Object tracking and at least partially automated guidance of a vehicle

Abstract

For object tracking, first and second sensor data generated by means of a surroundings sensor system (3) are obtained. On the basis of the first sensor data, a plurality of first geometric figures (16a, 16b, 16c, 16d, 25) are determined, which each approach a region of at least one object (5, 14, 20, 22, 24) represented by the first sensor data, and on the basis of the second sensor data, a plurality of second geometric figures (16a, 16b, 16c, 16d, 25) are determined, which each approach a region of at least one object (5, 14, 20, 22, 24) represented by the second sensor data. One of the second geometric figures (16a, 16b, 16c, 16d, 25) is selected as a feature to be tracked, and the feature to be tracked is assigned to a reference figure of the first geometric figures (16a, 16b, 16c, 16d, 25), which reference figure matches the feature to be tracked, said assignment being carried out on the basis of a comparison of the plurality of second geometric figures (16a, 16b, 16c, 16d, 25) with the plurality of first geometric figures (16a, 16b, 16c, 16d, 25).

Description

[0001] 2023PF02450

[0002] 1

[0003] Object tracking and at least partially automated vehicle control

[0004] The present invention relates to a computer-implemented method for object tracking, wherein first sensor data corresponding to a first detection period and second sensor data corresponding to a second detection period are obtained by means of an environmental sensor system of a vehicle, each representing an environment of the vehicle. The invention further relates to a method for at least partially automatic guidance of a vehicle, wherein such a computer-implemented method is carried out, a data processing system for carrying out said computer-implemented method, an electronic vehicle guidance system with such a data processing system, and a corresponding computer program product.

[0005] Environmental sensor systems, for example active optical sensor systems, especially lidar systems such as laser scanners or flash lidar systems, can be mounted on vehicles and used to implement a wide variety of driver assistance functions and / or other functions for partially automated vehicle control. In particular, distances to objects in the vehicle's vicinity can be determined, for example, by measuring the time of flight of light using active optical sensor systems of this type, and these distances can be used for driver assistance or other functions for at least partially automated vehicle control.

[0006] Object tracking involves continuously or cyclically determining the position, and potentially other state variables such as speed, yaw rate, and so on, of an object over time. These tracked state variables can then be used as the basis for a wide variety of driver assistance functions or other functions for at least partially automated vehicle control.

[0007] Well-known object tracking algorithms are based on Kalman filters or methods derived from them. These approaches share the common feature of relying on a predefined model that allows the prediction of the object's movement. This prediction is then refined or corrected through measurements. One disadvantage (2023PF02450)

[0008] 2. The problem is that tracking an object that does not behave according to the model cannot be done, or cannot be done reliably.

[0009] It is an object of the present invention to provide a means of object tracking in a vehicle environment that is independent of a model for the movement of the object.

[0010] This problem is solved by the subject matter of the independent claim. Advantageous further developments and preferred embodiments are the subject matter of the dependent claims.

[0011] The invention is based on the idea of ​​generating a symbolic representation of the environment for each of the various acquisition periods, also referred to as frames, of the environmental sensor system. This representation contains a multitude of geometric figures, each approximating an area of ​​at least one object represented by the corresponding sensor data. By comparing the symbolic representations of different acquisition periods, a feature to be tracked, defined by one of the geometric figures, can be tracked by assigning it to one of the geometric figures from a previous acquisition period.

[0012] According to one aspect of the invention, a computer-implemented method for object tracking is described. In this method, first sensor data, generated by an environmental sensor system of a vehicle and corresponding to a first detection period, and second sensor data, generated by an environmental sensor system and corresponding to a second detection period, are obtained. The first and second sensor data each represent an environment of the vehicle, in particular three-dimensionally. Based on the first sensor data, a plurality of first geometric figures are determined, each of which approximates a region of at least one object represented by the first sensor data. Based on the second sensor data, a plurality of second geometric figures are determined, each of which approximates a region of at least one object represented by the second sensor data.A second geometric figure from the set of second geometric figures is selected as the feature to be tracked. Depending on a comparison of the set of second geometric figures with the set of first geometric figures, and specifically depending on the result of the comparison, the feature to be tracked is assigned to a reference figure from the first set of geometric figures that matches the feature to be tracked. 2023PF02450.

[0013] 3

[0014] Unless otherwise specified, all steps of the computer-implemented method can be performed by a data processing system comprising at least one data processing device, in particular by a data processing system of the vehicle. Specifically, the at least one data processing device is configured or adapted to perform the steps of the computer-implemented method. For this purpose, the at least one data processing device may, for example, store a computer program containing instructions which, when executed by the at least one data processing device, cause it to execute the computer-implemented method. The terms "data processing system" and "at least one data processing device" may be used interchangeably.

[0015] All data processing devices of the at least one data processing device can be part of the vehicle. However, it is also possible that all data processing devices of the at least one data processing device are part of an external computing system outside the vehicle, for example, a mobile electronic device, a backend server, or a cloud computing system. It is also possible that the at least one data processing device comprises both at least one vehicle data processing device of the vehicle and at least one external data processing device of the external computing system.The at least one vehicle data processing device may, for example, comprise one or more electronic control units (ECUs), and / or one or more zone control units (ZCLIs), and / or one or more domain control units (DCLIs) of the vehicle and / or the environmental sensor system.

[0016] In the event that the at least one data processing device comprises two or more data processing devices, certain steps performed by the at least one data processing device may, for example, be understood as different data processing devices performing different steps or different parts of a step. In particular, it is not necessary for each data processing device to perform the steps completely. In other words, the performance of the steps may be distributed among the two or more data processing devices. 2023PF02450

[0017] 4

[0018] Each embodiment of the computer-implemented method results in a corresponding embodiment of a method for characterizing the at least one object that is not purely computer-implemented, by including corresponding steps for generating the sensor data using the active optical sensor system.

[0019] The environmental sensor system is, for example, a vehicle's environmental sensor system, particularly a motor vehicle such as a car, van, truck, motorcycle, and so on. The environmental sensor system is specifically mounted on the vehicle, so the sensor data, which represent the environment of the environmental sensor system, constitutes the vehicle's environment, specifically its external environment. In other words, the environment of the environmental sensor system is the vehicle's environment.

[0020] The environmental sensor system can, in particular, be an active optical sensor system. By definition, an active optical sensor system has a light source for emitting light or light pulses. The light source can, in particular, be a laser, for example, an infrared laser. Furthermore, by definition, an active optical sensor system, in this case the detector array, has at least one optical detector to detect reflected components of the emitted light. The active optical sensor system is, in particular, configured to generate and process or output one or more sensor signals based on the detected components of the light.

[0021] Here and in the following, the term "light" can be understood to encompass electromagnetic waves in the visible, infrared, and / or ultraviolet ranges. Accordingly, the term "optical" can also be understood to refer to light as defined in this way.

[0022] It should be noted that a single detector pixel of the detector array does not necessarily consist of a single optical detector. Rather, it is also possible for a group of several adjacent optical detectors to form a detector pixel. The latter is particularly possible when using single-photon avalanche diodes (SPADs). In other embodiments, however, it is also possible for a pixel to consist of exactly one optical detector, for example, a single photodiode or avalanche photodiode (APD). 2023PF02450

[0023] 5

[0024] The active optical sensor system is specifically designed as a lidar sensor system, for example as a laser scanner.

[0025] Laser scanners are a well-known type of lidar sensor system in which a laser beam is deflected by a light deflection device, allowing for various deflection angles. The light deflection device can, for example, contain a rotatably mounted mirror. Alternatively, the light deflection device can have a mirror element with a tiltable and / or swiveling surface. The mirror element can be designed, for example, as a microelectromechanical system (MEMS). In the environment, the emitted laser beams can be partially reflected, and the reflected portions can then strike the laser scanner, specifically the light deflection device, which can direct them onto the detector array of the laser scanner. Each detector pixel of the detector array, or each optical detector of the detector array, generates a corresponding detector signal based on the respective detected portions.Based on the spatial arrangement of each detector pixel, together with the current position of the light deflection device, in particular its rotational or tilting and / or swiveling position, the direction of incidence of the detected reflected components can be determined. An evaluation unit can also, for example, perform a time-of-flight measurement to determine the radial distance of the reflecting object. Alternatively or additionally, a method can be used to determine the distance by evaluating the phase difference between emitted and detected light.

[0026] Other types of lidar sensor systems are flash lidar sensor systems. These are non-scanning systems that do not require such a light deflection arrangement. Instead, the laser light generated by the light source is scattered by an optical element, so that it is emitted in a single flash over a wide angle.

[0027] Typically, data generated by an active optical sensor system can be represented as a point cloud, which is understood as a set of points, each identified by corresponding coordinates in a two- or three-dimensional coordinate system. In the case of a three-dimensional point cloud, the three-dimensional coordinates could be, for example, 2023PF02450

[0028] 6. The coordinates of a point cloud can be determined by the direction of incidence of the reflected light components and the corresponding light travel time or radial distance measured for that particular point. However, the information can also be preprocessed to obtain three-dimensional Cartesian coordinates for each point. In general, the points in a point cloud can be displayed in an unordered or unsorted manner, unlike, for example, a camera image. In addition to the spatial information, namely the two- or three-dimensional coordinates, the point cloud can also contain additional information or measurements for each point, such as the echo pulse width (EPW) of the respective sensor signal.

[0029] The data generated by the active optical sensor system during a frame can be interpreted as a 2.5-dimensional point cloud, provided the radial spacing is appropriately determined. This means that while a three-dimensional position is stored for each point in the point cloud, the entire three-dimensional space is not represented within the active optical sensor system's field of view. This is because the emitted light pulses only reach the side of objects in the environment facing the active optical sensor system. Therefore, parts of the objects behind this side, or other objects or parts thereof obscured by the objects, are not represented by the point cloud, just as with a camera image. Thus, even with a 2.5-dimensional point cloud, the sensor data can potentially represent the environment in three dimensions.

[0030] The view of data generated by an active optical sensor system as a point cloud stems, among other things, from the fact that the detector arrays of earlier active optical sensor systems had only a very small number of detector pixels, for example, two to four detector pixels arranged in a row. In modern active optical sensor systems, however, this is sometimes different, as the detector pixels are arranged in a large number of rows and columns, for example, in 100 to 1,000 rows and / or 100 to 1,000 columns. As a result, a point cloud of a frame can also be viewed or displayed as an image, where the number of image pixels corresponds to the number of detector pixels, and the spatial two-dimensional arrangement of the detector pixels defines the corresponding image plane and directly corresponds to the arrangement of the image pixels.The respective pixel values ​​of the image can be given by the radial distance of the reflecting point, which was measured using the corresponding detector pixel, or by the corresponding depth, which corresponds to the perpendicular distance of the reflecting point from the detector pixel and is calculated from the radial distance 2023PF02450.

[0031] 7. The image can then be described as a depth image, in particular a monochromatic depth image.

[0032] However, it is also possible, for example, to generate corresponding images, especially monochromatic images, based on other parameters instead of depth, such as the EPW or the light intensity of the corresponding reflected components as detected by the detector pixel. It is also possible to use the detector array to capture ambient light without emitting light pulses and detecting reflected components, essentially like a 2D camera, particularly a 2D infrared camera.

[0033] The multitude of initial geometric figures can be understood, in particular, as a symbolic representation of the environment, or a part of the environment, for the first recording period. Similarly, the multitude of secondary geometric figures can be understood, in particular, as a symbolic representation of the environment, or a part of the environment, for the second recording period. These are symbolic representations in the sense that a symbolic representation does not necessarily contain a realistic image of the at least one object, but rather the area and the further areas of the at least one object are symbolically represented by the geometric figure and the further geometric figure(s).

[0034] The second recording period, for example, occurs after the first recording period. The first and second recording periods can be immediately consecutive, but this is not mandatory. It is also possible that one or more further recording periods occur between the first and second recording periods, particularly if there is a correlation between the first and second recording periods, meaning the number of further recording periods is small, for example, in the range of one to five. The at least one object represented by the first sensor data partially corresponds to the at least one object represented by the second sensor data. However, a complete match is not necessarily given.In principle, it is also possible that there is no partial agreement, which would mean that the comparison of the multitude of second geometric figures with the multitude of first geometric figures does not yield 2023PF02450.

[0035] 8

[0036] This would allow the assignment of the feature to be tracked to one of the first geometric figures as a reference figure.

[0037] It should also be noted that the reference figure is not predetermined, but is identified through comparison. The fact that the reference figure corresponds to the feature being tracked can be understood, for example, as follows: the similarity of the reference figure to the feature being tracked, according to a given similarity measure, is greater than a given minimum similarity measure; or the deviation of the reference figure from the feature being tracked, according to a given deviation measure, is less than a given maximum deviation measure; or the reference figure corresponds to the feature being tracked within a given tolerance. In particular, it is not necessarily required that a sufficiently corresponding first geometric figure exists for all second geometric figures.It is also not necessarily required that the number of the second geometric figures equals the number of the first geometric figures. The similarity measure, the measure of deviation, or the tolerance can accordingly also be defined for partial similarity.

[0038] A geometric figure can, in this and subsequent discussion, be a curve or a surface in three-dimensional space. The curve or surface is fundamentally defined in three-dimensional space. However, this does not preclude the possibility that, in some cases, the curve or surface lies in a two-dimensional plane. This two-dimensional plane is not necessarily parallel to the image plane but can, for example, extend wholly or partially in the depth direction.

[0039] For example, the type of geometric figure and / or at least one other geometric figure can be specified. It can be specified, for instance, that the geometric figure and / or at least one other geometric figure are straight lines, segments of circular arcs, semicircles, segments of elliptical arcs, arcs of ellipses, parts of a sphere's surface or an ellipsoid's surface, and so on.

[0040] The first geometric figures are all given in a common coordinate system. In other words, the symbolic representation for the first recording period includes not only the shapes of the first geometric figures 2023PF02450

[0041] 9

[0042] The symbolic representation for the second recording period includes not only the shapes of the second geometric figures, but also, explicitly or implicitly, their relative positions, in particular the position of the feature to be tracked with respect to the other first geometric figures within the multitude of first geometric figures, or vice versa. The second geometric figures are also all given in the common coordinate system. In other words, the symbolic representation for the second recording period includes not only the shapes of the second geometric figures, but also, explicitly or implicitly, their relative positions, in particular the position of the reference figure with respect to the other second geometric figures within the multitude of second geometric figures, or vice versa.

[0043] Mathematically, the symbolic representations, i.e., the multitude of first geometric figures and the multitude of second geometric figures, can each be represented as a vector, for example, where entries of the vector contain or are derived from corresponding properties or parameters of the first geometric figures or the second geometric figures.

[0044] A region of the at least one object can, for example, be an edge of the at least one object. An edge can be approximated, in particular, by a curve as a geometric figure. A region of the at least one object can also be a surface region, in particular a surface region of the at least one object. A surface region can be approximated, in particular, by a surface as a geometric figure.

[0045] The at least one object can be exactly one object. The at least one object can also contain two or more objects. Each first geometric figure of the plurality of first geometric figures approximates, in particular, a different region of the at least one object, and each second geometric figure of the plurality of second geometric figures approximates, in particular, a different region of the at least one object.

[0046] The result of comparing the multitude of second geometric figures with the multitude of first geometric figures includes, in particular, the information as to whether one of the first geometric figures can serve as a reference figure for the feature to be tracked, and if so, which of the first geometric figures. However, the comparison of the multitude of second geometric figures with the multitude of first geometric figures is not limited to a comparison of the feature to be tracked with the multitude of first geometric figures. Rather, in some cases, 2023PF02450

[0047] 10

[0048] Embodiments also include other second geometric figures and, for example, their position with respect to the feature to be tracked with the multitude of first geometric figures, which in turn can allow a more reliable assignment of the feature to be tracked to the reference figure.

[0049] The assignment of the tracked feature to the reference figure can be understood as a result of the computer-implemented object tracking procedure, since the task of object tracking essentially consists of determining how an object or a part of the object—here, the area of ​​the at least one object approximated by the tracked feature—behaves over different recording periods. From this assignment, it can therefore be deduced how the state of the tracked feature, including, for example, its position, orientation, and / or velocity, has changed from the first recording period to the second. This assignment can be stored, in particular, in a computer-readable format.

[0050] The selection of the feature to be tracked can be random or according to a predefined rule. In particular, it is possible to perform the described steps of the computer-implemented object tracking procedure for the first and second recording periods for several of the multiple second geometric figures as the feature to be tracked, for example, for all second geometric figures in the multiple second geometric figures. This allows multiple objects or parts of objects to be tracked individually. Alternatively or additionally, it is also possible to repeat the described steps of the computer-implemented object tracking procedure, especially multiple times, adjusting the recording periods accordingly.

[0051] Since the assignment of the feature to be tracked to the reference figure is carried out by comparing geometric figures, the type of movement of the feature to be tracked is irrelevant or largely irrelevant for the computer-implemented object tracking method according to the invention. In particular, no model is required that can approximately predict the movement.

[0052] A further advantage of the computer-implemented method according to the invention is that not an entire object is tracked as such, but rather an area or, if applicable, several areas of the at least one object are tracked individually. 2023PF02450

[0053] 11. Accordingly, the partial occlusion of the at least one object is of less relevance for the computer-implemented method according to the invention than for conventional object tracking algorithms. In particular, when using the computer-implemented method according to the invention, a prior object classification or semantic segmentation can be dispensed with, which would be resource-intensive and, in some cases, for example with partial occlusion, would not be reliably possible.

[0054] According to at least one embodiment, to assign the feature to be tracked to the reference figure, the reference figure is identified by comparing the feature to be tracked with the plurality of first geometric figures and, in particular thereafter, for a further second geometric figure of the plurality of second geometric figures, i.e., a second geometric figure of the plurality of second geometric figures which is not the feature to be tracked, a further first geometric figure of the plurality of first geometric figures, i.e., a first geometric figure of the plurality of first geometric figures which is not the reference figure, is identified, the shape and relative position of which corresponds to the shape and relative position of the further second geometric figure with respect to the feature to be tracked.

[0055] In other words, the reference figure is only assigned to the feature to be tracked if there is a reference figure that corresponds to the feature to be tracked, and if, in addition, two further figures from the plurality of first geometric figures or the plurality of second geometric figures correspond with respect to their shape and with respect to their respective position with respect to the reference figure or with respect to the feature to be tracked.

[0056] The identification of the reference figure can be achieved, for example, by comparing the feature to be tracked with the multitude of first geometric figures, particularly according to the aforementioned similarity measure, the deviation measure, or the aforementioned tolerance. However, by not only selecting the correspondence of the feature to be tracked with the reference figure as a condition for assignment, the context of the feature to be tracked with respect to the other second geometric figures is also effectively taken into account, thus enabling a more reliable and less error-prone assignment. 2023PF02450

[0057] 12

[0058] In some embodiments, the verification regarding shape conformity and conformity in relative position can also be carried out for several further second geometric figures, in particular with the exception of the feature to be tracked itself, for all second geometric figures of the plurality of second geometric figures, thereby further increasing the reliability.

[0059] In particular, in some embodiments, at least for some of the remaining second geometric figures of the plurality of second geometric figures, a further first geometric figure of the plurality of first geometric figures can be identified whose shape and relative position with respect to the reference figure corresponds to the shape and relative position of the respective second geometric figure with respect to the feature to be tracked.

[0060] For example, it may be stipulated that the reference figure is assigned to the characteristic being tracked if the aforementioned identification is possible or successful.

[0061] According to at least one embodiment, the plurality of first geometric figures and the plurality of second geometric figures each include a corresponding curve in three-dimensional space.

[0062] In other words, every first geometric figure of the multitude of first geometric figures and every second geometric figure of the multitude of second geometric figures each contains a curve in three-dimensional space.

[0063] This can reduce the required storage space and computing power.

[0064] In particular, the respective area approximated by a first geometric figure is an edge of the at least one object represented by the first sensor data, and the respective area approximated by a second geometric figure is an edge of the at least one object represented by the second sensor data.

[0065] According to at least one embodiment, the plurality of second geometric figures is transformed such that a transformed starting point of the curve of the feature to be tracked and / or a transformed endpoint of the curve of the feature to be tracked 2023PF02450

[0066] 13

[0067] The features lie at predefined positions of the common coordinate system. After identifying the reference figure, the multitude of initial geometric figures is transformed such that a transformed starting point of the curve of the reference figure and / or a transformed endpoint of the curve of the reference figure lie at the predefined positions of the common coordinate system.

[0068] The comparison of the set of second geometric figures with the set of first geometric figures is performed based on the transformed set of second geometric figures and the transformed set of first geometric figures. In particular, the further first geometric figure of the set of first geometric figures is identified based on the transformed set of second geometric figures and the transformed set of first geometric figures.

[0069] Because, for example, the type of curve is defined, the path of each curve between its start and end points is not arbitrary. Representing the curve based on its start and end points thus allows for a compact and simple symbolic representation of at least one object.

[0070] In some embodiments, the representation of the respective curve is generated not only based on the aforementioned start and end points, but also on an intermediate point along the curve. This allows the curve's path between the start and end points to be further restricted or, depending on the type of curve, even uniquely defined. This reduces the likelihood of ambiguous interpretation.

[0071] In some embodiments, particularly with a semi-ellipse or semicircle as a curve or further curve, the respective intermediate point can, for example, lie centrally or symmetrically between the starting point and the endpoint. For example, the respective intermediate point can correspond to a vertex of the respective semi-ellipse or semicircle.

[0072] The transformation may include, in particular, a rotation and / or a translation and / or a scaling.

[0073] In an exemplary embodiment, the transformed starting point or the transformed endpoint of the curve of the feature to be tracked lies at the origin of the common coordinate system and / or the transformed 2023PF02450

[0074] 14

[0075] The starting point or the transformed endpoint of the curve of the reference figure lies at the origin of the common coordinate system.

[0076] According to at least one embodiment, to determine a first geometric figure from the plurality of first geometric figures, the corresponding area of ​​the at least one object represented by the first sensor data, in particular the corresponding edge, is approximated by a polyline, and the polyline is approximated by the corresponding curve. This applies in particular to each first geometric figure from the plurality of first geometric figures. Alternatively or additionally, to determine a second geometric figure from the plurality of second geometric figures, the corresponding area of ​​the at least one object represented by the second sensor data, in particular the corresponding edge, is approximated by a polyline, and the polyline is approximated by the corresponding curve. This applies in particular to each second geometric figure from the plurality of second geometric figures.

[0077] The respective edge itself can be identified, for example, by applying an edge detection algorithm, especially one that is known per se, such as a threshold-based edge detection algorithm.

[0078] For example, starting from a result of the edge detection algorithm, the polyline that approximates the respective edge is determined, and then the curve is determined so that it approximates the polyline.

[0079] A polyline can be understood as two or more line segments, each connected at one end. The polyline can, for example, be a polynomial chain, also known as a spline. In some embodiments, the individual line segments can be straight lines, i.e., first-degree polynomial segments or linear polynomial segments. In this case, the polyline can also be called a traverse. The term traverse does not imply, in particular, that the polyline is a closed polyline.

[0080] For example, the curve can be a smooth curve, meaning it can be differentiable any number of times except for a corresponding starting and ending point. This simplifies processing. On the other hand, the polyline, especially with straight lines as line segments, is generally not differentiable at the connection points. 2023PF02450

[0081] 15 differentiable. The polyline may allow for a more precise approximation of the edge.

[0082] According to at least one embodiment, the polyline is determined as a polygonal chain.

[0083] For example, the polyline is defined as a polygonal path and the curve as a semi-ellipse.

[0084] This allows for a computationally efficient yet comparatively accurate approximation of the object.

[0085] According to at least one embodiment, the corresponding area of ​​the at least one object represented by the first sensor data is approximated by two or more line segments, and the two or more line segments are approximated by the corresponding curve. This applies in particular to each first geometric figure of the plurality of first geometric figures. Alternatively or additionally, the corresponding area of ​​the at least one object represented by the second sensor data is approximated by two or more line segments, and the two or more line segments are approximated by the corresponding curve. This applies in particular to each second geometric figure of the plurality of second geometric figures.

[0086] The two or more line segments differ from a polyline in that they are not necessarily connected to each other.

[0087] In particular, the two or more line segments can each be polynomial segments, for example, straight line segments. The above explanations and embodiments regarding the polyline can be applied analogously to corresponding embodiments with two or more line segments.

[0088] According to at least one embodiment, the first sensor data and the second sensor data are each generated by an active optical sensor system of the vehicle. In other words, the environmental sensor system is an active optical sensor system.

[0089] According to at least one embodiment, the active optical sensor system comprises a two-dimensional detector array with a plurality of detector pixels. The initial sensor data includes a first depth image of the 2023PF02450 generated by the detector array.

[0090] 16

[0091] The environment, or the first depth image, is generated based on the initial sensor data. The second sensor data includes a second depth image of the environment, generated by the detector array, or the second depth image is generated based on the second sensor data. Both the first and second depth images contain a multitude of image pixels, where each image pixel corresponds to one of the detector pixels, and a pixel value of each image pixel corresponds to a distance from the detector array determined by the corresponding detector pixel. The multitude of first geometric figures is determined based on the first depth image, and the multitude of second geometric figures is determined based on the second depth image.

[0092] The detector pixels are arranged in a detector plane, for example as a multitude of rows and a multitude of columns, which corresponds to the image plane of the depth image. The distance from the detector array can, for example, be a distance perpendicular to the detector plane.

[0093] In some embodiments, filtering of the data generated by the detector array can also be carried out to create the respective depth image, for example noise filtering and / or filtering to remove certain objects or other components of the data that are not considered relevant.

[0094] If the image plane of the depth image is defined by a transverse direction y and a vertical direction z perpendicular to it, then the depth is defined along a longitudinal direction x perpendicular to both y and z. The rows of the detector array and the depth image are then arranged side by side parallel to the vertical direction z, and the columns of the detector array and the depth image are arranged side by side parallel to the transverse direction y. The three-dimensional space is defined by x, y, and z.

[0095] A geometric figure, for example a curve, and the correspondingly approximated area, for example the corresponding edge, of at least one object lie in three-dimensional space, and therefore not necessarily in a plane parallel to the xz-plane.

[0096] The area, in particular the edge, of the at least one object is defined by the two-dimensional position of the image pixels belonging to the area in the image plane in combination with their pixel values, which correspond to the depth. The geometric figure, in particular the curve, is then determined, for example, such that it defines the 2023PF02450

[0097] 17. The respective area is approximated. The geometric figure can be determined, for example, by defining a basic shape or type of geometric figure and adjusting the parameters that remain free so that the geometric figure approximates the area, for example, by regression or optimization in a known manner. The area itself can be identified, in particular, by applying an edge detection algorithm, especially one known per se, such as a threshold-based edge detection algorithm.

[0098] According to at least one embodiment, the respective curve is a plane curve. In other words, the first geometric figures of the plurality of first geometric figures each include a corresponding plane curve in three-dimensional space, and the second geometric figures of the plurality of second geometric figures each include a corresponding plane curve in three-dimensional space.

[0099] For a first geometric figure, the plane curve lies, for example, in a respective first curve plane that forms an angle in the range [0°, 180°] with the image plane of the first depth image. For a second geometric figure, the plane curve lies, for example, in a respective first curve plane that forms an angle in the range [0°, 180°] with the image plane of the second depth image. The value of the angle depends on the path of the edge in three-dimensional space.

[0100] In this way, the complexity of the curve and the number of parameters needed to describe the curve remain comparatively low, which further reduces the storage and computing effort.

[0101] According to at least one embodiment, for each curve of the plurality of first geometric figures and for each curve of the plurality of second geometric figures, a respective second curve plane is constructed, which intersects the respective first curve plane. For each image pixel of the respective depth image, a further pixel value is determined, which is a predetermined first binary value if a position defined by the respective image pixel in three-dimensional space lies on a first side of the respective second curve plane, and a predetermined second binary value if the position defined by the respective image pixel in three-dimensional space lies on a second side of the respective second curve plane opposite the first side. Based on the determined further pixel values, a respective binary image is generated. The respective curve is referred to here as a first curve, and the resulting 2023PF02450

[0102] 18. An approximate region of the at least one object is defined as a first region, or the edge as the first edge. Based on the respective binary image, a second curve is determined in three-dimensional space, which approximates a second region, in particular a second edge, of the at least one object. The respective symbolic representation contains, in particular, the second curve.

[0103] For example, the first binary value is equal to one and the second binary value is equal to zero, or vice versa.

[0104] The second edge can also be interpreted as the contour of the object in the binary image. The second curve is therefore a planar curve in three-dimensional space that lies in the second curve plane. The second curve plane can, for example, be perpendicular to the first curve plane and / or perpendicular to the xy-plane. The second curve plane can also be parallel to the xz-plane.

[0105] The position defined by an image pixel in three-dimensional space is given, in particular, by the position of the image pixel in the image plane of the depth image and the corresponding pixel value of the depth image, i.e., the corresponding depth. For image pixels whose position in three-dimensional space lies exactly on the second curve plane, various predefined procedures can be used. For example, the next pixel value in this case can, by definition, be the first binary value or the second binary value.

[0106] The symbolic representation is generated and stored, for example, in such a way that the first curve and the second curve are assigned to each other. If a further first curve is determined for each additional first edge of the at least one object, then, in various embodiments, a corresponding further second curve can be generated for each additional first curve analogously to the first and second curves.

[0107] By representing at least one object using pairs of curves, the at least one object can be represented more accurately without significantly increasing the required storage and computing effort.

[0108] According to at least one embodiment, the first curve is mirror-symmetric and the second curve plane is constructed as a mirror plane of the first curve. 2023PF02450

[0109] 19

[0110] The second curve plane therefore passes, in particular, through a midpoint of the first curve. If the first curve is a semi-ellipse, the second curve plane passes, in particular, through a vertex of the semi-ellipse.

[0111] This enables a particularly simple construction of the second curve plane and achieves a particularly meaningful representation for at least one object.

[0112] According to at least one embodiment, the first curve and / or the second curve is an elliptical arc, in particular a semi-ellipse.

[0113] This also includes, in particular, the limiting cases of a straight line, which corresponds to a semi-ellipse with a first semi-axis length equal to half the length of the line and a second semi-axis length equal to zero, and of a semicircle, which corresponds to a semi-ellipse with two equal semi-axis lengths. This can be understood, in particular, to mean that the respective curve is generally assumed to be a semi-ellipse, but when fitting it to the corresponding edge, one of the two limiting cases may prove to be optimal.

[0114] By using an elliptical arc or a semi-ellipse, a good approximation can be achieved with comparatively low complexity.

[0115] According to a further aspect of the invention, a method for at least partially automatic vehicle control is described, wherein a computer-implemented method according to the invention is carried out. At least one control signal for at least partially automatic vehicle control is generated depending on the position and / or speed of the feature to be tracked relative to the vehicle, and / or driver assistance information to support the driver of the vehicle in controlling the vehicle is generated depending on the position and / or speed of the feature to be tracked relative to the vehicle. The position corresponds in particular to a position and / or orientation.

[0116] The at least one control signal can be provided, for example, to one or more actuators of the vehicle, including, for example, one or more brake actuators and / or one or more steering actuators and / or one or more drive motors of the vehicle. The one or more actuators 2023PF02450

[0117] 20 can influence longitudinal and / or lateral control of the vehicle based on at least one control signal in order to control the vehicle at least partially automatically.

[0118] The driver assistance information can be output via a vehicle output device, such as a display and / or an audio output system and / or a haptic output system.

[0119] Further embodiments of the inventive method for at least partially automatic vehicle guidance follow directly from the various configurations of the inventive computer-implemented method for object tracking and vice versa. In particular, individual features and corresponding explanations as well as advantages relating to the various embodiments of the inventive computer-implemented method for object tracking can be transferred analogously to corresponding embodiments of the inventive method for at least partially automatic vehicle guidance.

[0120] According to another aspect of the invention, a data processing system is specified which is configured to carry out a computer-implemented method for object tracking according to the invention.

[0121] The terms "data processing system" and "at least one data processing device" may be used interchangeably within the scope of this disclosure. In this disclosure, a data processing device may, for example, be understood as a device with processing circuits for processing data. A data processing device can thus perform arithmetic operations to process data. Indexed access to a data structure, such as a lookup table (LUT) or a database, may also be considered an arithmetic operation. Similarly, data processing that is partially or fully implemented in hardware may be considered an arithmetic operation.

[0122] A data processing device may, in particular, comprise one or more computers, one or more microcontrollers, and / or one or more integrated circuits, for example, one or more application-specific integrated circuits (ASICs), one or more field-programmable gate arrays (FPGAs), and / or one or more systems-on-a-chip (SoCs). A data processing device may also include one or more processors, for example, one or more 2023PF02450 processors.

[0123] 21

[0124] Microprocessors, one or more central processing units (CPUs), one or more graphics processing units (GPUs), and / or one or more signal processors, in particular one or more digital signal processors (DSPs). The data processing equipment may also include a physical or virtual cluster of computers or other equipment of the aforementioned type.

[0125] A data processing device may also include one or more hardware and / or software interfaces, for example for receiving and / or providing data.

[0126] A data processing device may also include one or more storage devices. A storage device may be implemented as volatile memory, such as dynamic random access memory (DRAM) or static random access memory (SRAM), or as non-volatile data storage, such as read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), flash memory or flash EEPROM, ferroelectric random access memory (FRAM), magnetoresistive random access memory (MRAM), or phase-change random access memory (PCRAM).

[0127] According to a further aspect of the invention, an electronic vehicle guidance system for a vehicle is provided. The electronic vehicle guidance system comprises a data processing system according to the invention and a control system configured to generate at least one control signal for at least partially automatic guidance of the vehicle depending on the position and / or speed of the feature to be tracked relative to the vehicle, and / or to generate driver assistance information to support a driver of the vehicle in guiding the vehicle depending on the position and / or speed of the feature to be tracked relative to the vehicle.

[0128] The control system can be part of the data processing system or part of another data processing system.

[0129] An electronic vehicle guidance system can be understood as an electronic system designed to control a vehicle fully automatically or autonomously, in particular without requiring any intervention by a driver. The vehicle performs all necessary functions, such as steering, 2023PF02450

[0130] 22

[0131] The electronic vehicle control system automatically performs braking and / or acceleration maneuvers, monitors and records road traffic, and reacts accordingly. In particular, the system can implement a fully automated or fully autonomous driving mode according to Level 5 of the SAE J3016 classification. An electronic vehicle control system can also be understood as an advanced driver assistance system (ADAS), which supports the driver in partially automated or semi-autonomous driving. Specifically, the electronic vehicle control system can implement a partially automated or semi-autonomous driving mode according to Levels 1 to 4 of the SAE J3016 classification. Here and in the following, "SAE J3016" refers to the corresponding standard in the April 2021 version.

[0132] At least partially automated vehicle control can therefore include driving the vehicle in accordance with a fully automated or fully autonomous driving mode of Level 5 according to SAE J3016. At least partially automated vehicle control can also include driving the vehicle in accordance with a partially automated or semi-autonomous driving mode according to Levels 1 to 4 of SAE J3016.

[0133] According to at least one embodiment of the electronic vehicle guidance system, this includes the environment sensor system, in particular the active optical sensor system.

[0134] According to at least one embodiment, the active optical sensor system is designed as a lidar sensor system, in particular as a laser scanner.

[0135] According to at least one embodiment, the active optical sensor system comprises a two-dimensional detector array with a plurality of detector pixels. The plurality of detector pixels is arranged according to the plurality of columns and the plurality of rows of the detector array. The total number of rows of the detector array is at least 100, for example, 100 to 5,000, and / or the total number of columns of the detector array is at least 100, for example, 100 to 5,000.

[0136] Further embodiments of the electronic vehicle guidance system according to the invention follow directly from the various configurations of the computer-implemented method according to the invention and the method according to the invention for at least partially automatic guidance of a vehicle and 2023PF02450

[0137] 23 Conversely. In particular, individual features and corresponding explanations as well as advantages regarding the various embodiments of the computer-implemented method according to the invention and the method according to the invention for at least partially automatic guidance of a vehicle can be transferred analogously to corresponding embodiments of the electronic vehicle guidance system according to the invention. In particular, the electronic vehicle guidance system according to the invention is configured or programmed to carry out a computer-implemented method or method according to the invention. In particular, the electronic vehicle guidance system according to the invention carries out the computer-implemented method or method according to the invention.

[0138] According to another aspect of the invention, a computer program with instructions is specified. When the instructions are executed by a data processing system, the instructions cause the data processing system to carry out a computer-implemented method according to the invention.

[0139] The instructions can be provided, for example, as program code. This program code can be provided, for example, as binary code or assembly language, and / or as source code in a programming language such as C, and / or as a program script, such as Python.

[0140] According to a further aspect of the invention, a further computer program with additional commands is specified. When the additional commands are executed by an electronic vehicle guidance system according to the invention, in particular by the data processing system of the electronic vehicle guidance system, the commands cause the electronic vehicle guidance system to carry out a method according to the invention for at least partially automatic guidance of a vehicle.

[0141] The additional instructions can be provided, for example, as program code. This program code can be provided, for example, as binary code or assembly language, and / or as source code in a programming language such as C, and / or as a program script, such as Python.

[0142] According to a further aspect of the invention, a computer-readable storage medium is provided that stores a computer program according to the invention and / or a further computer program according to the invention. 2023PF02450

[0143] 24

[0144] The computer program, the further computer program, and the computer-readable storage medium are each computer program products containing the commands and / or the further commands.

[0145] Further features of the invention are evident from the claims, the figures, and the description of the figures. The features and combinations of features mentioned above in the description, as well as those mentioned below in the description of the figures and / or illustrated in the figures, may be encompassed by the invention not only in the combinations specified, but also in other combinations. In particular, embodiments and combinations of features that do not include all the features of an originally formulated claim may also be encompassed by the invention. Furthermore, embodiments and combinations of features that go beyond or deviate from the combinations of features mentioned in the claims may also include the invention.

[0146] The invention is explained in more detail below with reference to specific exemplary embodiments and corresponding schematic drawings. Identical or functionally equivalent elements in the drawings may be provided with the same reference numerals. The description of identical or functionally equivalent elements is not necessarily repeated with respect to the different figures.

[0147] The figures show,

[0148] Fig. 1 shows a schematic representation of a vehicle with an exemplary embodiment of an electronic vehicle guidance system according to the invention;

[0149] Fig. 2 shows a schematic representation of an active optical sensor system of another exemplary embodiment of an electronic vehicle guidance system according to the invention;

[0150] Fig. 3 schematically shows a depth image corresponding to an exemplary embodiment of a computer-implemented method for object tracking according to the invention; 2023PF02450

[0151] 25

[0152] Fig. 4 schematically shows a depth image and curves according to a further exemplary embodiment of a computer-implemented method for object tracking according to the invention;

[0153] Fig. 5 schematically shows a depth image and curves corresponding to a further exemplary embodiment of a computer-implemented method for object tracking according to the invention;

[0154] Fig. 6 schematic curve shapes for use in a further exemplary embodiment of a computer-implemented method for object tracking according to the invention;

[0155] Fig. 7 schematically shows objects according to a further exemplary embodiment of a computer-implemented method for object tracking according to the invention;

[0156] Fig. 8 schematically shows further objects according to another exemplary embodiment of a computer-implemented method for object tracking according to the invention; and

[0157] Fig. 9 schematically shows first geometric figures corresponding to a first detection period and second geometric figures corresponding to a second detection period according to a further exemplary embodiment of a computer-implemented method for object tracking according to the invention.

[0158] Figure 1 schematically depicts a motor vehicle 1 with an exemplary embodiment of an electronic vehicle guidance system 2 according to the invention. The electronic vehicle guidance system 2 comprises a data processing system 4 and an active optical sensor system 3.

[0159] In particular, a computer-implemented method for object tracking according to the invention can be carried out using the data processing system 4.

[0160] In this process, the active optical sensor system 3 generates initial sensor data corresponding to a first acquisition period and a second 2023PF02450.

[0161] 26

[0162] During the acquisition period, corresponding second sensor data are received, each representing an environment of vehicle 1. Based on the first sensor data, a plurality of first geometric figures 16a, 16b, 16c, 16d, 25 are determined, each approximating an area of ​​at least one object 5, 14, 20, 22, 24 represented by the first sensor data. Based on the second sensor data, a plurality of second geometric figures are determined, each approximating an area of ​​at least one object represented by the second sensor data. One of the plurality of second geometric figures is selected as the feature to be tracked. Depending on a comparison of the plurality of second geometric figures with the plurality of first geometric figures 16a, 16b, 16c, 16d, 25, the feature to be tracked is assigned to a reference figure of the first geometric figures 16a, 16b, 16c, 16d, 25 that corresponds to the feature to be tracked.

[0163] An exemplary embodiment of the active optical sensor system 3 is shown schematically in Fig. 2. The active optical sensor system 3 comprises a housing 7 and an emitter unit 9 for emitting light 6a into an external environment of the active optical sensor system 3. The emitted light 6a passes, for example, through a window 8 of the housing 7. The active optical sensor system 3 has a control unit 11, which is configured to control a deflection device 18, for example, a rotatable mirror, of the active optical sensor system 3, to deflect the light 6a in different directions, and thereby to scan the environment of the active optical sensor system 3.

[0164] The active optical sensor system 3 has a detector unit with a two-dimensional detector array 10, which comprises a multitude of detector pixels arranged, for example, in a multitude of rows and a multitude of columns. If components 6b of the emitted light 6a are reflected by an object 5 in the vicinity, these components 6b can be reflected back towards the active optical sensor system 3 and, for example, re-enter through the window 8, where they are directed by the deflection device 18 onto the detector array 10 and detected by it. Depending on the corresponding detector signals, the control unit 11 can then calculate a radial distance of the object 5 from the active optical sensor system 3, for example, by measuring the time of flight.

[0165] The active optical sensor system 3 is specifically designed as a lidar sensor system in the form of a laser scanner. The emitter unit 9 therefore comprises one or more laser light sources, in particular laser diodes, for generating the light 6a. The 2023PF02450

[0166] 27

[0167] Each detector pixel has one or more photodetectors, which can be configured, for example, as photodiodes, avalanche photodiodes (APDs), or single-photon avalanche photodiodes (SPADs). Possible optical components in the optical path are represented in Fig. 2 by a lens 12.

[0168] In some embodiments, the first sensor data includes a first depth image 13 of the environment generated by the detector array 10, and the second sensor data includes a second depth image of the environment generated by the detector array 10. The first depth image 13 and the second depth image each have a plurality of image pixels, with each image pixel corresponding to one of the detector pixels and a pixel value of the respective image pixel corresponding to a depth determined by the corresponding detector pixel in a direction perpendicular to the detector array 10. In Fig. 3, the first depth image 13 is schematically represented as a binary image for simplicity, with the hatched area corresponding to a background 15, particularly at infinity, and the unhatched area corresponding to an object 14. In the general case, the first depth image 13 and the second depth image would be monochromatic images, for example, grayscale images.Each image plane of the first depth image 13 and the second depth image is defined by a transverse direction y and a vertical direction z; a longitudinal direction x is oriented perpendicular to the transverse direction y and the vertical direction z. In the simplified example of Fig. 3, the object 14 has a flat rectangular surface on the side facing the active optical sensor system 3.

[0169] Based on the first depth image 13, a first curve 16a, 16b, 16c, 16d, 25 is determined in three-dimensional space, which approximates a first edge of the object 5, 14, 20, 22, 24 as represented by the first depth image 13. This is shown schematically in Fig. 4 for the first depth image 13 from Fig. 3 for four first curves 16a, 16b, 16c, 16d. In this simplified example, the first curves 16a, 16b, 16c, 16d lie in the y-z plane, which is not necessarily the case in general. In general, the first curve 16a, 16b, 16c, 16d, 25 lies, for example, in a first curve plane that forms an angle in the range [0, 180°] with an image plane of the first depth image 13. Similarly, based on the second depth image, a first curve 16a', 16b', 16c', 16d' is determined in three-dimensional space, which approximates a first edge of the object as represented by the second depth image. This is also shown schematically in Fig. 9. 2023PF02450

[0170] 28

[0171] As shown schematically in Fig. 7, it is also possible to distinguish between edges 21, 23 of different objects 20, 22 based on the corresponding depth image 13, although it may happen that these cannot be distinguished in the image plane of the corresponding depth image 13 if the objects 20, 22 partially overlap or are directly adjacent to each other.

[0172] The first curves 16a, 16b, 16c, 16d, 25 are defined, for example, as respective semi-ellipses 19, which also includes the limiting case of straight lines, as schematically shown in Fig. 6 and in the simplified example of Fig. 4. The limiting case of a semicircle is also included, as shown in the example of Fig. 8 for a spherical object 24 and the corresponding first curve 25.

[0173] In preferred embodiments, for each first curve 16a, 16b, 16c, 16d, 25, a corresponding second curve 17, 26 is generated, as shown in Fig. 8 for the first curve 25 and in the simplified example of Fig. 5 for the first curve 16a. The second curve 17, 26 can also be defined as a semi-ellipse.

[0174] To determine the second curve 17, 26, a second curve plane is constructed that intersects the first curve plane. For example, the second curve plane is perpendicular to the first curve plane. For example, the second curve plane is a mirror plane with respect to a mirror symmetry of the first curve 16a, 16b, 16c, 16d, 25.

[0175] The second curve plane divides three-dimensional space into two halves, defined by a region on the first side of the second curve plane and a region on the opposite side of the second curve plane. For each image pixel of the depth image 13, for example, an additional pixel value is determined. This additional pixel value is a predetermined first binary value if a position defined by the respective image pixel in three-dimensional space lies on the first side of the second curve plane, and a predetermined second binary value if the position defined by the respective image pixel in three-dimensional space lies on the second side of the second curve plane. Based on these determined additional pixel values, a binary image is generated. Based on this binary image, the second curve 17, 26 is determined, which approximates a second edge of the object 5, 14, 20, 22, 24.

[0176] In some embodiments, the present invention provides a universal solution for object tracking without requiring a motion model, object classification, or semantic segmentation. 2023PF02450

[0177] 29

[0178] In some embodiments, the vehicle's own motion can also be taken into account when assigning the reference figure to the object being tracked, in addition to the result of comparing the multitude of second geometric figures with the multitude of first geometric figures. This allows for further improved reliability, particularly when tracking objects that are geometrically difficult to distinguish from one another, especially static objects such as posts or guardrails.

Claims

2023PF02450 30 Patent claims 1. A computer-implemented method for object tracking, wherein first sensor data corresponding to a first detection period and second sensor data corresponding to a second detection period are obtained by means of an environment sensor system (3) of a vehicle (1), each representing an environment of the vehicle (1); based on the first sensor data, a plurality of first geometric figures (16a, 16b, 16c, 16d, 25) are determined, each approximating an area of ​​at least one object (5, 14, 20, 22, 24) represented by the first sensor data; based on the second sensor data, a plurality of second geometric figures (16a, 16b, 16c, 16d, 25) are determined, each approximating an area of ​​at least one object (5, 14, 20, 22, 24) represented by the second sensor data; one of the many second geometric figures (16a, 16b, 16c, 16d, 25) is selected as the feature to be tracked;The feature to be tracked is assigned to a reference figure of the first geometric figures (16a, 16b, 16c, 16d, 25) which corresponds to the feature to be tracked, depending on a comparison of the multitude of second geometric figures (16a, 16b, 16c, 16d, 25) with the multitude of first geometric figures (16a, 16b, 16c, 16d, 25).

2. Computer-implemented method according to claim 1, wherein, to assign the feature to be tracked to the reference figure, the reference figure is identified by comparing the feature to be tracked with the plurality of first geometric figures (16a, 16b, 16c, 16d, 25); and for a further second geometric figure (16a, 16b, 16c, 16d, 25) of the plurality of second geometric figures (16a, 16b, 16c, 16d, 25), a further first geometric figure (16a, 16b, 16c, 16d, 25) of the plurality of first geometric figures (16a, 16b, 16c, 16d, 25) is identified, the shape and relative position of which 2023PF02450 31 with respect to the reference figure corresponds to the shape and relative position of the further second geometric figure (16a, 16b, 16c, 16d, 25) with respect to the feature to be tracked.

3. Computer-implemented method according to one of the preceding claims, wherein the plurality of first geometric figures (16a, 16b, 16c, 16d, 25) and the plurality of second geometric figures (16a, 16b, 16c, 16d, 25) each comprise a corresponding curve in three-dimensional space.

4. Computer-implemented method according to claim 2 and claim 3, wherein the plurality of second geometric figures (16a, 16b, 16c, 16d, 25) is transformed such that a transformed starting point of the curve of the feature to be tracked and / or a transformed endpoint of the curve of the feature to be tracked lies at predefined positions of a common coordinate system; after identifying the reference figure, the plurality of first geometric figures (16a, 16b, 16c, 16d, 25) is transformed such that a transformed starting point of the curve of the reference figure and / or a transformed endpoint of the curve of the reference figure lies at the predefined positions of the common coordinate system;and the further first geometric figure (16a, 16b, 16c, 16d, 25) of the plurality of first geometric figures (16a, 16b, 16c, 16d, 25) is identified based on the transformed plurality of second geometric figures (16a, 16b, 16c, 16d, 25) and the transformed plurality of first geometric figures (16a, 16b, 16c, 16d, 25).

5. Computer-implemented method according to claim 3 or 4, wherein, to determine a first geometric figure (16a, 16b, 16c, 16d, 25) of the plurality of first geometric figures (16a, 16b, 16c, 16d, 25), the corresponding area of ​​the at least one object (5, 14, 20, 22, 24) represented by the first sensor data is approximated by a polyline and the polyline is approximated by the corresponding curve; or the corresponding area of ​​the at least one object (5, 14, 20, 22, 24) represented by the first sensor data is approximated by two or more line segments 2023PF02450 32 is approached and the two or more line segments are approximated by the corresponding curve.

6. Computer-implemented method according to one of the preceding claims, wherein the first sensor data and the second sensor data are each sensor data generated by means of an active optical sensor system (3) of the vehicle (1).

7. Computer-implemented method according to claim 6, wherein the active optical sensor system (3) comprises a two-dimensional detector array (10) with a plurality of detector pixels, and wherein the first sensor data includes a first depth image (13) of the environment generated by the detector array (10) or the first depth image (13) is generated depending on the first sensor data; the second sensor data includes a second depth image (13) of the environment generated by the detector array (10) or the second depth image (13) is generated depending on the second sensor data; the first depth image (13) and the second depth image (13) each comprise a plurality of image pixels, wherein each image pixel corresponds to one of the detector pixels and a pixel value of the respective image pixel corresponds to a distance from the detector array (10) determined by the corresponding detector pixel;and the plurality of first geometric figures (16a, 16b, 16c, 16d, 25) is determined based on the first depth image (13) and the plurality of second geometric figures (16a, 16b, 16c, 16d, 25) is determined based on the second depth image (13).

8. Computer-implemented method according to claim 7 and one of claims 3 to 5, wherein the respective curve is a planar curve.

9. Computer-implemented method of claim 8, wherein the respective curve is an elliptical arc or a semi-ellipse (19).

10. Method for at least partially automatic control of a vehicle (1) , wherein a computer-implemented method according to any of the preceding claims is carried out and 2023PF02450 33 at least one control signal for at least partially automatic control of the vehicle (1) is generated depending on the position and / or speed of the feature to be tracked relative to the vehicle (1); and / or driver assistance information to assist a driver of the vehicle (1) in controlling the vehicle (1) is generated depending on the position and / or speed of the feature to be tracked relative to the vehicle (1).

11. Data processing system (4) configured to carry out a computer-implemented method according to any one of claims 1 to 9.

12. Electronic vehicle guidance system (2) for a vehicle (1), comprising a data processing system (4) according to claim 11 and a control system configured to generate at least one control signal for at least partially automatic guidance of the vehicle (1) depending on the position and / or speed of the feature to be tracked relative to the vehicle (1); and / or to generate driver assistance information to support a driver of the vehicle (1) depending on the position and / or speed of the feature to be tracked relative to the vehicle (1).

13. Electronic vehicle guidance system (2) according to claim 12, wherein the electronic vehicle guidance system (2) comprises an environment sensor system (3) and the environment sensor system (3) is designed as an active optical sensor system.

14. Electronic vehicle guidance system (2) according to claim 13, wherein the active optical sensor system (3) is designed as a lidar sensor system.

15. Electronic vehicle guidance system (2) according to one of claims 13 or 14, wherein the active optical sensor system (3) comprises a two-dimensional detector array (10) with a plurality of detector pixels; the plurality of detector pixels is arranged according to a plurality of columns and rows of the detector array (10); and 2023PF02450 34. The total number of rows of the detector array (10) is at least 100 and / or the total number of columns of the detector array (10) is at least 100.

16. Comprising a computer program product Commands which, when executed by a data processing system (4), cause the data processing system (4) to perform a computer-implemented method according to any one of claims 1 to 9; and / or further commands which, when executed by an electronic vehicle guidance system (2) according to any one of claims 12 to 15, cause the electronic vehicle guidance system (2) to perform a method according to claim 10.

Images