lane 82 and passes through sections 66, 64, and 66, starting at times t0, t1, and t2. Vehicle 100 detects lane marking 88, which separates lanes 82 and 83, and lane boundaries 86, which laterally limit the drivable road area.

[0049] In section 64, a third lane, 81, is added to lanes 82 and 83. At least in its later stages, this third lane, 81, is also separated from lane 82 by a road marking, 88. Unlike the first driving situation, this third lane, 81, is an additional lane that follows the further course of the road. Furthermore, unlike the first driving situation, a road marking separating lanes 81 and 82 is completely absent for large parts of section 64, making it impossible to reliably determine the left lane boundary in this section.

[0050] In the exemplary second driving situation shown, an additional lane 81 is added to the lane 82 already in use by vehicle 100 on its left side. By default, vehicle 100 detects its own lane 82 based on the detection of the road marking 88 in area 88a (see section 66 below). Fig. 2) and based on the detection of the lane boundary 86 in area 86a (see also section 66 below in Fig. 2) as boundaries of its own lane 82. Based on this detection, the vehicle 100 determines the road model and a target guidance curve (or straight line) 68. This runs essentially in the middle of lane 82 as planned.

[0051] In the transition between area 66 and area 64 (in the "direction of travel" from bottom to top in Fig. 2) As already described, it is possible that the lane boundary 86 is perceived as the left boundary of one's own lane 82, for example due to a poorly perceptible or, as in Fig. 2 shown, a non-existent road marking that would separate the resulting lane 81 from lane 82.

[0052] As a result of detecting the lane boundary 86 as the left edge of lane 82, the vehicle 100 now determines a target guidance curve 68' that deviates from the actually correct target guidance curve 68, essentially analogous to the first driving situation already described. Based on the target guidance curve 68', which, as in Fig. As shown in Figure 2, if the vehicle 100 deviates significantly to the left from the correct target guidance curve 68 within a very short distance, the lateral guidance of the vehicle 100 induces a steering movement towards lane 81, exemplified by vehicle 100'. Again, a "steering error" can occur, in which vehicle 100' incorrectly changes lanes or at least begins to change lanes (see time t). xBased on the recorded target guidance curve 68', the longitudinal guidance can also induce a braking torque or further guidance or assistance functions can be triggered or influenced.

[0053] According to embodiments of the present disclosure, the driving behavior represented by vehicle 100' is also effectively prevented in the event of a widening of the roadway.

[0054] In the Fig. In the second case shown, the presence of a road widening is again determined, for example, based on map data (e.g., ADAS). Alternatively, this can also be done by detecting traffic signs that indicate a road widening. The road model is then adjusted again, as described below.

[0055] In a predetermined area 64, which essentially comprises the widening area, possibly including transition areas before and after the widening, the determination of the guidance curve 68 is adjusted by changing the weighting of recorded road characteristics. Initially, or as planned, the determination of the guidance curve 68 takes place in area 66 (see Fig. 2 below) based on the right-hand road marking 88 (see area 88a) and the left-hand lane boundary 86 (see area 86a). In the following area 64 in the direction of travel (see Fig. 2 (center) the road model then focuses primarily or exclusively on the right-hand road marking 88 (see area 88a), while the potentially problematic left-hand lane boundary 86 (see area 86i) is only considered secondarily or not at all. An example weighting could prioritize only the right-hand road marking 88 in area 88a, assigning it a weight of 100%, and completely ignore the left-hand lane boundary 86, corresponding to a weight of 0%. Depending on the situation, other weighting models or weightings are possible, enabling a desired guidance curve 68. As already mentioned, further processing of the determined guidance curve is also possible.

[0056] From a temporal perspective, the control sequence is essentially as follows. At time t0, the planned lateral guidance occurs without any special restrictions or adjustments. At time t1, based on navigation data (see above), an event is generated indicating the presence of a widening at a certain distance (e.g., within a range of up to 200 m, preferably within a range of up to 50 m). At or from time t1 onwards, the road model or the determination of the target guidance curve 68 can then be temporarily adjusted as described, and subsequently, for example at time t2, the planned lateral guidance resumes without restrictions or adjustments.

[0057] In some embodiments, a rule-based adaptation of the road model or the determination of the guidance curve can be implemented. In the present case, the presence of a widening can, based on rules, lead to only the elements of the road markings or lane boundaries facing away from the widening being used to determine the guidance curve.

[0058] In some embodiments of the present disclosure, the lane centerline can be ignored (weighting zero), with the lane markings remaining equally weighted. The lane centerline (HPP), estimated, for example, based on camera signals, ideally lies midway between two lane markings (e.g., 86, 88; cf. 68 in the Fig. 1, Fig. 2, Fig. 3) In the event of an incorrect detection of the HPP (e.g. along 68'), it should be discarded accordingly.

[0059] Fig. Figure 3 shows a schematic top view of a third driving situation of a vehicle 100 according to embodiments of the present disclosure. In this third exemplary driving situation, the vehicle 100 is on a roadway with initially three lanes 81, 82, 83 in the direction of travel (e.g., a motorway or a road with structurally separated carriageways), which then narrow to two lanes 82, 83.

[0060] As in the first and second example driving situations, information about the road model of the current lane (here lane 82) and, if applicable, the left (initially lane 81) and right lanes (here lane 83) is required for lateral guidance. The road model is typically derived primarily from the detected lane markings or structural features, the determined lane centerline, and information from the navigation data. To avoid an incorrect road model, the road model or the determination of the target guidance curve must also be adjusted in this third driving situation.

[0061] In the situation shown, vehicle 100 is on lane 82 and passes through sections 66, 64, and 66, starting at times t0, t1, and t2. Vehicle 100 detects a left lane marking 88, which separates lanes 81 and 82, and a right lane marking 88, which separates lanes 82 and 83.

[0062] In section 64, lane 81 ends, having initially been separated from lane 82 by a road marking 88. Unlike the first driving situation, in the area where lane 81 ends, a road marking separating lanes 81 and 82 is missing for a large portion of section 64, making it impossible to reliably determine the left lane boundary in this section.

[0063] In the exemplary third driving situation shown, the additional lane 81 on the left side of lane 82, which is being driven on by vehicle 100, ends. By default, vehicle 100 detects its own lane 82 based on the detection of the left-hand road marking 88 in the left-hand area 88a (see section 66 below). Fig. 3) and based on the detection of the right-hand road marking 88 in the right-hand area 88a (see also section 66 below in Fig. 3) as boundaries of its own lane 82. Based on this detection, the vehicle 100 determines the road model and a target guidance curve (or straight line) 68. This runs essentially in the middle of lane 82 as planned.

[0064] In the transition between area 66 and area 64 (in the "direction of travel" from bottom to top in Fig. 3) As already described, it is possible that the lane boundary 86 is perceived as the left boundary of one's own lane 82, for example due to a poorly perceptible or, as in Fig. Figure 3 shows a non-existent road marking that would separate the ending lane 81 from lane 82.

[0065] As a result of detecting the lane boundary 86 as the left edge of lane 82, the vehicle 100 now determines a target guidance curve 68' that deviates from the actually correct target guidance curve 68, essentially analogous to the first and second driving situations already described. Based on the target guidance curve 68', which, as in Fig. As shown in Figure 3, the vehicle 100, within a very short distance, deviates significantly from the actually correct target guidance curve 68 to the left and then returns to it. This lateral guidance induces a steering movement towards lane 81 and then back to lane 82, exemplified by vehicle 100'. Again, a "steering error" can occur, in which vehicle 100' incorrectly changes lanes or at least begins to change lanes (see time t). xIn the third driving situation, the situation is further exacerbated by the fact that the leftward steering maneuver must be corrected shortly afterward by steering back, which can result in a further load transfer and potentially cause the vehicle to oscillate 100'. Based on the detected target guidance curve 68', the longitudinal guidance system can also induce a braking torque, or other guidance or assistance functions can be triggered or influenced. Possibly in combination with the steering and / or oscillation of the vehicle 100', this can create dangerous driving maneuvers.

[0066] According to embodiments of the present disclosure, the driving behavior represented by vehicle 100' is also effectively prevented in the event of a narrowing of the roadway.

[0067] In the Fig. In the case described in section 3, the presence of a lane widening is again determined, for example, based on map data (e.g., ADAS). Alternatively, this can also be done by detecting traffic signs that indicate a narrowing of the roadway or by using collective vehicle movement data that shows vehicles changing lanes from the left lane 81 to the middle lane 82 in the relevant area. The road model is then adjusted as described below.

[0068] In a predetermined area 64, which essentially comprises the area of ​​the narrowing, possibly including transition areas before and after the narrowing, the determination of the guidance curve 68 is adjusted by changing the weighting of the recorded road characteristics. Initially, or as planned, the determination of the guidance curve 68 takes place in area 66 (see Fig. 3 below) based on the left road marking 88 (see left area 88a) and the right road marking 88 (see right area 88a). In the area 64 following in the direction of travel (see Fig. 3 (center) The road model then focuses primarily or exclusively on the right-hand road marking 88 (see area 88a), while the potentially problematic left-hand lane boundary 86 (see area 86i) is only considered secondarily or not at all. For example, one weighting could prioritize only the right-hand road marking 88 in area 88a, assigning it a weight of 100%, and completely ignore the left-hand lane boundary 86, corresponding to a weight of 0%. Depending on the situation, other weighting models or weightings are possible, enabling a desired guidance curve 68. As mentioned previously, further processing of the determined guidance curve is also possible.

[0069] From a temporal perspective, the control sequence is essentially as follows. At time t0, the planned lateral guidance occurs without any special restrictions or adjustments. At time t1, based on navigation data (see above), an event is generated indicating the presence of a narrowing at a certain distance (e.g., within a range of up to 200 m, preferably within a range of up to 50 m). At or from time t1 onwards, the road model or the determination of the target guidance curve 68 can then be temporarily adjusted as described, and subsequently, for example at time t2, the planned lateral guidance resumes without restrictions or adjustments.

[0070] In some embodiments, a rule-based adaptation of the road model or the determination of the guidance curve can be implemented. In the present case, the presence of a narrowing can, based on rules, lead to only the elements of the road markings or lane boundaries facing away from the narrowing being used to determine the guidance curve.

[0071] Fig. Figure 4 shows a block diagram of a system 400 according to embodiments of the present disclosure. As with respect to the Fig. As described in Figures 1 to 3, the system 400 generates one or more events 404 based on navigation data 402, indicating special driving situations. Based on the vehicle 100's sensors 406, elements 408 required for the (lateral) guidance of the vehicle 100, such as lane boundaries, are determined. This can be done, for example, by optically detecting road or lane boundaries. Subsequently, the events 404 and the elements 408 required for the (lateral) guidance of the vehicle 100 are selected, for example, based on a rule-based approach, and fused in a road model 412. The (lateral) guidance of the vehicle then takes place based on the fused road model, whereby various assistance systems (see Figures 422, 424, 426, and others 428) can also access the fused road model.

[0072] Fig.Figure 5 shows a flowchart of a method 500 for detecting a lane 82 for lateral guidance of a vehicle 100 according to embodiments of the present disclosure. The method 500 begins in step 501. Optionally, navigation data is received in step 502. The received navigation data can be based on map data (in particular ADAS data), as described above, and / or on other data sources (see above). In step 504, one or more features 404 are determined, optionally at least partially based on the received navigation data. The determined one or more features 404 are suitable for influencing the detection of the lane 82, as described above. Examples of features are areas of entrances, on-ramps, and exits, construction zones, intersections, and the like (see above).The features can preferably be suitable to influence the detection of lane 82 insofar as detection is made more difficult or prevented in the area of ​​these features, particularly by altered, changing, or otherwise modified road markings 86, 88. This preferably includes partial impediment or prevention. In step 506, a lane 82 is detected based on a sensor 406 of the vehicle 100. In this step, a known method of lane detection can be used, for example, based on one or more optical sensors (e.g., cameras). In step 508, the road model 412 is determined based on the detected lane 82 and the determined one or more features 404. As described in detail, in this step, specifically detected road markings can be made higher or lower.The weighting is reduced to avoid or completely prevent the aforementioned disadvantages of potentially erroneous lane detection. Optionally, in step 510, the vehicle 100 is guided laterally based on the determined road model 412. The procedure 500 ends in step 512.

[0073] Although the invention has been further illustrated and explained in detail by means of preferred embodiments, the invention is not limited by the disclosed examples, and other variations can be derived from them by a person skilled in the art without departing from the scope of protection of the invention. It is therefore clear that a multitude of possible variations exist. It is also clear that the embodiments mentioned as examples are truly only examples and are not to be understood in any way as limiting, for example, the scope of protection, the possible applications, or the configuration of the invention.Rather, the preceding description and the description of the figures enable the person skilled in the art to implement the exemplary embodiments in concrete terms, whereby the person skilled in the art, with knowledge of the disclosed inventive concept, can make various changes, for example with regard to the function or the arrangement of individual elements mentioned in an exemplary embodiment, without leaving the scope of protection defined by the claims and their legal equivalents, such as further explanations in the description.

Claims

[1] Method (500) for detecting a lane (82) for lateral guidance of a vehicle (100), the lateral guidance of the vehicle (100) based on a road model (412), the method (500) comprising: Determine (504) of one or more characteristics (404), wherein one or several features (404) are suitable to influence the detection of the lane (82); and Detecting (506) a lane (82) based on sensors (406) of the vehicle (100); characterized by , that the procedure further includes: Determining a weighting of one or more elements (408; 86, 88) of the lane (82) detected by the sensors (406) of the vehicle (100); and Determining (508) the road model (412) based on the recorded lane (82), on the determined one or more features (404) and on the determined weighting. [2] Method according to the preceding claim, further comprising receiving (502) navigation data (402); wherein determining (504) one or more features (404) based on the navigation data (402) is carried out. [3] Method according to the preceding claim, wherein the navigation data (402) comprise one or more of the following data: - Map data, in particular Advanced Driver Assistance Systems (ADAS) data; - Data based on the collective driving behavior of a large number of vehicles; and - Data based on the recording of traffic signs, in particular the optical recording of traffic signs. [4] A method according to any of the preceding claims, further comprising determining a target guidance curve (68), the determination of the target guidance curve (68) preferably comprising smoothing the target guidance curve and / or performing a plausibility check of the target guidance curve (68), wherein the smoothing is configured to obtain an optimized shape of the target guidance curve (68) and wherein the plausibility check is configured to filter out one or more erroneous points of the target guidance curve (68); and Lateral and / or longitudinal guidance of the vehicle (100) based on the target guidance curve (68). [5] Method according to any of the preceding claims, wherein, if the one or more features (404) indicate the presence of an entrance or exit on a road section traveled by the vehicle (100), determining (508) the road model (412) comprises: Weights of elements (86, 88) located on one side of the carriageway opposite the entrance or exit (82) with a first factor; and Weights of elements (86, 88) located on one side of the lane facing the entrance or exit (82) with a second factor; wherein the first factor indicates a higher weighting than the second factor. [6] Method according to any of the preceding claims, wherein, if the one or more features (404) indicate the presence of a widening or narrowing on a road section traveled by the vehicle (100), the widening or narrowing relating to a beginning or ending lane (81), determining (508) the road model (412) comprises: Weights of elements (86, 88) located on the side of the lane (82) opposite the beginning or ending lane (81) with a first factor; and Weights of elements (86, 88) located on the side of the lane (82) facing one of the beginning or ending lanes (81) with a second factor; where the first factor indicates a higher weighting than the second factor. [7] Method according to one of the two preceding claims, wherein the first and second factors are configured to optionally display a weighting in the range of 0% to 100%, preferably wherein the first factor displays a weighting of 100% and wherein the second factor displays a weighting of 0%. [8] Method according to one of the preceding claims, further comprising lateral guidance (510) of the vehicle (100) based on the determined road model (412). [9] System for detecting a lane (82) for lateral guidance of a vehicle (100), comprising a control unit (120), wherein the control unit (120) is configured to perform the method according to one of the preceding claims. [10] Vehicle (100) comprising the system according to the preceding claim, optionally wherein the vehicle (100) comprises means for semi-autonomous or autonomous control of the vehicle, optionally wherein the means for semi-autonomous or autonomous control of the vehicle are configured to perform the lateral guidance of the vehicle (100).

Images