Method for operating a driver assistance system for a motor vehicle, driver assistance system and motor vehicle

Abstract

Method for operating a driver assistance system (1) for a motor vehicle (2) by which environmentally caused wind turbulence (3) can be detected, comprising the steps: - Generating an environment model (4) that describes road users (6) in an environment (7) of the motor vehicle (2) using sensor data from at least one vehicle-integrated sensor of the driver assistance system (1); (VS1) - Determining a respective lane assignment, lane curvature (8) and vehicle orientation (9) for the motor vehicle (2) and the road users (6); (VS2) - Determining the relative orientation (10) between track curvature (8) and vehicle orientation (9) for the motor vehicle (2); (VS3) - Determining the relative orientation (10) between lane curvature (8) and vehicle orientation (9) for road users (6) and combining the relative orientations (10); (VS4) - Segmenting the environment (7) into segments (11) such that an assignment of the respective relative orientation (10) per segment (11) is made; (VS5) characterized by the steps: - Generating a frequency representation (12) of the orientations (10) of road users (6) per segment (11); (VS6) - Determining an orientation distribution (13) based on the frequency representation (12); (VS7) - Deriving a turbulence orientation (14) from the orientation distribution (13); (VS8) and - Outputting a signal depending on the turbulence orientation (14). (VS9)

Description

[0001] The invention relates to a method for operating a driver assistance system for a motor vehicle, by which environmentally induced wind turbulence can be detected, according to claim 1. Furthermore, the invention relates to a driver assistance system for carrying out the method and a corresponding motor vehicle according to claims 9 and 10.

[0002] Crosswinds and wind turbulence can lead to critical driving situations for motor vehicles, especially passenger cars and trucks, as they require the driver to make appropriate steering or counter-steering adjustments. Therefore, it would be desirable to detect potential turbulence and issue relevant warnings to the driver. Strong turbulence can occur particularly near bridges, in open areas, and / or when overtaking trucks.

[0003] US patent 2021 / 0 291 829 A1 discloses a method for controlling a vehicle, comprising: determining an object, a property of the object, and the strength and direction of the wind affecting the object, based on an image of the vehicle's surroundings captured by an image sensor; defining a risk region for the object based on the specified property and the specified wind strength and direction; and controlling the vehicle's speed and steering according to the defined risk region.

[0004] US Patent 5,315,868 A discloses a sensor arrangement for detecting the influence of crosswinds on the driving behavior of a motor vehicle, with pressure measuring points located on opposite sides of the vehicle to measure the air pressure prevailing in the respective area of ​​the vehicle body's outer skin. A differential pressure sensor allows for the detection of crosswind gusts.

[0005] The object of the present invention is to advantageously calculate turbulence or wind gusts individually and adaptively for the motor vehicle by means of a method, a driver assistance system or a motor vehicle, in order to be able to visualize local turbulence in particular.

[0006] This problem is solved according to the invention by the subject matter of the independent claims. Advantageous embodiments and further developments of the invention are specified in the dependent claims, as well as in the description and in the drawings.

[0007] A first aspect of the invention relates to a method for operating a driver assistance system for a motor vehicle, wherein the method is intended to make environmentally caused wind turbulence detectable or detectable.

[0008] The inventive method comprises the following steps: In a first step, the driver assistance system generates an environment model that describes road users in the vicinity of the vehicle, in addition to the vehicle itself. This model is generated using sensor data from at least one sensor or sensor device, particularly one on the vehicle. In a second step, the system determines the lane assignment, lane curvature, and vehicle orientation for both the vehicle and the road users, specifically at their current positions. In a third step, the system determines the relative orientation between the lane curvature and the vehicle orientation for the vehicle. Finally, in a fourth step, the system determines the relative orientation between the lane curvature and the vehicle orientation for the road users and combines or accumulates these relative orientations.

[0009] In a fifth step, the environment is segmented into segments, allowing for the assignment of relative orientations to each segment. In a sixth step, a frequency representation of the orientations of road users per segment is generated, where the frequency representation describes at least a summary of the orientations of the road users recorded in the respective segment. In a seventh step, an orientation distribution is determined based on the frequency representation; that is, the relative orientations are determined in relation to the road users in each segment. Finally, in an eighth step, a turbulence orientation is derived from the orientation distribution; that is, it is determined in which segment, and especially from which direction, turbulence is to be expected.Finally, in a ninth step, a signal is output depending on the turbulence orientation.

[0010] The motor vehicle in question is, in particular, a passenger car or truck. The driver assistance system is specifically designed to monitor the vehicle's journey using the control unit. At least one of the vehicle's own sensors is, in particular, an environmental sensor, such as radar. The environmental model is thus an interpretation or evaluation of the sensor data, for example, using machine learning methods or a suitable model; in the case of a camera sensor, for example, using computer vision. Object recognition is used to determine both the roadway or its course and the other road users.

[0011] Lane assignments involve assigning a lane or track to each road user and vehicle. The lane on which the road user (in this case, another vehicle) or the vehicle itself is located is defined by its curvature. Lane curvature describes the radius of curvature of the road or the respective lane. Vehicle orientation, particularly within a global coordinate system that also incorporates lane curvature, describes the orientation of the vehicle, specifically its position. This orientation can include, for example, a pose or be a part of one. Relative orientation describes any deviation of the vehicle's orientation from the lane curvature to its current position, thus representing a potential discrepancy between the two.Segmentation divides the environment into multiple areas, allowing for individual analysis and the determination of the orientation distribution for each segment based on a frequency representation. This enables the determination of the turbulence levels for different segments. Subsequently, a signal is output, which can be used as a warning signal or, for example, as a control signal for the driver assistance system or another driver assistance system.

[0012] In other words, the first step of the process involves generating the environment using a driver assistance system, detecting the ego-vehicle's surroundings based, for example, on radar cameras or similar devices. The second step involves recognizing the roadway, detecting the vehicle's lane assignment and lane curvature at each position. The third step calculates the relevant vehicle orientation, for example, using the delta direction vector, the inter-lane curvature, and the vehicle's orientation. The fourth step involves accumulating the relative orientation across all detected vehicles, analogous to step 3. The fifth step involves spatial segmentation, particularly through normalization, concentrating the resulting orientation into spatial segments.In the sixth step, a frequency evaluation is performed based on the segments, specifically by calculating a polar histogram for the orientation sequence of all objects or road users in the respective segment. In the seventh step, frequency clusters are detected based on the polar histogram by identifying clusters within it. In the eighth step, turbulence orientations are detected by determining the orientation from the location within the polar histogram. Finally, in the ninth step, a turbulence warning or the turbulence itself is visualized, for example, on a display element such as an infotainment system screen, which can be operated, for instance, by the driver assistance system's control unit.

[0013] The present invention thus provides an approach to detect wind turbulence based on sensor signals and to visualize it for a motor vehicle driver.

[0014] This allows for a warning to be issued, which is relevant for the driver, as strong turbulence can occur over bridges, open areas, and similar surfaces, especially in adverse weather conditions. Such turbulence can lead to a strong lateral impulse, which in turn can result in critical driving situations. This method allows the driver to be not only shown turbulence based on weather information or windsocks, but also to calculate and visualize it individually and adaptively for the ego-vehicle.

[0015] The method according to the invention offers the advantage of adaptive driver warning in critical weather conditions. Furthermore, it allows for individual consideration of the current traffic environment. Another advantage is that no human intervention is required, and the method can be performed independently of weather and / or lighting conditions. Initial calibration can be carried out at the factory where the vehicle is produced and supplied, or adaptively in the environment during use. A further advantage of the method is that it does not require an additional connection to an information system to retrieve, for example, map data. Moreover, it is independent of the sensor modality and can be implemented, for example, using radar, lidar, cameras, or similar devices. In addition, sensors already present in the vehicle can be advantageously used.Furthermore, the procedure can be repeated adaptively and / or cyclically, thus always providing an update of the turbulence.

[0016] In an advantageous embodiment of the invention, a radar, lidar, and / or a camera are used as the sensor, or as at least one in-vehicle sensor. In other words, the method is independent of a specific sensor modality and can generate the environmental model based on various sensor data, such as images and / or 3D point clouds. This offers the advantage that the method can be carried out independently of sensor data and thus, for example, independent of weather conditions.

[0017] In a further advantageous embodiment of the invention, the signal is output as a control signal for a driver assistance system, in particular the driver assistance system for carrying out the method, and / or another driver assistance system. Additionally or alternatively, the signal is output as a visualization of the turbulence and / or as a warning signal, which can be, for example, acoustic, haptic, visual, and / or olfactory. In other words, the signal can be provided as a control signal for the driver assistance system or another driver assistance system, particularly if the vehicle is, for example, operating in semi-autonomous mode, so that, for example, lateral steering can be adjusted by the corresponding driver assistance system. Additionally or alternatively, the driver can advantageously be alerted to the turbulence by means of the visualization and / or the warning signal.This results in the advantage that the vehicle is controlled particularly advantageously and / or can be controlled by a driver.

[0018] In a further advantageous embodiment of the invention, a speed and / or acceleration for the motor vehicle and / or for the road users are additionally used to determine the relative orientation. Additionally or alternatively, a lane normalization is performed for the road users. In other words, the difference between lane curvature and vehicle orientation is determined based on a speed and / or acceleration of the motor vehicle or the respective road user. Additionally or alternatively, lane normalization is performed to avoid false positives in the lane curvature calculation, which can be achieved, for example, by subtracting the direction vectors of the lane curvature and the orientation of the respective vehicle.If the vectors are similar, the resulting vector is as small as possible, allowing for spatial normalization since the sensors detect objects at different distances, which can then be factored out. This offers the advantage that the method can determine a particularly advantageous relative orientation of the road users, thus enabling a highly precise determination of turbulence.

[0019] In an advantageous embodiment of the invention, a polar histogram is used for the frequency representation, thus determining an orientation sequence of the road users or all road users in the respective segment. Within a given area or segment, or for each segment, the vehicles are then identified using direction vectors. This offers the advantage of enabling a meaningful frequency representation in a particularly simple manner. This, in turn, allows the method to be carried out with exceptional efficiency.

[0020] In a further advantageous embodiment of the method, the orientation distribution is determined based on frequency clusters in the polar histogram. For example, frequent changes in the relative orientation, i.e., the positioning or vehicle orientation to the track curvature, can be decisive for determining the turbulence orientation. This results in the advantage that the method can be carried out particularly efficiently.

[0021] In a further advantageous embodiment of the invention, the segments are divided according to their distance from the vehicle. In other words, segmentation can be carried out depending on the distance to the vehicle, particularly in an area in front of the vehicle. For example, a first segment can be created for a distance of 5 to 50 meters from the vehicle, a second segment for a distance of 50 to 100 meters, and so on. This offers the advantage that a particularly efficient output can subsequently be generated when visualizing the turbulence, since the driver knows at what distance from the vehicle turbulence to expect.

[0022] In a further advantageous embodiment of the invention, static objects along a roadway are detected and taken into account when deriving the turbulence orientation. In other words, the environment model detects static objects, such as noise barriers and / or bridges, and their influence, for example due to a wind shadow, is considered. This offers the advantage that the turbulence orientation can be determined with particular precision.

[0023] A second aspect of the invention relates to a driver assistance system for a motor vehicle comprising a control unit and at least one sensor, which is configured to perform a method according to one of the preceding claims.

[0024] Advantageous embodiments, further developments and advantages of the first aspect of the invention are to be regarded as advantageous embodiments, further developments and advantages of the second aspect of the invention and vice versa.

[0025] A third aspect of the invention relates to a motor vehicle equipped with a driver assistance system according to the second aspect of the invention and / or equipped to carry out a method according to the first aspect of the invention.

[0026] Advantageous embodiments, further developments and advantages of the first and second aspects of the invention are to be regarded as advantageous embodiments, further developments and advantages of the third aspect of the invention and vice versa.

[0027] It shows: Fig. 1. Schematic flowchart for a procedure for operating a driver assistance system by which environmentally caused wind turbulence can be detected; Fig. 2. Schematic environmental model of the environment around the motor vehicle during a second step of the procedure; Fig. 3. Schematic environment model with vectors for a third step of the procedure; Fig. 4. Schematic environment model for a fourth step of the procedure; Fig. 5. Schematic representation of a segmented environment model for a fifth step of the procedure; Fig. 6. Schematic polar histogram for a sixth step of the procedure; Fig. 7. Schematic polar histogram for a seventh step of the procedure; Fig. 8 Schematic representation of a visualization for a ninth step of the procedure.

[0028] In the figures, identical or functionally equivalent elements are provided with the same reference symbols.

[0029] Fig. Figure 1 shows a schematic flowchart for a procedure for operating a driver assistance system 1, which can perform the presented procedure steps VS1 to VS9 of the procedure, for a motor vehicle 2, whereby the procedure can detect environmentally caused wind turbulence 3.

[0030] The procedure will also be used to present a corresponding driver assistance system 1 for a motor vehicle 2, which is designed to perform the procedure presented here, particularly in combination with at least one vehicle-integrated sensor. Similarly, a motor vehicle 2 will be presented which has said driver assistance system 1 and / or is also designed to perform the procedure presented here.

[0031] The process includes the following steps: In a first step VS1, an environment model 4 is generated, which describes road users 6 on a roadway 5 in an environment 7 of the motor vehicle 2. The generation of the environment model 4 is carried out using sensor data from at least one vehicle-integrated sensor, in particular by a control unit of the driver assistance system 1.

[0032] In a second step VS2, a respective lane assignment, a lane curvature 8 and a vehicle orientation 9 are determined for the motor vehicle 2 and the road users 6.

[0033] In a third step VS3, the relative orientation 10, between track curvature 8 and vehicle orientation 9, is determined for the motor vehicle 2.

[0034] In a fourth step VS4, the relative orientation 10 is determined between track curvature 8 and vehicle orientation 9 for the road users 6, and the relative orientations 10 are combined or accumulated.

[0035] In a fifth step VS5, the environment 7 is segmented into segments 11, so that an assignment of the respective relative orientation 10 per segment 11 takes place.

[0036] In a sixth step VS6, a frequency representation is generated of 12 orientations of 10 road users per segment 11.

[0037] In a seventh step VS7, an orientation distribution 13 is determined based on the frequency representation 12.

[0038] In an eighth step VS8, a turbulence orientation 14 is derived from the orientation distribution 13.

[0039] In a ninth step VS9, a signal is output depending on the turbulence orientation 14.

[0040] The motor vehicle 2, which may be a passenger car or a truck, is hereinafter also referred to as the ego vehicle. The other road users 6 are, in particular, other motor vehicles or motorized road users. The at least one vehicle-integrated sensor used to store or acquire the sensor data for the environment model 4 may be a lidar sensor, a radar sensor and / or a camera sensor or similar; the method is, in particular, independent of the specific sensor modality.

[0041] The signal can be output as a control signal for driver assistance system 1 and / or another driver assistance system. Furthermore, the signal can be used, as described in the Fig. 8 shown, output as visualization 15, additionally or alternatively also as a warning signal, for example acoustic, optical, haptic and / or olfactory.

[0042] Fig. Figure 2 shows the environment model 4 during the second step VS2 of the procedure, in which a lane assignment is made for the roadway 5 based on the sensor data for the road users 6. Furthermore, a lane curvature 8 and a respective vehicle orientation 9 are determined.

[0043] The environment generation, which was carried out for environment model 4, is based on the detection of the ego vehicle's surroundings using vision, radar, etc. During detection, lane assignments and lane curvatures are also determined at the corresponding current position.

[0044] Fig. Figure 3 shows a section of the environment model from Fig. 2 with the two road users 6 shown there, whereby a determination of a relative orientation 10 between track curvature 8 and vehicle orientation 9 has already been carried out both for the motor vehicle 2, according to procedure step VS3, and additionally for the road users 6, also a relative orientation 10 has been determined in each case according to procedure step VS4.

[0045] Thus, the relative orientation 10 is determined, in particular by calculating a direction vector, between the track curvature 8 and the vehicle orientation 9. In addition, in the process step VS4, the relative orientations 10 are accumulated over all driving objects, i.e., the road users 6 detected by the driver assistance system 1 on the basis of the environment model 4.

[0046] Fig. Figure 5 schematically shows the environment 7 with the ego vehicle or motor vehicle 2 and the road users 6, each of which is already assigned to a segment 11 and a lane normalization has already been carried out for the road users 6.

[0047] As an interim conclusion at this stage of the procedure, it can be summarized that the environment model 4 records the road users 6, whereby, for example, especially for determining the respective relative orientation 10, a speed or acceleration for the respective motor vehicle 2 or the road users 6 can also be recorded. Furthermore, the environment model 4 specifies the lane assignment and the respective lane curvature 8.

[0048] If the relevant data is available, a lane normalization is performed, particularly to avoid false positive results in case of deviations from the lane curvature 8 or the relative orientation 10. This is done, for example, by subtracting the direction vectors of the lane curvature 8 and the orientation 10 of the respective vehicle (motor vehicle 2 or road user 6). If the vectors are similar, the resulting vector is as small as possible.

[0049] Subsequently, spatial normalization is performed, since the sensors typically detect or record the road users 6 at different distances from motor vehicle 2, resulting in a representation according to Fig. This spatial normalization can lead to a result of 5. This spatial normalization is achieved in particular by storing the previously derived direction vectors along with their distance from the ego vehicle. An array of dimension n*r can then be constructed, where n equals the number of road users (6) and r equals the number of direction vectors detected over a distance, for example, a maximum detection range of 200 meters. This results in a 2-dimensional array containing the direction vector values ​​for each surrounding vehicle or for each road user (6).

[0050] Corresponding calculations are repeated for each of the segments 11. Through normalization of the segments 11, a concentration of the resulting orientation 10 can be achieved in the spatial segment 11, whereby the segments 11 can thus represent distance-dependent sub-areas of the environment 7. For example, the area between 50 and 100 meters from the ego-vehicle can be evaluated. In this area, the direction vector values ​​are considered for all vehicles or road users 6, and it is assumed that the change in the direction vector corresponds to a clustering in a polar histogram 16, which is significant as soon as all road users 6 in the vicinity exhibit these changes. A separate polar histogram 16 is also created for each segment 11.

[0051] In particular, as a representation variant for the frequency representation 12, only the orientation 10 of the road users 6 per time is determined, and this is advantageously presented as a polar histogram 16, as in the Fig. 6 and Fig. Figure 7 shows that if all road users 6 travel along the road orientation, meaning their respective relative orientation 10 corresponds almost exactly to the curvature of the lane 8, then no significant change can be observed, resulting in a relatively large even distribution. The response of the respective polar histogram 16 allows us to determine, firstly, whether a large change in swarm direction is present, and secondly, in which direction this change tends.

[0052] This shows Fig. Procedure step VS6 essentially generates a frequency representation of the relative orientation 10 of the road users 6 per segment 11 by means of a frequency evaluation based on the segments 11, whereby a polar histogram 16 is calculated for an orientation sequence of all objects or road users 6 in the respective segment 11. Thus, in particular, procedure steps VS6 and VS8 allow for the determination of clusters and their orientation from the polar distributions of the direction vectors according to procedure steps VS3 and VS4 for segment 11.

[0053] Fig. Figure 7 thus shows the seventh process step VS7 when using a polar histogram 16, where the detection of frequency clusters reflecting the orientation distribution 13 is indicated, and the turbulence orientation 14 can be determined from this. The detection of the turbulence orientation 14 thus builds upon process step VS7, whereby the orientations 10 can be detected from a localization within the polar histogram 16.

[0054] The responses from the polar histogram 16 generated for each segment 11 allow the aforementioned change in swarm direction to be recorded and the direction in which this change tends to be determined. Based on the feedback from the polar histogram 16, a corresponding warning, in particular a signal to the driver of vehicle 2, can then be issued.

[0055] In this process, a given action can be advantageously parameterized based on the response of polar histogram 16 (Turbulence Search Area), or rather, should be parameterized by defining at which response in polar histogram 16 a turbulence warning should be issued. The corresponding processing of the signal is described in Fig. Figure 8 shows the signal as visualization 15.

[0056] The corresponding reaction regarding the turbulence warning can be initiated initially at the factory, but also online in the field, for example, in the case of a known strong wind warning and the detection of bridges and thus static objects based on environmental perception or using a map, as shown in map section 17. It should be emphasized again, however, that an advantage of the method presented here is that traffic-critical situations are forwarded to the driver without considering additional external vehicle sensors or map data. Furthermore, the method shown is robust against weather and lighting conditions and allows for initial application and / or calibration at the factory or adaptively in the field.

[0057] In summary, the advantages shown here include adaptive driver warnings in critical weather conditions, individual consideration of the current traffic environment, the aforementioned independence from weather and lighting conditions, the need for no additional map data, and independence from sensor modalities, which generally also means that no hardware modifications are required. Furthermore, the procedure can be adaptively and / or cyclically tested.

Claims

[1] Method for operating a driver assistance system (1) for a motor vehicle (2) by which environmentally caused wind turbulence (3) can be detected, comprising the steps: - Generating an environment model (4) that describes road users (6) in an environment (7) of the motor vehicle (2) using sensor data from at least one vehicle-integrated sensor of the driver assistance system (1); (VS1) - Determining a respective lane assignment, lane curvature (8) and vehicle orientation (9) for the motor vehicle (2) and the road users (6); (VS2) - Determining the relative orientation (10) between track curvature (8) and vehicle orientation (9) for the motor vehicle (2); (VS3) - Determining the relative orientation (10) between lane curvature (8) and vehicle orientation (9) for road users (6) and combining the relative orientations (10); (VS4) - Segmenting the environment (7) into segments (11) so that an assignment of the respective relative orientation (10) is made to each segment (11); (VS5) characterized by the steps: - Generating a frequency representation (12) of the orientations (10) of road users (6) per segment (11); (VS6) - Determining an orientation distribution (13) based on the frequency representation (12); (VS7) - Deriving a turbulence orientation (14) from the orientation distribution (13); (VS8) and - Outputting a signal depending on the turbulence orientation (14). (VS9) [2] Method according to claim 1, characterized by , that a radar, a lidar and / or a camera is used as a sensor, which allows the environmental model to be generated based on different sensor data, such as images and / or 3D point clouds. [3] Method according to claim 1 or 2, characterized by, that the signal is output as a control signal for a driver assistance system (1) and / or as a visualization (15) of the turbulence orientation (14) and / or as a warning signal. [4] Method according to any one of the preceding claims, characterized by , that in order to determine the relative orientation (10) additionally a speed and / or an acceleration for the motor vehicle (2) and / or for the road users (6) are used and / or that a respective lane normalization is carried out for the road users (6). [5] Method according to any one of the preceding claims, characterized by , that for the frequency representation (12) a polar histogram (16) is used and thus an orientation sequence of the road users (6) in the respective segment (11) is determined. [6] Method according to claim 5, characterized by , that the orientation distribution is determined based on frequency clusters in the polar histogram (16). [7] Method according to any one of the preceding claims, characterized by , that the segments (11) are divided according to a distance from the motor vehicle (2). [8] Method according to any one of the preceding claims, characterized by , that static objects along a roadway are detected and used in deriving the turbulence orientation (14). [9] Driver assistance system (1) for a motor vehicle (2) comprising a control unit and at least one sensor, which is configured to perform a method according to one of the preceding claims. [10] Motor vehicle (2) with a driver assistance system (1) according to claim 9 and / or equipped to perform a method according to any one of claims 1 to 8.

Images