Method for generating a view of the surroundings, method for training a neural network, controller, vehicle and computer program

Abstract

The invention relates to a method for generating a view of the surroundings located at least partially behind a trailer (3) of a vehicle combination (1) comprising the trailer (3) and a towing vehicle (2) from camera images obtained with the aid of a camera arrangement, wherein the camera arrangement has at least one first camera (4) recording at least one section of the trailer and at least one second camera (5) recording at least one subregion of the surroundings behind the trailer, comprising the following steps: - recording at least one first camera image (13) using the at least one first camera (4) and at least one second camera image using the second camera (5), - identifying at least one segment (15) of the first camera image (13) that corresponds to an end face, facing the towing vehicle (2), of the trailer (3) in the first camera image (13) by segmenting the first camera image (13) and - generating the view of the surroundings from the first camera image (13) and the second camera image on the basis of the identified segment (15), wherein at least part of the segment (15) is supplemented with image data from the second camera image.

Description

[0001] Description

[0002] Method for generating an environmental view, method for training a neural network, control unit, vehicle and computer program

[0003] The invention relates to a method for generating an environmental view of an environment located at least partially behind a trailer of a vehicle combination comprising the trailer and a towing vehicle from camera images acquired using a camera arrangement. The camera arrangement has at least one first camera capturing at least a portion of the trailer and at least one second camera capturing at least a partial area of ​​the environment behind the trailer. Furthermore, the invention relates to a method for training a neural network, a control unit, a vehicle, and a computer program.

[0004] In a vehicle combination consisting of a tractor unit and a trailer, a camera that captures the area behind the tractor unit, such as a rearview camera, is of limited use, as the camera's field of view is at least partially blocked by the trailer. However, a clear view to the rear is desirable, especially with such a combination, as the trailer can also impede the driver's safety when facing backward or using a rearview or side mirror.

[0005] It is known from the prior art that the images from a rear-view camera of the towing vehicle are supplemented by images recorded by a camera arranged on the rear of the trailer, facing away from the towing vehicle. In this way, a view of the surroundings can be generated in which a part of the surroundings blocked by the trailer becomes visible in the images from the rear-view camera by displaying the trailer at least partially transparently, thus enabling an at least substantially unblocked view of the surroundings behind the vehicle. Such a method is described, for example, in the document WO 2021 / 046379 A1.

[0006] To generate a quasi-transparent trailer in the surrounding view, it is necessary to know the trailer's dimensions to enable seamless stitching of the images from the various cameras. For this purpose, it is common to assume that the trailer's representation in an image from the towing vehicle's rear-view camera is a four-sided surface whose dimensions are known, for example, due to a user of the vehicle combination entering these dimensions. However, generating a transparent view of the trailer in the surrounding image is limited to box-shaped trailers with known and constant dimensions.

[0007] The invention is based on the object of specifying an improved method for generating an environmental view of an environment located at least partially behind a trailer of a vehicle combination, which in particular facilitates use of the method with different trailers.

[0008] To achieve this object, the invention provides for a method of the type mentioned at the outset to comprise the following steps:

[0009] - capturing at least one first camera image with the at least one first camera and at least one second camera image with the second camera,

[0010] - Determining at least one segment of the first camera image corresponding to a front surface of the trailer facing the towing vehicle in the first camera image by segmenting the first camera image,

[0011] - Generating the environmental view from the first camera image and the second camera image depending on the determined segment, wherein at least part of the segment is supplemented with image data from the second camera image.

[0012] The first camera can, for example, be arranged on the towing vehicle and capture an area behind the towing vehicle. The first camera can, for example, be a rear-view camera of the towing vehicle. The camera image recorded by the first camera shows the front of the trailer facing the towing vehicle. The geometry of the front of the trailer facing the towing vehicle depends in particular on a pivot angle between the towing vehicle and the trailer and can, in addition to a front side or a section of a front side of the trailer, also completely or at least partially encompass one or more sides or side surfaces of the trailer.The front surface of the trailer within the meaning of the invention is not only limited to a front side of the trailer body, but can also include the load of the trailer arranged on the trailer and visible from the first camera, for example if the load is arranged on a trailer without a cover or the like.

[0013] By segmenting the first camera image to detect the trailer's front end depicted in the first camera image, it is advantageously possible to detect trailer front ends with any geometry. Due to the segmentation of the first camera image to detect the trailer's representation in the first camera image, the trailer can thus advantageously be detected independently of a predefined geometry and independently of dimensions entered in advance or provided in another way, and can be hidden in the surrounding view or supplemented with image data from the second camera image. This advantageously makes it possible to achieve an at least partially transparent representation of the trailer in the surrounding view for trailers of any shape.

[0014] Advantageously, in addition to box-shaped trailers, other types of trailers, for example trailers for transporting vehicles such as boat trailers, motorcycle trailers, car trailers or similar, can be completely or partially hidden in the surrounding view, with or without a load. The same applies to trailers for transporting other loads, such as flatbed trailers, trailers for transporting bulk goods or containers, and so on, which can also have different front surfaces when unloaded and when fully or partially loaded. A further advantage of the solution according to the invention is that it is possible to dispense with determining the dimensions of the trailer, for example by measuring the trailer. This simplifies the application of the method within the framework of a driver assistance system, since it can be used with different trailers without user input of trailer dimensions or the like.Furthermore, the method can be used on a trailer without any further action, both in unloaded and in differently loaded conditions.

[0015] By means of segmentation, individual pixels of the first camera image are assigned to the front face of the trailer depicted in the first camera image. For the pixels or pixel areas identified as belonging to the front face of the trailer, a contour enclosing the pixels can then be determined, which encloses the segment.

[0016] Depending on the characteristics of the trailer or its frontal area, one or more segments can be determined. Pixels from the first camera image that are not assigned to the frontal area of ​​the trailer can, for example, be assigned to other sections of the trailer and / or the surroundings of the vehicle combination.

[0017] The second camera image, which is recorded by the second camera, shows the area around the vehicle combination behind the trailer. The second camera can be arranged, for example, on the back or rear of the trailer. The at least one segment determined in the first camera image is then supplemented with image information from the second camera image. The segment can be partially or completely supplemented with image data from the second camera image. The supplementation is carried out in particular in such a way that image sections of the area within the segment, which are not visible in the first camera image due to the trailer, are filled with corresponding image sections of the second camera image. In this way, an image of the area behind the vehicle combination, in particular an at least essentially distortion-free and continuous image, can be generated as an environmental view.

[0018] The environmental view can be generated depending on position information that describes the relative arrangement between the first camera and the second camera. In this way, the image data from the second camera image can be inserted into the segment determined in the first camera image as seamlessly and as distortion-free as possible. The position information can, in particular, describe a spatial offset between the first camera and the second camera, for example, through a translation and / or a rotation. The position information can be initially specified and determined, for example, by a measurement and / or continuously determined, in particular during a driving movement of the vehicle combination.

[0019] Continuous determination of the position information can be carried out, for example, by odometry and / or by machine vision, for example based on optical flow or highlighted features in the first and / or second camera images. Extrinsic camera parameters of the first camera and / or extrinsic camera parameters of the second camera and / or a vehicle model of the towing vehicle, the trailer, and / or the vehicle combination can also be used to determine the position information.

[0020] The segmentation takes place continuously, in particular, for a plurality of first camera images, in particular for first camera images recorded in a video stream. Accordingly, image data from second camera images, also continuously recorded, are also added to the continuously acquired and segmented first camera images, so that the environmental view can also be displayed as an image stream or video. The environmental view can be displayed, for example, on a display device arranged in the towing vehicle and visible to a driver of the towing vehicle.According to the invention, it can be provided that a lateral swivel angle of the trailer is determined depending on the position of one or more edges of the at least one segment in the first camera image and the position of a coupling point at which the trailer is laterally swivel-coupled to the towing vehicle, wherein the surrounding view is generated depending on the lateral swivel angle and / or wherein, when a swivel angle limit value is exceeded, a warning is issued to a driver of the towing vehicle and / or a control command is issued to an actuator of the towing vehicle.

[0021] The edges can be determined, for example, using a line-fitting algorithm. The coupling point can be determined, for example, from a three-dimensional vehicle model stored in a computing device configured to implement the method and / or from the first camera image using an image recognition algorithm. Determining the lateral swivel angle between the towing vehicle and the trailer—i.e., the angle of rotation of the trailer around the coupling point, which can occur, for example, when the vehicle combination is cornering—from the at least one segment determined in the first camera image has the advantage that no additional devices or algorithms are required to determine this angle.

[0022] Advantageously, sensors that measure the swivel angle as well as additional evaluation of the first camera image and / or other image data can be dispensed with.

[0023] In a preferred embodiment of the invention, it can be provided that the respective intersection points of two lateral edges of the segment, which correspond to the lateral edges of the trailer, with a lower edge of the segment, which corresponds to a lower edge of the trailer, are determined, wherein the lateral pivot angle is determined from a projection of the intersection points and the coupling point onto a ground plane extending parallel to a roadway plane of the vehicle combination. It is assumed that the towing vehicle and the trailer are moving on the same roadway or on a common plane. Two intersection points can be determined from the lateral edges of the trailer, in particular from the lateral edges of a front side of the trailer facing the towing vehicle, and the lower edge of the trailer, which depend on the orientation of the trailer to the towing vehicle and thus on the lateral pivot angle.These intersection points and the coupling point can then be projected onto the ground plane. The swivel angle can be determined, for example, as the angle between a longitudinal direction of the towing vehicle and a perpendicular from the projection of the coupling point onto the straight line between the projections of the intersection points.

[0024] According to the invention, at least one further segment can be determined by segmenting the first camera image, wherein the further segment corresponds to a coupling section of the trailer. The representation of the coupling section of the trailer cannot be replaced by the camera image of the second, trailer-side camera, so that this continues to be displayed in the surrounding view. Furthermore, the representation of the coupling section also enables a driver to recognize the current angle between the towing vehicle and the trailer by depicting the coupling section in the surrounding view, even when the front surface of the trailer is displayed completely transparently.

[0025] In a preferred embodiment of the invention, it can be provided that the segmentation of the first camera image is carried out by a neural network and / or that the segmentation of the first camera image is carried out by semantic segmentation, in particular a content-guided network algorithm.

[0026] A description of the content-guided network algorithm can be found, for example, in the article "Cgnet: A lightweight context-guided network for semantic segmentation." IEEE Transactions on Image Processing, 2020, Vol. 30, pp. 1169-1179. However, the invention is not limited to this type of algorithm; other algorithms for semantic segmentation of image data can also be used to determine the at least one segment in the first camera image.

[0027] According to the invention, it can be provided that a camera arrangement is used which uses one or more third cameras, each of which captures a lateral surrounding area of ​​the towing vehicle and / or the trailer, wherein the surrounding view is additionally generated from one or more third camera images of the third cameras.

[0028] By using one or more third cameras, each of which captures part of the lateral surroundings of the vehicle combination, the surrounding view can be expanded so that areas to the side of and / or to the side behind the vehicle combination can also be imaged.

[0029] For a method according to the invention for training a neural network for segmenting an image, in particular for use in a method according to the invention, it is provided that it comprises the following steps:

[0030] - Providing at least one training data set comprising a plurality of training images, each of which depicts at least one frontal surface of a trailer directed towards a recording position of the training image as a ground truth,

[0031] - Assigning at least one segment to each of the training images, wherein the segment corresponds to the front surface of the trailer in the training image, and

[0032] - Optimizing the neural network with regard to a match between the at least one segment and the front surface of the trailer described by the ground truth.

[0033] A neural network trained according to the method described above can preferably be used in a method according to the invention for generating an environmental view of an environment located at least partially behind a trailer of a vehicle combination comprising the trailer and a towing vehicle. The neural network can be used to segment the at least one first camera image, so that the at least one segment of the first camera image comprising the front end of the trailer can advantageously be determined.

[0034] To train the neural network, a training dataset is used, which comprises a large number of training images. Each of the training images shows the frontal surface of a trailer directed toward a recording position of the respective training image, which represents a ground truth of the respective training image. The training images can, for example, be images taken by the rearview camera of the towing vehicle of a vehicle combination, in which a trailer of the vehicle combination can be seen.

[0035] At least one segment can then be assigned to each of these training images, corresponding to the frontal area of ​​the trailer depicted in the respective training image. The neural network can then be trained in such a way that the neural network's parameters are adjusted to ensure that the segments match the images of the trailer's frontal areas present as ground truth in the training images. The training images preferably include representations of different trailer types and / or different trailer loading states.

[0036] According to the invention, it can be provided that at least some of the training images depict a coupling section of the trailer as a further ground truth, wherein a further segment is assigned to the coupling section and the neural network is also optimized with regard to a match between the at least one further segment and the coupling section described by the further ground truth.

[0037] Because the training images are each provided with a coupling section of the

[0038] Since the ground pointing to the trailer was assigned, the parameters of the neural network can also be advantageously adjusted to ensure a match between the additional segments corresponding to the coupling section and the ground truth. This results in the neural network being trained to also detect coupling sections of trailers.

[0039] For a control device according to the invention, it is provided that it is designed to receive at least one first camera image and at least one second camera image, wherein the control device is set up to carry out a method according to the invention for generating an environmental view.

[0040] The control unit can be connected, in particular, to at least one first camera, at least one second camera, and optionally also to one or more third cameras. The connections between the control unit and the cameras can be implemented, for example, as wired or wireless communication connections. This allows the first camera images, the second camera images, and, if applicable, also the images from the third cameras to be transmitted to the control unit.

[0041] A vehicle according to the invention is provided to include a control unit according to the invention. The vehicle can be a motor vehicle, for example, a passenger car or a truck, or a trailer.

[0042] A computer program according to the invention comprises instructions which cause a computing device to carry out a method according to the invention. The computer program according to the invention can in particular be stored on a non-transient data carrier or can be retrieved and / or downloaded by a computing device via a communication connection, for example the Internet. The computing device can be, for example, a control unit of a vehicle or a vehicle-external computing device which communicates with such a control unit. All advantages and configurations described above in relation to the method according to the invention for generating an environmental view also apply accordingly to the method according to the invention for training a neural network, the control unit according to the invention, the vehicle according to the invention and the computer program according to the invention, and vice versa.

[0043] Further advantages and details of the invention will become apparent from the exemplary embodiments described below and from the drawings. These are schematic representations and show:

[0044] Fig. 1 shows a vehicle combination comprising an embodiment of a vehicle according to the invention,

[0045] Fig. 2 shows the relative arrangement of a first camera and a second camera,

[0046] Fig. 3 an example of a first camera image,

[0047] Fig. 4 shows the example of the first camera image after determining the at least one segment,

[0048] Fig. 5 is a graphical representation of a first step of determining a swivel angle between towing vehicle and trailer from the segmented first camera image,

[0049] Fig. 6 a graphical representation of a second step of determining the swivel angle and

[0050] Fig. 7 shows an embodiment of a method for training a neural network for image segmentation. Fig. 1 shows a vehicle combination 1. The vehicle combination 1 comprises an embodiment of a vehicle according to the invention as a towing vehicle 2 and a trailer 3 coupled to the towing vehicle 2. The towing vehicle 2 is designed as a motor vehicle, for example, as a passenger car or a truck.

[0051] A first camera 4 is arranged at the rear of the towing vehicle 2. The first camera 4 can, for example, be a rear-view camera of the towing vehicle 2. A second camera 5 is arranged at a rear of the trailer 3, which in this case faces away from the towing vehicle 2. The first camera 4 and the second camera 5 form a camera arrangement of the vehicle combination 1. The camera arrangement can optionally also comprise one or more third cameras 6, which, for example, record the surroundings of the vehicle 1 to the side of the towing vehicle 2 or to the side of the trailer 3. The cameras 4-6 of the camera arrangement are connected, for example wirelessly or via a wired communication connection, to a control unit 7, which is designed to carry out a method for generating an environmental view of an environment located at least partially behind the trailer 3 of the vehicle combination 1.

[0052] The first camera 4, arranged on the towing vehicle 2, captures at least a portion of the trailer 3 within a capture area 8 of the first camera 4, which is shown in dashed lines. The second camera 5 captures at least a portion of the surroundings behind the trailer 3 and has a capture area 9, also shown in dashed lines in Fig. 1. The third cameras 6, which are also optionally present, each have a capture area 10 that covers a portion of the surroundings to the side of the vehicle combination 1.

[0053] The trailer 3 comprises a coupling section 11, with which it is coupled to a coupling point 12, for example, its trailer coupling, of the towing vehicle 2. The trailer 3 can be pivoted at least laterally via the coupling section 11. In this context, "laterally pivotable" refers to pivoting of the trailer 3 relative to the towing vehicle 2 in a plane spanned by the vehicle's longitudinal direction x and the vehicle's transverse direction y.

[0054] To generate the surrounding view, the control unit 7 receives the camera images from the first camera 4, the second camera 5, and possibly also the third camera 6. From the received camera images, the control unit 7 generates a surrounding view that at least partially depicts the surroundings of the vehicle combination. The surrounding view can, for example, be displayed to a driver of the towing vehicle 2 via a display device (not shown) arranged inside the towing vehicle 2.

[0055] The control unit 7 is configured to receive, in a first step, at least a first camera image from the first camera 4 and at least a second camera image from the second camera 5. The control unit 7 then determines at least one segment in the first camera image 4 that corresponds to a front face of the trailer 3 facing the towing vehicle 2. The segment is determined by segmenting the first camera image, which will be described in more detail below. The control unit 7 then generates the surrounding view from the first camera image and the second camera image depending on the determined segment, with at least part of the segment being supplemented with image data from the second camera image.

[0056] In order to be able to supplement the segment with the image data of the second camera image, position information which describes the relative arrangement between the first camera 4 and the second camera 5 is stored in the control unit 7. Even when using a plurality of first cameras 4, a plurality of second cameras 5 and / or the optional use of one or more third cameras 6, the respective relative, initial arrangements of the cameras 4, 5, 6 to one another can be stored in the form of position information in order to enable the respective image data from the camera images to be combined as seamlessly as possible. Fig. 2 shows the position information schematically for the first camera 4 and the second camera 5. The position information includes, for example, for the first camera 4 a translation vector T and a rotation vector R, which describe the translational or rotational position.Describe the rotational arrangement of the first camera 4 relative to the coupling point 12 of the trailer 3. Accordingly, the position information for the second camera 5 describes the translational offset of the second camera 5 relative to the coupling point 12 with a translation vector T' and the rotational offset relative to the coupling point 12 with a rotation vector R'.

[0057] The respective components x, y, z of the vectors T, T' and the components a, b, c of the vectors R and R' can initially be determined, for example, by a measurement and / or based on image data acquired by the first camera 4 and / or the second camera 5. During driving operation of the vehicle combination 1, the vectors and their components can be at least partially redetermined in order to be able to display the surrounding view at least approximately correctly in real time. Corresponding vectors can also be determined accordingly for additional first cameras 4, additional second cameras 5 and / or optionally present one or more third cameras 6.

[0058] To explain the segmentation of a first camera image, Fig. 3 schematically shows a first camera image 13, which was recorded, for example, with a first camera 4 comprising a fisheye lens. The first camera image 13 shows a partial area of ​​the surroundings of the vehicle combination 1 behind the towing vehicle 2, wherein a front end of the trailer 3 facing the towing vehicle can also be seen. Furthermore, the coupling section 11, the coupling point 12 and other features of the surroundings are also depicted in the first camera image 13. All images in the first camera image 13 exhibit the distortion characteristic of a fisheye lens.

[0059] The first camera image 13 is then segmented by a segmentation algorithm implemented in the control unit 7. An example of a segmented camera image 14 is shown in Fig. 4. With the aid of the segmentation, at least one segment 15 of the first camera image 13 corresponding to the front face of the trailer 3 facing the towing vehicle 2 is determined. The size and geometry of the segment 15 depend on the actual dimensions of the trailer 3. In addition to a structure of the trailer 3, the segment 15 can also comprise a load of the trailer 3 arranged uncovered on the trailer 3, so that different types of trailers and / or different loading states of a trailer can be advantageously recognized and overlaid with image data from a second camera image of the second camera 5 when generating the surroundings view. In this way, an surroundings view can be advantageously generated in which the trailer 3 orthe front surface of the trailer 3 facing the towing vehicle 2 appears transparent.

[0060] By segmenting the first camera image 13, at least one further segment 16 is determined, wherein the further segment 16 corresponds to the coupling section 11 of the trailer 3. The background of the first camera image 13 forms a background segment 17. The surrounding view is then generated by replacing the entire segment 15 or at least part of the segment 15 with image data from a second camera image generated using the second camera 5. For example, the image data of the first camera image 13 can remain in the background segment 17, so that with the help of the image data of the second camera image inserted into the segment 15, a rear view from the towing vehicle 2 is obtained which is not blocked by the trailer 3.

[0061] To generate the transparent view, the segment 15 corresponding to the front face of the trailer 3 in the first camera image 13 is supplemented entirely or at least partially with image data from the second camera image. The image area outside of the segment 15, i.e. the further segment 16 and the background segment 17, can in particular continue to contain the image data from the first camera image, since this depicts the surroundings of the vehicle combination 1, which in these areas is not obscured by the trailer 3. To supplement the segment 15 with the image data from the second camera image, the second camera image can be edited, e.g. by cropping and rotating or by rotational and / or translational transformation.Additionally or alternatively, further changes and / or processing can be carried out on the second camera image, for example a projection onto a plane corresponding to the segment 15 in the first camera image 13 and / or comparable transformations.

[0062] The segmentation of the first camera image 13 is performed, for example, by a neural network implemented in the control unit 7. The segmentation can be performed, for example, by semantic segmentation, in particular a content-guided network algorithm.

[0063] The shape of the segment 15 depends not only on the geometry of the trailer 3 and any load transported on the trailer 3, but also on the lateral swivel angle of the trailer 3, i.e. the angle at which the trailer is swiveled about the coupling point 12. The replacement of at least part of the segment 15 in the first camera image 13 with image data from the second camera image, which was described above, takes place in particular depending on the swivel angle of the trailer 3. This allows the image data suitable for the current driving situation of the vehicle combination 1 to be selected from the second camera image for a seamless representation of the surroundings in the surroundings view and inserted into the segment 15. By segmenting the first camera image 13, the lateral swivel angle between the towing vehicle 2 and the trailer 3 can also be advantageously determined.

[0064] Fig. 5 graphically illustrates the first step of determining the swivel angle from the segmented first camera image 14. The lateral swivel angle can be determined depending on the position of one or more edges of the at least one segment 15 in the first camera image 13 and a position P1 of the coupling point 12 at which the trailer 3 is laterally pivotably coupled to the towing vehicle 2. For this purpose, the respective intersection points P2, P3 of two lateral edges 18, 19 of the segment 15, which correspond to the lateral edges of the trailer 3, with a lower edge 20 of the segment 15, which corresponds to the lower edge of the trailer 3, are determined. The points P1, P2 and P3 are then projected onto a ground plane extending parallel to a roadway plane of the vehicle combination 1.

[0065] Fig. 6 shows a projection of the intersection points P2, P3 and the position P1 of the coupling point 12 onto the ground plane extending in the vehicle longitudinal direction x and in the vehicle transverse direction y, wherein the projections of the points P1, P2 and P3 onto the ground plane are designated P1', P2' and P3'. The lateral swivel angle a can then be determined as the angle between the longitudinal direction x of the towing vehicle 2 and a perpendicular 21 which is dropped through the point P1' onto a straight line 22 between the projections P2', P3' of the intersection points P2, P3.

[0066] The surrounding view can be generated, in particular, depending on the lateral swivel angle a. Additionally or alternatively, if the determined swivel angle a exceeds a swivel angle limit, it is possible to issue a warning to a driver of the towing vehicle 2 and / or to issue a control command to an actuator of the towing vehicle 2.

[0067] Fig. 7 shows an embodiment of a method for training a neural network for image segmentation. The method comprises the following steps:

[0068] - Providing (S1) at least one training data set comprising a plurality of training images, each of which depicts at least one frontal surface of a trailer 3 directed towards a recording position of the training image as a ground truth,

[0069] - Assigning (S2) at least one segment 15 to each of the training images, wherein the segment 15 corresponds to the front face of the trailer 3 in the training image, and - Optimizing (S3) the neural network with regard to a match between the at least one segment 15 and the front face of the trailer 3 described by the ground truth. It is possible for at least some of the training images to depict the coupling section 11 of the trailer 3 as a further ground truth, wherein a further segment 16 is assigned to the coupling section 11 and the neural network is also optimized with regard to a match between the at least one further segment 16 and the coupling section 11 described by the further ground truth.

Claims

Patent claims 1 . Method for generating an environmental view of an environment located at least partially behind a trailer (3) of a vehicle combination (1) comprising the trailer (3) and a towing vehicle (2) from camera images obtained with the aid of a camera arrangement, wherein the camera arrangement has at least one first camera (4) capturing at least a portion of the trailer and at least one second camera (5) capturing at least a partial area of ​​the environment behind the trailer, comprising the steps: - capturing at least one first camera image (13) with the at least one first camera (4) and at least one second camera image with the second camera (5), - Determining at least one segment (15) of the first camera image (13) corresponding to a front face of the trailer (3) directed towards the towing vehicle (2) in the first camera image (13) by segmenting the first camera image (13) and - generating the environmental view from the first camera image (13) and the second camera image depending on the determined segment (15), wherein at least a part of the segment (15) is supplemented with image data from the second camera image.

2. Method according to claim 1, characterized in that depending on the position of one or more edges (18, 19, 20) of the at least one segment (15) in the first camera image (13) and the position (P1) of a coupling point (12) at which the trailer (3) is laterally pivotably coupled to the towing vehicle (2), a lateral pivot angle of the trailer (3) is determined, wherein the surrounding view is generated depending on the lateral pivot angle and / or wherein when a swivel angle limit value, a warning is issued to a driver of the towing vehicle (2) and / or a control command is issued to an actuator of the towing vehicle (2).

3. Method according to claim 2, characterized in that the respective intersection points (P2, P3) of two lateral edges (18, 19) of the segment (15), which correspond to the lateral edges of the trailer (3), with a lower edge of the segment (15), which corresponds to a lower edge (20) of the trailer (3), are determined, wherein the lateral pivot angle is determined from a projection of the intersection points (P2, P3) and the position (P1) of the coupling point (12) onto a base plane extending parallel to a roadway plane of the vehicle combination (1).

4. Method according to one of the preceding claims, characterized in that by segmenting the first camera image (13) at least one further segment (16) is determined, wherein the further segment (16) corresponds to a coupling section (11) of the trailer (3).

5. Method according to one of the preceding claims, characterized in that the segmentation of the first camera image (13) is carried out by a neural network and / or that the segmentation of the first camera image (13) is carried out by semantic segmentation, in particular a content-guided network algorithm.

6. Method according to one of the preceding claims, characterized in that a camera arrangement is used which uses one or more third cameras (6), each of which captures a lateral surrounding area of ​​the towing vehicle (2) and / or the trailer (3). wherein the environmental view is additionally generated from one or more third camera images of the third cameras (6). A method for training a neural network for the segmentation of an image, in particular for use in a method according to one of the preceding claims, comprising the steps: - Providing at least one training data set comprising a plurality of training images, each of which depicts at least one frontal surface of a trailer (3) directed towards a recording position of the training image as a ground truth, - Assigning at least one segment (15) to the training images, wherein the segment (15) corresponds to the front surface of the trailer (3) in the training image, and - Optimizing the neural network with regard to a match between the at least one segment (15) and the end face of the trailer (3) described by the ground truth. Method according to claim 7, characterized in that at least some of the training images depict a coupling section (11) of the trailer (3) as a further ground truth, wherein a further segment (16) is assigned to the coupling section (11), and the neural network is also optimized with regard to a match between the at least one further segment (16) and the coupling section (11) described by the further ground truth. Control unit configured to receive at least one first camera image (13) and at least one second camera image, wherein the control unit (7) is set up to carry out a method according to one of claims 1 to 6. Vehicle comprising a control unit (7) according to claim 9. Computer program comprising instructions which cause a computing device to carry out a method according to one of claims 1 to 8.

Images