Procedure for operating a motor vehicle

Abstract

Procedure for operating a motor vehicle (1), comprising the steps: - Determining whether a section of road (2) that is expected to be travelled by the motor vehicle (1) in the future includes a junction (3) where the section of road (2) crosses a road (4) or where a road (4) merges into the section of road (2), by means of a recognition algorithm (5), - Checking a trigger condition (6) which is fulfilled or can only be fulfilled if, after the detection of the intersection (5) and before the motor vehicle (1) passes through the detected intersection (5), the presence of a road user (7, 8) on the roadway (4) is detected, and - Issuance of a notice (9) concerning the road user (7, 8) and / or execution of an automated driving intervention (10) when the trigger condition (6) is met.

Description

[0001] The invention relates to a method for operating a motor vehicle and a motor vehicle.

[0002] Road safety at intersections with poor visibility is typically ensured through a combination of structural measures, traffic engineering, and driver behavior. Clear and highly visible traffic signs and road markings help to direct drivers' attention and alert them to hazards. These include, for example, right-of-way rules, speed limits, warning signs, and traffic lights.

[0003] At some intersections, there may be obstructions to visibility, such as trees, bushes, and structural obstacles. If such obstructions cannot be removed, traffic mirrors can be used to give drivers a view of areas that are otherwise difficult or impossible to see.

[0004] According to numerous studies, despite the widespread use of the aforementioned measures, a significantly increased risk of accidents results at confusing intersections. While various driver assistance systems are known to minimize the risk of accidents, specialized assistance systems, such as blind spot monitoring or lane keeping assist, can only achieve a noticeable risk reduction for some of the driving situations relevant at such intersections. Known collision warning systems, which serve to detect potential collisions and warn the driver in time, aim to avoid a frontal impact, whereas at confusing intersections, the primary risk is a side impact.

[0005] To better inform a driver about potentially obscured objects, US document 2024 / 4142597 A1 proposes that, when detecting moving objects, such as a pedestrian, with a radar sensor, stationary objects should also be detected using a camera. If it is determined that the moving object should actually be obscured by the stationary object at the position determined based on the radar sensor data, it is assumed that the detection is based on a reflection of the radar radiation from the stationary object, and the coordinates of the moving object are corrected accordingly.

[0006] The invention is based on the objective of further improving the support provided to a driver in potentially confusing driving situations.

[0007] The problem is solved according to the invention by a method for operating a motor vehicle, which comprises the following steps: - Determining, using a recognition algorithm, whether a section of road that is expected to be traveled by the motor vehicle in the future includes a junction where the section of road crosses a road or where a road merges into the section of road. - Checking a trigger condition that is fulfilled, or can only be fulfilled, if, after the detection of the intersection and before the motor vehicle passes through the detected intersection, the presence of a road user on the road is detected, and - Issuance of a warning to the road user and / or execution of an automated driving intervention when the trigger condition is met.

[0008] The inventive method allows the system to take into account whether a detected road user is located in the vicinity of a detected intersection while the vehicle is approaching or entering the intersection. This enables the implementation of a driver assistance system that provides warnings or intervenes in the driving process, specifically designed to support the driver at intersections, particularly those with poor visibility. Because the system focuses on a specific driving situation, the threshold for providing warnings or implementing automated driving interventions can be considerably lower than would be possible with an assistance system that, for example, only considers the relative movement between the vehicle and the road user or potential collision objects to indicate detected collision risks.This makes it possible to react early to road users on the other road who are potentially not yet visible to the driver of the motor vehicle, without reducing the acceptance of the driver assistance system through excessively frequent warnings and / or driving interventions.

[0009] As will be explained in more detail later, the detection of road users can be achieved primarily through vehicle-mounted sensors. Therefore, depending on the surroundings of the intersection, detection may only be possible once the vehicle enters the intersection area, such as a junction or intersection. However, since the sensors used to detect road users, such as radar sensors, can be located at the front of the vehicle, the proposed notification still provides the driver with additional reaction time, or the automated intervention can begin even before the driver can perceive the road user.

[0010] The transport route can be any transport route that is not part of the section of road to be traveled and that crosses or merges into it. Merging here can be understood as any convergence of the road section and the transport route.

[0011] The steps of the procedure can be carried out during the operation of the motor vehicle, in particular repeatedly, for each preceding section of the route.

[0012] The vehicle can be at least 3 m, at least 5 m, or at least 10 m away from the track segment at the point in time when all input data for the recognition algorithm are available and / or the execution of the recognition algorithm for the track segment begins. Additionally or alternatively, this point in time can be at least 0.5 s or at least 1 s before the expected arrival at the track segment. In particular, the recognition algorithm can thus be executed for each track segment that is still at a certain temporal or spatial distance ahead of the vehicle along its expected route. This allows for a sufficiently long computation time to complete the recognition algorithm before the track segment, and thus a potential intersection within it, is reached.

[0013] The recognition algorithm can use map data of the route segment as input data, and optionally, image data from a camera. Detecting intersections and their properties can be relatively computationally intensive, especially when image data is evaluated and / or input data from multiple sources is fused. Near-real-time capability, where the detection of a potentially present intersection occurs only at or immediately before reaching it—for example, only when entering an intersection or junction, or just a few tens of milliseconds before entering—would therefore require significant computing power and thus increase the vehicle's costs and energy consumption.

[0014] However, it has been recognized that reliable detection and optional classification of intersections is already possible based on input data that can be determined at a significant temporal or spatial distance from the intersection or the section of road encompassing the intersection. For example, the presence of the intersection can be detected well before reaching the respective section of road, based on the vehicle's current position, a predicted route, and map data. Furthermore, a sufficiently accurate visual assessment of the area surrounding a potential intersection is possible several meters or seconds before reaching the intersection.By starting the execution of the recognition algorithm earlier, a longer computing time is available for its execution, thus requiring a lower computing speed and therefore less hardware and / or energy expenditure.

[0015] Upon detecting a junction in the section of road, the recognition algorithm can perform a classification to determine whether the detected junction is an obscured junction where the view of the road for a driver of the motor vehicle is restricted by at least one obstacle before and / or upon reaching the junction, whereby the fulfillment of the trigger condition depends on the classification.

[0016] The issuance of a warning to road users and / or automated driving intervention can only occur if the driver's view of the road is likely to be restricted. This allows driver assistance to be limited to intersections with potentially poor visibility, either always or, for example, when a driver assistance system implementing the described procedure is configured accordingly. This can be advantageous because early warnings and / or automated driving interventions in situations where the driver can identify relevant road users early and react accordingly may be perceived as disruptive by drivers and could therefore reduce the acceptance of such a driver assistance system.

[0017] The classification of an intersection can depend, for example, on the vehicle's position relative to the intersection, from which point a relevant section of the road and / or other road users are likely to become visible to the driver. A relevant section of the road can be a section of the road adjacent to the intersection or a lane of the road leading towards the intersection, with a predetermined length. This length can be fixed or determined based on a maximum speed limit on the road, which may be known from map data. Specifically, this length is chosen to encompass all positions from which other road users are likely to reach the intersection before the vehicle has completely crossed it.

[0018] Preferably, the recognition algorithm processes as input data both image data from at least one camera of the motor vehicle and map data.

[0019] It is generally known to determine the section of road a vehicle is likely to travel on in the future based on map data and its current position, for example, based on a route planned or predicted by a navigation system. Since such map data also describes other roads and their intersections with or junctions with the route segment, intersections in the upcoming section can be identified with relatively little computational effort. Map data that is kept up-to-date, for example, through wireless data transmission to the vehicle, is particularly useful for this purpose, as it allows for the consideration of temporary changes to the road network, such as temporary construction sites.

[0020] The map data can directly provide the classification of the intersection, or at least parts of the information considered within this classification. However, since not all factors that can influence the driver's view of the road are typically described with sufficient accuracy or up-to-date information by the available map data—for example, the presence of parked vehicles and / or relevant vegetation—it is advisable to also consider the camera image data for classification purposes.

[0021] For example, map data can be used to estimate the temporal or spatial distance from the node at which image data relevant for its classification can likely be acquired. Image acquisition can then be repeated until the image data shows that the node and its surroundings are sufficiently well represented to allow for classification. This is typically possible well before reaching the node, ensuring that even with relatively limited computing power, sufficient processing time remains to determine the classification.

[0022] In particular, at the time when the camera image data is captured, which is processed as input data for the recognition algorithm, and / or when the execution of a sub-algorithm of the recognition algorithm begins, which performs the classification as to whether the recognized intersection is a confusing intersection, the motor vehicle may be at least 3 m or at least 5 m or at least 10 m away from the intersection and / or this time may be at least 0.5 s or at least 1 s before the expected arrival at the intersection.

[0023] For example, during operation, it can be repeatedly checked whether a junction is located on the respective section of track to be traveled in the future, in order to carry out the acquisition of the image data underlying the classification or the classification itself in good time before reaching the junction.

[0024] Preferably, radar data from at least one front corner radar of the vehicle is processed to determine whether the road user is present in the roadway. A front corner radar is typically located laterally in the area of ​​the vehicle's front or in the transition area between the front and side, thus covering the area to the front left or front right of the vehicle. Such a corner radar is therefore particularly well suited to detecting road users in the roadway at an early stage.

[0025] If front corner radars are present on both the left and right sides of the vehicle, the radar data from which of the corner radars is considered under the trigger condition can be selected, depending on the classification of the intersection. For example, if the intersection is a junction, only the radar data from the corner radar on the side where the road merges can be considered.

[0026] Even when the roadway crosses the section of track, it may be advantageous to consider radar data from only one of the radar sensors within the triggering condition, for example, if obstacles in the vicinity of the intersection only restrict the driver's view of the roadway on one side. However, it is also possible to evaluate radar data from both a left and a right corner radar, particularly if the intersection is a junction and especially if visibility is poor on both sides of the roadway.

[0027] As explained above, the point in time at which the input data for the recognition algorithm is available and / or at which the execution of the recognition algorithm for the section of the route begins may be, in particular, before the point in time from which a relevant section of the road and / or the road user itself can be detected by sensors of the motor vehicle, in particular by the corner radar.

[0028] The recognition algorithm can be or comprise a function trained through machine learning. The trained function can be, for example, a neural network (especially a deep one) or a transformer. In particular, the trained function can process image data from the camera as input, enabling pattern and / or object recognition. This allows, for example, the identification of obstacles that could potentially obstruct the driver's view of the road and thus the classification of the intersection described above.

[0029] However, it may also be useful to process both the map data regarding the actual position of the vehicle and the image data from the camera as input data for the trained function, for example, to implement the entire recognition algorithm, such as the recognition and classification of the junction, by the trained function, or to take the map data into account when classifying the junction.

[0030] While it is theoretically possible to use manually programmed or parameterized algorithms for object or pattern recognition, such as well-known algorithms for recognizing scale-invariant features, a trained function often achieves better robustness and accuracy. Furthermore, trained functions are well-suited for estimating the three-dimensional positions and / or dimensions of objects based on two-dimensional image data. Trained functions are also well-suited for recognizing and considering complex relationships in input data, such as relationships between map data and image data.

[0031] The training of the function can be carried out in a known manner using supervised learning, for example, through error feedback, specifically using a gradient descent method. Training datasets can be provided, for example, by specifying input data for the algorithm for a large number of real-world route segments—in particular, map data and / or image data—as well as target output data, in particular, information on whether a junction exists in the route segment and / or its position and / or classification. The target output data can, for example, be based on an assessment of the input data or the underlying real-world route segment by a human evaluator or expert.

[0032] Preferably, directional information is determined that indicates in which direction the road user is located with respect to the section of road.

[0033] The warning can then be displayed by a vehicle display that extends laterally over at least 50% or at least 70% of the vehicle's interior width, the position of which depends on the directional information. In particular, a laterally elongated display strip, which may be located on the dashboard or formed by a windshield root display, can be used as the display. The width of such a display strip in the transverse direction can be at least five, ten, or twenty times its height.

[0034] Alternatively, the warning can be displayed by one of two indicators spaced apart from each other in the transverse direction of the vehicle, with the selection of which indicator is used to display the warning depending on the directional information. For example, two elongated display strips can be used, positioned at the same height to the left and right of the instrument cluster or a display that represents an instrument cluster.

[0035] By selecting the display position or display medium as described above, depending on the directional information, the driver can, for example, be alerted to a road user approaching the intersection from the left by a message displayed to the left of the instrument cluster, while information about a road user approaching from the right can be displayed to the right of the instrument cluster. This allows the driver's gaze to be intuitively directed towards the direction from which the road user is approaching.

[0036] The display device(s) can be graphically capable, allowing a suitable symbol, such as an image of a motor vehicle or a warning symbol, to be displayed as an indication. Alternatively, the indication can be provided by a simple light source, such as one or more LEDs.

[0037] When the presence of a road user is detected on the road, a relevance measure can be determined that describes a collision risk for a collision between the motor vehicle and the road user, whereby on the one hand the fulfillment of the trigger condition and / or on the other hand the indication given when the trigger condition is fulfilled depends on the relevance measure.

[0038] For example, in the case of a visual output of the warning, a luminous color and / or a luminous brightness of a displayed warning symbol or an activated light source can be selected depending on the relevance level and thus the criticality of the driving situations.

[0039] The relevance measure can depend, for example, on the distance of the road user from the intersection, and / or on the direction of travel, and / or speed, and / or acceleration of the road user, and / or on the maximum speed permitted on the roadway, and / or on the distance of the vehicle from the intersection. Approaches to determining the probability of a collision or the criticality of a driving situation are well-known and will therefore not be explained in detail. For example, it can be assumed that the vehicle is moving along the road segment and the other road user along the roadway, whereby the relevance measure can be particularly high the shorter the time interval between the predicted passage of the road user and the predicted passage of the roadway.For example, it can be assumed that the road user and the motor vehicle continue to move at a constant speed or acceleration.

[0040] In addition to the method according to the invention, the invention relates to a motor vehicle with a control unit configured to carry out the method according to the invention. The motor vehicle may have sensors, in particular at least one corner radar, configured to detect the road user in the method according to the invention.

[0041] Preferably, the motor vehicle includes a position sensor, for example a GPS sensor, for recording the vehicle's current position. This can be used, as explained above, particularly in conjunction with map data available on the vehicle, to check whether the intersection exists in the route segment and / or to classify detected intersections.

[0042] Preferably, the motor vehicle comprises at least one camera in order to provide image data, as already explained above in relation to the inventive method.

[0043] Further advantages and details of the invention will become apparent from the following exemplary embodiments and the accompanying drawings. These schematically illustrate: Fig. 1 an exemplary traffic situation in which an embodiment of the method according to the invention is carried out by an embodiment of the motor vehicle according to the invention, and Fig. 2 a flowchart of an embodiment of the method according to the invention.

[0044] Fig. Figure 1 shows a traffic situation in which motor vehicle 1 is approaching a junction 3, in this example an intersection. The intersection, or junction 3, is obscured, at least with regard to the part of the intersecting road 4 to the right of the section of road 2 that motor vehicle 1 will be traveling on, because the view of the driver 14 of motor vehicle 1 is still restricted by the obstacle 15 even at position 29, where motor vehicle 1 is already entering the intersection. Therefore, without additional assistance, driver 14 can only react very late if road user 7 is present, potentially entering the intersection almost simultaneously with motor vehicle 1, thus creating a significant risk of collision.

[0045] In many cases, structural measures are therefore planned at confusing intersections 3 to assist the driver, for example the traffic mirror 30, which is located in Fig. 1 is shown. However, since such structural measures are not present in all cases and, moreover, the recognition of relevant road users at confusing intersections may be difficult despite such measures, the motor vehicle 1 includes a control device 26 which is designed to support the driver 14 in such driving situations.

[0046] This will be discussed below with additional reference to Fig. Section 2 explains the process, which shows a flowchart of the assistance procedure implemented by the control unit 26. For better understanding, the central processes in the procedure will first be briefly discussed before the individual steps of the process are described. Fig. The exemplary design of such a procedure shown in section 2 will be discussed.

[0047] Steps S1 to S5 serve to recognize, by means of a recognition algorithm 5, whether the section of road 2 which is expected to be travelled by the motor vehicle 1 in the future includes a junction 3 where the section of road 2 crosses a road 4 or where a road 4 merges into the section of road 2, whereby in the example an additional classification 13 of recognized junctions 3 is carried out.

[0048] If a junction 3 is detected, then, as part of the trigger condition 6, it is checked whether the presence of a road user 7, 8 on the road 4 is detected before the motor vehicle 1 passes through the detected junction 3.

[0049] If this is the case, at least if the detected road user 7,8 is likely to be difficult to recognize for the driver 14 of the motor vehicle 2 and is relevant for the operation of the motor vehicle 1, a notification 9 regarding the road user 7, 8 will be issued to the driver 14 and / or an automated driving intervention 10 will be carried out in order to reduce the risk of a collision.

[0050] The individual steps are explained in more detail below. In the illustrated embodiment, in step S1, the actual position 27 of the vehicle 1 is first recorded using the GPS sensor 28 of the vehicle 1.

[0051] In step S2, map data 18 of the area surrounding the vehicle 1 are provided as input data 11 for the recognition algorithm 5. This data describes the route that the vehicle 1 will travel in the future and the roads 4 that merge into or cross it. Suitable map data 18 can be provided by selecting a map section from a map stored in the control unit 26 or provided, for example, by a navigation device (not shown), depending on the previously determined actual position 27 of the vehicle 1. Alternatively, it would be possible, for example, to retrieve or receive the map data from an infrastructure facility or a server.

[0052] In step S3, further input data 12 for the recognition algorithm 5 includes image data 16 from at least one camera 17 of the motor vehicle 1, which, as will be explained in more detail later, primarily serve to classify 13 of recognized nodes 3 in the example.

[0053] In step S4, a first sub-algorithm of the recognition algorithm 5 checks whether a road section 2, which is expected to be traveled by the motor vehicle 1 in the future, includes a junction 3 where the road section 2 crosses a road 4 or where a road 4 merges into the road section 2. If it can be assumed that the map data 18 is sufficiently up-to-date to also reflect changes in traffic flow due to construction sites, for example, this can preferably be done exclusively on the basis of the map data 18, since junctions can be recognized there with minimal effort and the recognition of junctions is also possible at a relatively large distance before the respective junction.

[0054] To provide sufficient computing time for the detection of nodes 3 and the subsequent processing steps, especially for the classification 13 of nodes 3, even though the control unit 26 only provides relatively low computing power, the example considers nodes 3 and thus also track sections 2 that are several meters away from the current position of the vehicle and are therefore expected to be reached only in 1 to 2 seconds. The distance of the considered track section 2 from the current position of the vehicle 1 is in Fig. 1 shown schematically. If no nodes 3 are detected in the currently considered section 2, the procedure is repeated from step S1, especially after a short waiting period.

[0055] In step S5, a further sub-algorithm of the recognition algorithm 5 then performs a classification 13 of the node 3. If several nodes 13 are detected in the track segment 2, it is typically sufficient to initially perform the classification only for the nearest of the detected nodes 3.

[0056] Within the framework of classification 13, it is determined whether the identified intersection 3 is an obscured intersection 3, where the view of the roadway 4 for a driver 14 of the motor vehicle 1 is restricted by at least one obstacle 15 before and / or upon reaching the intersection 3. Fig. Figure 1 represents a building in the intersection area as such an obstacle.

[0057] Such a classification 13 serves to restrict warnings or driving interventions at confusing intersections. This can be useful because many drivers find early warnings or driving interventions disruptive when the relevant road users 7, 8 are readily identifiable by the driver, allowing them to react in time. However, it would also be possible, for example by selecting a different operating mode of the control unit 26, to always provide a warning about road users 7, 8 on intersecting or merging roads 4, or to react to their presence with a driving intervention as needed.

[0058] For classification 13, approaches known in themselves can be used to identify objects that represent potential obstacles 15, such as vegetation, buildings, parked vehicles and signs, and to estimate their positions and dimensions with respect to the map data in order to determine, on the basis of the resulting model, from what distance before the intersection a relevant area of ​​the road 4 is likely to be visible to the driver 14.

[0059] Since machine learning-trained algorithms have proven particularly suitable for identifying objects in image data and estimating their positions or dimensions in three-dimensional space, the classification in this example is performed using such an algorithm. In principle, it would be possible to perform object recognition using only such an algorithm. However, it can be advantageous to consider both the image data 16 and the map data 18 as input data for the trained function in order to perform the entire classification of node 3, or even the entire recognition and classification of nodes 3, by a single trained function. The training of a suitable function has already been explained in more detail in the general section of the description.

[0060] In step S6, radar data 19 from at least one front corner radar 20, 21 of the motor vehicle 1 are acquired in order to image road users 7, 8 on the roadway. Since the corner radars 20, 21 are located significantly in front of the driver's seat position 14, for example, in Fig. One of the road users 8 is already detected by the corner radar 21, while it is still obscured by the obstacle 15 for the driver 14. A warning or driving intervention based on the radar data 19 can therefore be possible before the driver 14 can even perceive the road user, thus increasing the reaction time available to the driver to avoid critical situations or collisions.

[0061] The acquisition of radar data 19 can, in principle, take place continuously during driving. In the example discussed, however, only radar data 19 that are already assumed to pertain to relevant areas of the road are processed further. Such radar data 19 are typically only received shortly before the vehicle 1 enters the area of ​​intersection 3, for example at the Fig. 1 schematically represented position 29.

[0062] If, at node 3, it were the case, contrary to the example in Fig. 1. If the traffic route 4 were to merge into the section 2 from one side, it would be sufficient to process only the radar data of the corner radars 20, 21 facing this junction.

[0063] In step S7, a trigger condition 6 is then checked, which in the example can only be fulfilled if, on the one hand, node 3 has been classified as an unclear node 3, which in the example according to Fig. 1 due to the obstacle 15, and on the other hand, the presence of a road user 7, 8 on the crossing or merging road 4 is detected based on the radar data 19.

[0064] To avoid unnecessary warnings or interventions in driving operations, a relevance measure 25 is also determined in the example, which describes the collision risk of a collision between motor vehicle 1 and road user 7, 8, and the trigger condition 6 can only be fulfilled if the relevance measure 25 is sufficiently high. As already discussed in more detail in the general section, the relevance measure 25 can depend on whether the vehicle 1 and the respective road user 7, 8 are expected to pass intersection 3 almost simultaneously or with a large time interval between them.

[0065] In the Fig. In the example shown, the intersection shown as junction 3 is only unclear with respect to that part of the roadway 4 which lies to the right of the roadway segment 2 and is therefore obscured by the obstacle 15. The control unit 26 can, for example, be configured so that driving interventions or warnings within the framework of the described procedure are only made for road users 7, 8 who are not recognizable in advance by the driver 14 himself, i.e., in the example of the Fig. 1 only for road user 7 in the section of road 4 to the right of section 2. Therefore, in the trigger condition 6 in step S7, additional direction information 22 is evaluated, which indicates in which direction road user 7, 8 is located with respect to section 2, i.e., for example, by means of which of the corner radars 20, 21 the respective road user 7, 8 was detected.

[0066] If the trigger condition is not met, the procedure is repeated from step S1. If, however, trigger condition 6 is met, a notification 9 should be issued to road user 7, 8 in step S8, as in the example. For the purposes of the following explanation, it is assumed by way of example that in the Fig. The traffic situation shown is intended to draw attention exclusively to road user 7, as road user 8 is already recognizable early on by driver 14 himself.

[0067] The output of the warning 9 was achieved, for example, by controlling a display 23 of the vehicle 1, which extends laterally across a large part of the width of the interior 24 of the vehicle 1, for example, across the entire width of the dashboard. In this example, the display 23 is a graphic display. The warning is presented in the form of a warning symbol, the position of which depends on the directional information 22 explained above. Thus, for example, it is possible to display a warning symbol regarding road user 7 to the right of the instrument cluster of the vehicle 1 in order to draw the driver's attention in that direction.

[0068] As already discussed in the general part of the description, alternatively, separate display devices spaced apart from each other in the transverse direction of the motor vehicle can be used, whereby in this case it can be selected, based on the direction information 22, which of the display devices provides the information.

[0069] The way the warning is displayed, for example, the color of the indicator and / or whether the warning symbol is continuously illuminated or flashing, depends on the relevance level explained above. The driver is thus not only alerted that a road user 7 who is difficult or impossible to see is present and in which direction it is located, but can also immediately recognize how relevant or critical the presence of this road user 7 is likely to be for further driving. In cases of very high relevance or criticality, it is also possible, for example, to additionally emit a warning tone or similar signal.

[0070] In addition to or as an alternative to issuing the instruction 9, a driving intervention 10 can also be carried out in step S9, for example to brake the motor vehicle 1 before the junction 3 or to carry out an evasive maneuver. QUOTES INCLUDED IN THE DESCRIPTION

[0000] This list of documents cited by the applicant was automatically generated and is included solely for the reader's convenience. The list is not part of the German patent or utility model application. The DPMA accepts no liability for any errors or omissions. Cited patent literature

[0000] US 2024 / 4142597 A1

[0005]

Claims

[1] Method for operating a motor vehicle (1) comprising the steps: - Determining whether a section of road (2) that is expected to be travelled by the motor vehicle (1) in the future includes a junction (3) where the section of road (2) crosses a road (4) or where a road (4) merges into the section of road (2), by means of a recognition algorithm (5), - Checking a trigger condition (6) which is fulfilled or can only be fulfilled if, after the detection of the intersection (5) and before the motor vehicle (1) passes through the detected intersection (5), the presence of a road user (7, 8) on the roadway (4) is detected, and - Issuance of a notice (9) concerning the road user (7, 8) and / or execution of an automated driving intervention (10) when the trigger condition (6) is met. [2] Method according to claim 1, characterized by, that the motor vehicle (1) is at least 3 m or at least 5 m or at least 10 m away from the section of track (2) at the time when all input data (11, 12) for the recognition algorithm (5) are available and / or the execution of the recognition algorithm (5) for the section of track (2) begins, and / or that this time is at least 0.5 s or at least 1 s before the expected reaching of the section of track (2). [3] Method according to claim 1 or 2, characterized by, that the recognition algorithm (5) upon detection of the junction (3) in the track section (2) performs a classification (13) to determine whether the detected junction (3) is an obscured junction (3) where the view of the roadway (4) for a driver (14) of the motor vehicle (1) is restricted by at least one obstacle (15) before and / or upon reaching the junction (3), wherein the fulfillment of the trigger condition (6) depends on the classification (13). [4] Method according to any of the preceding claims, characterized by , that the recognition algorithm (5) processes as input data (12, 13) both image data (16) from at least one camera (17) of the motor vehicle (1) and map data (18). [5] Method according to any of the preceding claims, characterized by, that radar data (19) from at least one front corner radar (20, 21) of the motor vehicle (1) are processed to determine whether the road user (7, 8) is present on the roadway (4). [6] Method according to any of the preceding claims, characterized by , that the recognition algorithm (5) is or comprises a function trained by machine learning. [7] Method according to any of the preceding claims, characterized by , that direction information (22) is determined which indicates in which direction the road user (7, 8) is located with respect to the section of road (2), whereby the output of the hint (9) - either by means of a display means (23) of the motor vehicle (1) which extends in the transverse direction of the motor vehicle (1) over at least 50% or over at least 70% of the width of the interior (24) of the motor vehicle (1), wherein a display position in the transverse direction at which the indication (9) is displayed on the display means (23) depends on the direction information (22), - or by one of two display means (23) which are arranged spaced apart from each other in the transverse direction of the motor vehicle (1), wherein the selection of which of the display means (23) is used to display the notice (9) depends on the direction information (22). [8] Method according to any of the preceding claims, characterized by, that when the presence of a road user (7, 8) on the road (4) is detected, a relevance measure (25) is determined which describes a collision risk for a collision of the motor vehicle (1) with the road user (7, 8), whereby on the one hand the fulfillment of the trigger condition (6) and / or on the other hand the indication (9) given when the trigger condition (6) is fulfilled depends on the relevance measure (25). [9] Motor vehicle with control unit (26), characterized by that the control device (26) is set up to carry out the procedure according to one of the preceding claims.

Images