Methods, apparatus, and storage media for determining the overlap between an object and a vehicle belt.

Abstract

This invention relates to a method for determining the overlap between at least one object and at least one lane by presenting the surrounding environment of a platform as a two-dimensional pixel data domain, comprising the steps of: assigning at least one lane-pixel group to corresponding pixels representing at least one lane in the two-dimensional pixel data domain; assigning at least one object-pixel group to corresponding pixels representing at least one object in the two-dimensional pixel data domain; defining at least one object-pixel pair in the two-dimensional pixel data domain S3, the object-pixel pair representing the width of at least one object-pixel group; and comparing the object-pixel pair with the lane-pixel group in the two-dimensional pixel data domain S4. The invention also relates to methods for identifying target objects, methods for assigning objects to lanes, methods for sending control signals for manipulating vehicles and / or warning signals for warning vehicle occupants, and apparatus, computer programs, and storage media for implementing the method of the invention.

Description

Technical Field

[0001] The present invention relates to a method for determining the degree of overlap between at least one object and at least one carriageway by presenting the surrounding environment of a platform as a two-dimensional pixel data domain. Background Technology

[0002] Driver assistance systems typically use radar sensors to adjust distance. These radar sensors are capable of accurately determining the distance between the radar sensor and an object, such as a motor vehicle. For distance measurement with measurement-based distance adjustment, it is necessary to select a target object relevant to the distance adjustment from multiple objects that can be sensed by the radar sensor. Since the radar sensor directly determines the distance, the target vehicle is selected in three-dimensional space according to existing technology.

[0003] Here, in a single-sensor strategy for spacing adjustment using radar data, the selection of a target vehicle is derived solely from time tracking of the object without knowing the direction of the driving lane.

[0004] In a dual-sensor system for spacing adjustment (which has a radar sensor and a video sensor with video-based vehicle strip marking recognition), vehicle strip information is transformed into the three-dimensional measurement space of the radar sensor. Summary of the Invention

[0005] The transformation from two-dimensional image space to three-dimensional space is ambiguous and noisy, causing video-based single-sensor strategies that transform measurements of lane markings from image sensors, such as those of vehicle lane markings, into three-dimensional space to suffer from the following drawbacks: The transformation from lane markings identified in two-dimensional space, i.e., in a two-dimensional (2D) representation of the vehicle's surroundings, to three-dimensional (3D) space can introduce significant errors or violate model assumptions. This results in lane markings being incorrectly identified in the three-dimensional representation of the surroundings. However, the subsequent selection of target vehicles can only compensate for the bias and strong noise to a limited extent. This leads to incorrect objects or no objects being selected as target objects.

[0006] The impact of these problems is particularly pronounced over greater distances. This is partly because marker recognition has a limited effective range compared to vehicle or object detection. Furthermore, in the case of video sensors, the noise from the transition from a two-dimensional to a three-dimensional representation of the surrounding environment increases exponentially with distance. Thus, a stationary vehicle may be detected very early at the end of a congestion, even if, under certain circumstances, it is not assigned to the traffic lane. Consequently, the necessary intervention for guiding at least partially automated vehicles cannot be comfortably implemented.

[0007] This invention discloses a method for determining the overlap between an object and a conveyor belt, a method for identifying a target object, a method for assigning at least one object to a conveyor belt, an apparatus, a computer program, and a machine-readable storage medium. Advantageous configurations are those described in the preferred embodiments and the following specification.

[0008] This invention is based on the unexpected realization that the size of elements in the representation of the platform's surrounding environment, such as the size of a vehicle, can be analyzed, evaluated, and compared more accurately in a two-dimensional representation than in a three-dimensional representation.

[0009] According to one embodiment, the method determines the overlap between at least one object and at least one carriageway by representing the platform's surrounding environment as a two-dimensional pixel data domain. In one step, at least one carriageway-pixel group is assigned to pixels in the two-dimensional pixel data domain that correspondingly represent at least one carriageway. In another step, at least one object-pixel group is assigned to pixels in the two-dimensional pixel data domain that correspondingly represent at least one object. In yet another step, at least one object-pixel pair is defined in the two-dimensional pixel data domain, representing the width of the at least one object-pixel group. In yet another step, the object-pixel pair and the carriageway-pixel group are compared in the two-dimensional pixel data domain.

[0010] Throughout this description, the order of the method steps is shown to facilitate understanding of the method or to correspond to a preferred embodiment. However, those skilled in the art will recognize that many of these method steps can be performed in a different order and result in the same outcome. In this sense, the order of the method steps can be changed accordingly.

[0011] For at least partially automated vehicles, and for driver assistance systems, it is crucial to use sensor systems to assign different types of identified objects, such as vehicles, to the driving lanes within the surrounding traffic environment. This assignment can be quantified using overlap, as it is important not only that a vehicle is traveling entirely within the driving lane, but also, and more importantly, that lane changes increase or decrease the overlap with the various driving lanes during these changes, which are crucial for safe driving. Furthermore, the identification of another vehicle partially traveling within the driving lane is important, for example, for decisions regarding the freedom of movement within the driving lane.

[0012] For example, the degree of overlap between an object and the driving lane indicates how similar the dimensions of the object's extension are to the dimensions of the driving lane's extension. As an example of an object, if the vehicle is traveling entirely within the driving lane, there is complete overlap, meaning the degree of overlap is 100%. When an object, such as a vehicle, changes lanes, this degree of overlap can range from 0% to 100%, thus indicating the extent to which the lane change is completed. Similarly, the degree of overlap can also be reduced accordingly.

[0013] For example, the representation of the platform's surrounding environment for further analysis and evaluation, such as that of at least partially automated vehicles, robots, or vehicles with driver assistance systems, can be achieved using a two-dimensional representation of the three-dimensional surrounding environment. In particular, this two-dimensional representation can be achieved using a two-dimensional data domain, in which a single location characterized in two dimensions carries information describing and / or representing the three-dimensional surrounding environment after its transformation into two dimensions. Specifically, this information may relate to the optical properties of the three-dimensional surrounding environment, but these single locations defined in two dimensions may also contain other information. For example, this information may relate to the output values ​​of an optical imaging system, such as a digital camera or digital video system, such that this information has brightness and color information. In this case, the optical imaging of the surrounding environment is, for example, a representation of the surrounding environment that can be described using a two-dimensional pixel data domain. Furthermore, information about the three-dimensional surrounding environment based on different sensor types can be used alternatively or additionally for this two-dimensional representation of the three-dimensional surrounding environment. In particular, any camera system can be used, such as a front-facing single camera, a stereo camera, a fisheye camera, or any multi-camera system. Another example of an imaging sensor system is a lidar sensor.

[0014] To make the method easier to understand, we will refer to the two-dimensional pixel data domain in the context of this paper. In this sense, it should be understood as pixel data of a two-dimensional field, and not limited to optical pixel data.

[0015] As objects, not only static objects but also dynamic objects can be identified and assigned. In the context of the surrounding environment, such as traffic conditions, these objects can especially be motor vehicles, such as passenger cars or trucks.

[0016] Such a platform can be, in particular, a mobile platform, and examples of such a mobile platform include at least partially automated vehicles, drones, robotic systems, or service robots. However, such a mobile platform can also be a vehicle equipped with driver assistance systems.

[0017] A driving strip describes an area of ​​a road where vehicles can travel in one direction. A lane is a road strip marked on a road for travel in a defined direction. Thus, the concept of "driving strip" includes the concept of "lane" in this invention. A driving strip typically has only elements of a defined lane in the form of architectural elements, such as curbs, fences, or walls, or has different surface properties, such as a transition from asphalt to grass.

[0018] Lane markings, or lane markings, can be identified using various methods, such as within an optical image, through mathematical convolution operations that enhance brightness contrast or other suitable techniques. This identification can be achieved using a two-dimensional pixel data domain, or more particularly, through other methods that analyze and evaluate information from the three-dimensional surrounding environment. The identified lane markings are assigned to a lane-pixel group in the two-dimensional pixel data domain, which therefore represents the lane marking.

[0019] Object identification in the platform's surrounding environment can be achieved using various methods, not only with a two-dimensional pixel data domain but also, for example, with a three-dimensional representation of the surrounding environment. Identification can be performed using methods based on rule-based logic or neural network techniques. These identified objects are assigned to an object-pixel group in the two-dimensional pixel data domain, and the object-pixel group thus represents the corresponding object.

[0020] To characterize the width of at least one object-pixel group, object-pixel pairs are defined in a two-dimensional pixel data domain. Depending on the nature of the identified object, the object-pixel pair can be defined in different ways. For example, a rectangular frame enclosing the object-pixel pair can be defined around the object-pixel group in the two-dimensional pixel data domain. The object-pixel pair can be selected on this rectangle such that it defines the width of the object-pixel group. As another example, the object-pixel pair can also be defined by pixels in the two-dimensional pixel data domain that are furthest apart from each other at the lower edge of the object-pixel group.

[0021] By comparing object-pixel pairs with carriageway-pixel groups in a two-dimensional pixel data domain, the ratio of the object's width to the extended scale, or width, of the carriageway-pixel group, which represents the carriageway in the two-dimensional pixel data domain, can be determined.

[0022] This simple method achieves high accuracy because it does not require transforming the surrounding environment from a two-dimensional to a three-dimensional representation. Transforming the surrounding environment from a two-dimensional to a three-dimensional representation is ambiguous and therefore prone to error. In this method, information about the object and information about the conveyor belt are fused in the two-dimensional pixel data domain and can therefore be correlated or compared in a simple step. This method can be used in parallel with other methods also used to determine the overlap between an object and the conveyor belt. In some existing systems, this method can also be implemented by modifying the software if necessary.

[0023] According to one configuration of this method, at least one object-pixel pair is compared with a traffic strip-pixel group within exactly one pixel row of the pixel data field.

[0024] This allows for the minimization of offset when determining the degree of overlap with respect to the surrounding environment. Because perspective distortion always occurs when transforming a natural three-dimensional representation of the surrounding environment into a two-dimensional one, this distortion can introduce errors when determining the degree of overlap, especially when offset is present.

[0025] According to one configuration of this method, a first set of pixels containing an object-pixel pair and a second set of pixels containing a carriageway-pixel group are compared between object-pixel pairs in a two-dimensional pixel data domain, the second set of pixels being located in the same pixel row as the object-pixel pair in the pixel data domain.

[0026] Therefore, the width of the object can be easily described and calculated using the first set of pixels, and the width of the vehicle belt can be easily described and calculated using the second set of pixels in the two-dimensional pixel data domain.

[0027] According to one configuration of this method, in another step, the pixel intersection of a first pixel set and a second pixel set is determined. In another step, the overlap between at least one object and at least one carriageway is determined by calculating the ratio of the number of pixels in the determined intersection to the number of pixels in the first set.

[0028] Therefore, the degree of overlap between the width of the object and the conveyor belt can be quantitatively determined.

[0029] According to one configuration of this method, at least one represented object belongs to a category, and the degree of overlap is determined in relation to the classification results.

[0030] This classification system can help determine whether overlap should be determined, or whether the object width should be determined in relation to the classification when determining overlap. For example, it can help determine the width of trucks and passenger cars in different ways.

[0031] According to one configuration of this method, a pixel box surrounding at least one object-pixel group is assigned in a two-dimensional pixel data domain. Object-pixel pairs representing the width of at least one object-pixel group are arranged on the pixel box such that the object-pixel pair represents the width of the object-pixel group.

[0032] Furthermore, the pixel box can also be adapted to the shape of an object, or rather, the shape of an object-pixel group. This corresponds to the outer contour of the object, and the width of the object-pixel group can be characterized by object-pixel pairs that are furthest apart from each other in a row on the pixel box adapted to the shape. Additionally, contact points between the object and the road surface, such as the contact points between a tire and the road surface, can be identified, and these points can be used to characterize the width of the object-pixel group.

[0033] According to another configuration of this method, at least one object represented by an object-pixel group is identified by rule-based logic or by a neural network. The identification can be performed using a two-dimensional pixel data domain or by a method operating in three-dimensional space, wherein the identified object is then assigned to pixels of the object-pixel group in the two-dimensional pixel data domain in another step. By employing these different methods for identifying objects in the platform's surrounding environment, the most suitable method for object identification can be selected for each surrounding environment and each object.

[0034] According to another configuration of this method, at least one lane-pixel group is assigned to pixels in the two-dimensional pixel data field by means of at least one marker-pixel group representing the lane boundary of the two-dimensional pixel data field.

[0035] If only a carriageway boundary is identified by representing the surrounding environment in a two-dimensional pixel data domain or otherwise, the carriageway boundary can be assigned to a pixel in the two-dimensional pixel data domain and, for example, by using an assumption about the carriageway width, the carriageway boundary can be assigned to a carriageway-pixel group in the two-dimensional pixel data domain.

[0036] According to another configuration of this method, at least one marker-pixel group is extended by extrapolation in the two-dimensional pixel data domain.

[0037] Since object identification in practice may be performed at a greater distance from the platform than the identification of lane boundaries, such as lane markings, this configuration of the method can reliably determine the overlap between objects and lanes at a greater distance from the platform, and thus determine this overlap for a larger number of identified objects when necessary. This extrapolation method can obviously be implemented particularly simply in a two-dimensional representation in a two-dimensional pixel data domain.

[0038] According to another configuration of this method, at least one driving strip-pixel group or lane mark in the two-dimensional pixel data field is assigned by the extrapolated mark-pixel group and / or the mark is increased.

[0039] According to another configuration of this method, extrapolation is proposed for the driving strip-pixel group or the mark-pixel group using linear or polynomial functions.

[0040] According to one configuration of this method, the left and / or right boundaries of the driving lane are extrapolated in the image space of the two-dimensional pixel data domain. Here, the driving lane is the driving lane on which the platform travels, or the lane on which the platform itself is located.

[0041] Since this travel belt is particularly important to the platform, such as at least partially automated vehicles or driver assistance systems, safer or more comfortable driving behavior can be achieved by determining the overlap between the identified object and this travel belt at a greater distance from the platform. This greater distance is achieved by extending the travel belt using an extrapolation method.

[0042] According to one configuration of this method, a virtual driving strip boundary is used to assign a driving strip-pixel group to a two-dimensional pixel data domain. This virtual driving strip boundary is generated in parallel with the marker-pixel group and / or an extended marker-pixel group.

[0043] Furthermore, the consistency of the left and right boundaries can be determined by checking whether right and left lane markings are detected in the image. If only one lane marking is detected, the other lane is associated based on assumptions, such as those about the width of the driving lane, since a driving lane always has left and right boundaries. Additionally, the reliability of the derived driving lane markings must be verified. For example, lane markings composed of horizontal and vertical lines are contradictory.

[0044] According to one configuration of this method, at least one lane-pixel group is assigned in the two-dimensional pixel data domain using data from a sensor system.

[0045] Data or measurements from additional sensor systems can be fused in a two-dimensional pixel data domain, that is, integrated into the two-dimensional pixel data domain corresponding to the surrounding environment, in order to achieve, for example, higher accuracy in the position of objects or conveyor belts and thus improve the methods used to determine the degree of overlap.

[0046] Sensor fusion systems typically consist of different sensors, such as radar and video systems. These systems have different measurement principles and different measurement spaces. Therefore, radar sensors directly measure metric values, such as signal propagation time and quantization intervals. To fuse these measurements, a common reference frame is needed. To represent measurements from different reference frames, these measurements must be transformed. Transforming three-dimensional measurement parameters yields three measurements to determine two-dimensional parameters, resulting in mathematical overdefinition. This mathematical overdefinition does not occur when fusing two-dimensional parameters into three-dimensional space. Additional data is required here. In the case of camera images, this data is obtained using a second image of the same point along with information about the distance the platform has traveled between these images. Due to the errors introduced by the additional required data, the transformed result is less accurate.

[0047] According to another configuration of this method, at least one lane-pixel group is identified in a two-dimensional pixel data domain using map data of the platform's surrounding environment.

[0048] Therefore, just as other sensor data can be integrated into the two-dimensional pixel data domain, map data can also be integrated into the two-dimensional pixel data domain. This can improve the accuracy of the driving lane position, especially when the map data contains information about the driving lane and / or lane markings.

[0049] As a result, the method for determining the degree of overlap becomes more accurate and is thus improved, because the driving lanes corresponding to the map data are known at any distance along the direction of travel.

[0050] For this purpose, map data can be integrated with two-dimensional pixel data domains by matching the map data to the surrounding environment represented by the two-dimensional pixel data domains.

[0051] A method for identifying a target object is described, wherein in one step, pixels of a two-dimensional pixel data domain are assigned to at least two lane-pixel groups, which represent at least two lanes of the surrounding environment in the two-dimensional pixel data domain. In another step, a lane-pixel group is selected from the at least two lane-pixel groups. And in yet another step, the overlap between multiple assigned object-pixel groups in the two-dimensional pixel data domain and the lane-pixel group is determined according to the method described above.

[0052] For safe and comfortable driving on at least partially automated platforms, or for driver assistance systems that adjust the distance to vehicles ahead, target objects are identified; that is, objects, such as vehicles traveling in the same lane as the vehicle. To this end, the degree of overlap between multiple identified and associated objects and the lane is determined. Thus, it is determined that a target object exists in the lane.

[0053] To detect which lane the vehicle is traveling on, the lane boundary is tracked over a certain period of time using imaging methods. The lane can be inferred by knowing the camera's mounting position relative to the platform or vehicle and other vehicle detections in the image. Here, the object closest to the vehicle is located closest to the lower edge in the image or corresponding pixel data domain of the imaging system.

[0054] In other words, in an image stored in a two-dimensional pixel data domain, the pixel spacing of observed objects along the horizontal (u) and vertical (v) directions can be used to infer the relative arrangement of objects to each other.

[0055] For example, the distance to an object can be assumed in image space using the v-coordinate. Therefore, an object at the lower edge of the image is closer to the vehicle than an object at the upper edge. This is because the object closest to the vehicle is ultimately particularly important for adjustment. The distance can be calculated in the image using a simple "pixel count".

[0056] Therefore, for the selection of target objects, a two-dimensional pixel data domain is used to represent the platform's surrounding environment. This two-dimensional pixel data domain contains information about the two-dimensional representation of the three-dimensional surrounding environment, such as information derived from an imaging system representing the platform's surrounding environment. This two-dimensional pixel data domain is assigned pixel groups representing objects, which can be identified not only within the two-dimensional pixel data domain but also as already identified objects. Furthermore, the two-dimensional pixel data domain is assigned pixel groups representing the vehicle belt, which can be identified not only within the two-dimensional pixel data domain but also as already identified vehicle belts. Additionally, this vehicle belt is identified. If the pixel group assigned to the vehicle belt in the two-dimensional pixel data domain does not include the object's pixel group, for example because the identification of the vehicle belt did not reach the object, the above method can be used to extrapolate the pixel group assigned to the vehicle belt. This information is used to determine the degree of overlap between this vehicle belt and the identified and assigned object. Then select the target object as an object that has a certain degree of overlap with the current driving lane and is highly relevant to it, such as an object with the minimum distance to the current vehicle.

[0057] A method for assigning at least one object to a traffic belt is described. In one step, pixels of a two-dimensional pixel data field are assigned to at least two traffic belt-pixel groups, which respectively represent at least two traffic belts of the surrounding environment in the two-dimensional pixel data field. In another step, at least one object-pixel group is assigned to pixels of the two-dimensional pixel data field that respectively represent at least one object. And in yet another step, the degree of overlap between at least one object and each of the at least two traffic belt-pixel groups is determined according to the method described above.

[0058] For example, this method can be used to associate the platform’s surrounding environment with the objects that are surrounded by that environment by determining the degree of overlap between each of the multiple identified and associated objects and the multiple traffic belts.

[0059] In addition, it is possible to check all objects identified and associated in the two-dimensional pixel data domain: whether these objects are partially or completely located within the current carriageway.

[0060] Furthermore, the most relevant vehicle can be selected from multiple objects on the current lane. Relevance can be determined by one or all of the following factors: distance to the vehicle, time-to-contact (TTC), number of adjacent lanes, directional arrows on the lane, hazard light detection, brake light detection, and / or information from schilderbrücke (traffic signs). Additionally, the minimum overlap between the object boundary, such as the bounding box, and the current lane can be used as a criterion for selection. Furthermore, predictions, i.e., estimates, of the further movement of the bounding box in the two-dimensional pixel data domain can also be considered.

[0061] A method is described in which control signals for operating at least a partially automated vehicle and / or warning signals for warning vehicle occupants are transmitted in relation to the degree of overlap between the driving belt and the object as determined by the above method.

[0062] Since the degree of overlap between the identified and assigned object and the driving lane indicates the extent to which the driving lane is used by the object, it is possible to react to the object, especially the vehicle, merging into or leaving the driving lane.

[0063] An apparatus is described, configured to implement the above-described method. This apparatus allows the method to be easily integrated into different systems.

[0064] A computer program is described, comprising instructions that, when executed by a computer, instruct the computer to perform one of the methods described above. This computer program enables the use of the methods in various systems.

[0065] A machine-readable storage medium is described, on which the aforementioned computer program is stored. Attached Figure Description

[0066] Reference Figures 1 to 3 Embodiments of the invention are shown and illustrated in detail below. The accompanying drawings show:

[0067] Figure 1 The surrounding environment of the platform with the driving belt;

[0068] Figure 2 A top-down view showing the surrounding environment of the vehicle;

[0069] Figure 3 Steps for determining the degree of overlap. Detailed Implementation

[0070] Figure 1 A simplified diagram of the platform's surrounding environment is shown, which is transformed into a two-dimensional pixel data domain by the imaging system. The outlines of static and dynamic objects correspond to the regions representing the respective objects in the two-dimensional pixel data domain. At least one object 20 is a simplified rear view of a truck located on the central carriageway 10b. A left carriageway 10a or a right carriageway 10c is adjacent to the left or right side of the central carriageway 10b. Due to the centering of the represented surrounding environment, it can be inferred that the central carriageway 10b is the carriageway in question.

[0071] At least one carriageway - pixel group 10b represents at least one carriageway, and according to as in Figure 3 The simplified method shown in the diagram belongs to the two-dimensional pixel data field S1. The middle carriageway 10b is bounded by the left carriageway boundary 12 or the right carriageway boundary 14. Pixels of at least one object-pixel group 20 represent at least one object and belong to the two-dimensional pixel data field S2. The width of the defined at least one object-pixel group 20 is characterized by object-pixel pairs 22, 24 and lies on the box 20 surrounding the object-pixel group 20. Figure 1 In the frame, object-pixel pairs 22 and 24 are particularly positioned at the bottom edge of the frame.

[0072] Object-pixel pairs 22 and 24 are defined S3 and compared with a carriageway-pixel group in the two-dimensional pixel data field, wherein at least one object-pixel pair 22 and 24 is compared with the carriageway-pixel group within exactly one pixel row of the pixel data field S4. Figure 1In this configuration, the carriageway-pixel group forms a horizontal line that is defined by the pixels of object-pixel pairs 22 and 24 and is bounded by the left carriageway boundaries 12 and 16 or the right carriageway boundaries 14 and 18. All pixels between object-pixel pairs 22 and 24, which together constitute the first pixel set, are completely located within the carriageway-pixel group 10b, which forms the second pixel set within the pixel row defined by object-pixel pairs 22 and 24. Therefore, the intersection of the first pixel set and the second pixel set is equal to the first pixel set. In this case, they completely overlap with a 100% overlap.

[0073] exist Figure 1 In the middle, lines E1 to E3 are simply shown at equal intervals on the driving belts 10a, 10b and 10c.

[0074] Figure 1 This also briefly illustrates how the carriageway-pixel group 10b can be assigned using at least one marker-pixel group, such as the right carriageway boundary 14, 18 represented by a two-dimensional pixel data field. For this purpose, a virtual carriageway boundary, such as the left carriageway boundary 12 of carriageway 10b, is constructed as follows: a constant carriageway width is assumed for the middle carriageway 10b. The pixel group located between these two carriageway boundaries 12, 14 forms the carriageway-pixel group 10b.

[0075] also, Figure 1 This illustrates, for example, how marker-pixel group 14, in this case the right lane boundary 14 of the middle lane 10b, can be extended by extrapolation in a two-dimensional pixel data domain. Figure 1 As indicated by dashed line 18, the marker-pixel group 14 can be extended using other pixels in the two-dimensional pixel data field. Correspondingly symmetrically, as indicated by dashed line 16, the left-hand lane boundary 12 can be extended by extrapolation.

[0076] By using the interpolated marker-pixel groups 16 and 18, at least one intermediate carriageway-pixel group 10b can be assigned and / or enlarged in the two-dimensional pixel data domain. This allows the overlap between object 20 and intermediate carriageway 10b to be determined even when the object is further away from the vehicle.

[0077] Figure 2 The diagram briefly illustrates how vehicle 32 enters lane 50, but the identified left and right lane boundaries 36 and 38 of that lane do not reach the identified object, or vehicle 34. Errors in estimating the further direction of the left or right lane boundary 40 using methods in a three-dimensional coordinate system could lead to incorrect estimations of the overlap between the identified object 34 and lane 50. The intervals E', E0' to E5' are again given at equal intervals.

Claims

1. Method for determining the coincidence of at least one object with at least one traffic lane by presenting the surroundings of a platform as a two-dimensional pixel data field, the method having the following steps: S1 : attributing at least one traffic lane-pixel group (10a, 10b, 10c) to pixels of the two-dimensional pixel data field which correspondingly represent at least one traffic lane; S2: attributing at least one object-pixel group (20) to pixels of the two-dimensional pixel data field which correspondingly represent at least one object; S3: defining at least one object-pixel pair (22, 24) in the two-dimensional pixel data field, which object-pixel pair characterizes the width of the at least one object-pixel group (20); S4: comparing the object-pixel pair (22, 24) with the traffic lane-pixel group (10a, 10b, 10c) in the two-dimensional pixel data field, comparing a first set of pixels located between and containing the object-pixel pair (22, 24) in the two-dimensional pixel data field with a second set of pixels of the traffic lane-pixel group (10a, 10b, 10c) which are located in the same pixel row of the two-dimensional pixel data field as the object-pixel pair (22, 24), determining a pixel intersection of the first set of pixels with the second set of pixels, determining the coincidence of the at least one object with the at least one traffic lane by calculating the proportion of the number of pixels of the determined intersection to the number of pixels of the first set. comparing the object-pixel pair (22, 24) with the traffic lane-pixel group (10a, 10b, 10c) in exactly one pixel row of the two-dimensional pixel data field; and / or wherein in the two-dimensional pixel data field the width of the object is calculated by means of the first set and the width of the traffic lane is calculated by means of the second set. attributing the at least one traffic lane-pixel group (10a, 10b, 10c) to pixels of the two-dimensional pixel data field by means of at least one marker-pixel group (12, 14) of the two-dimensional pixel data field which represents a traffic lane boundary. extending at least one marker-pixel group (12, 14) by means of an extrapolation in the two-dimensional pixel data field. attributing and / or enlarging at least one traffic lane-pixel group (10a, 10b, 10c) in the two-dimensional pixel data field by means of the extrapolated marker-pixel group. wherein attributing the at least one traffic lane-pixel group (10a, 10b, 10c) in the two-dimensional pixel data field by means of data of a sensor system. recognizing the at least one traffic lane-pixel group (10a, 10b, 10c) in the two-dimensional pixel data field by means of map data of the surroundings of the platform.

8. Method for recognizing a target object, having the following steps: attributing pixels of a two-dimensional pixel data field to at least two traffic lane-pixel groups (10a, 10b, 10c) which correspondingly represent at least two traffic lanes of the surroundings in the two-dimensional pixel data field; 2. The method of claim 1, wherein, ​ ​ 3. The method of claim 1 or 2, wherein, ​ 4. The method of claim 3, wherein, ​ 5. The method of claim 4, wherein, ​ 6. The method of any one of claims 1, 2, 4, and 5, wherein, ​ 7. The method of any one of claims 1, 2, 4, and 5, wherein, ​ ​ ​ selecting a current lane-pixel group from the at least two lane-pixel groups; and determining a degree of coincidence of the at least one object-pixel group with each of the at least two lane-pixel groups according to the method of any one of claims 1 to 7.

9. A method for attributing at least one object to a lane, having the following steps: attributing pixels of a two-dimensional pixel data field to at least two lane-pixel groups (10a, 10b, 10c), which respectively represent at least two lanes of a surrounding environment in the two-dimensional pixel data field; attributing at least one object-pixel group (20) to pixels of the two-dimensional pixel data field, which respectively represent at least one object; and determining a degree of coincidence of the at least one object with each of the at least two lane-pixel groups (10a, 10b, 10c) according to the method of any one of claims 1 to 7.

10. A method wherein, sending a control signal for controlling an at least partially automated vehicle and / or a warning signal for warning a vehicle occupant based on the degree of coincidence of a lane and an object determined according to the method of one of claims 1 to 7.

11. An apparatus for implementing the method of any one of claims 1 to 10.

12. A computer program product comprising a computer program which comprises instructions arranged to cause a computer to implement the method of any one of claims 1 to 10 when the program is executed by the computer.

13. A machine-readable storage medium having stored thereon a computer program which comprises instructions arranged to cause a computer to implement the method of any one of claims 1 to 10 when the program is executed by the computer.

Images