Method and device for multi-sensor data fusion for automated and autonomous vehicles

Abstract

Computer-implemented method (30) for estimating a road course in an environment of a vehicle (11) based on a state function describing the road course, wherein the state function comprises a clothoid spline (20), the method (30) comprising: Providing (31) environmental measurement data describing the course of the road at a current position of the vehicle (11), wherein the environmental measurement data comprise at least one polynomial function; Transforming (32) the state function and the environmental measurement data into a common coordinate system; and Checking (33a) the environmental measurement data for an error; if no error was detected, adapting (33) the state function based on the environmental measurement data in the common coordinate system, wherein the adapting (33) of the state function comprises applying a measurement model to the clothoid spline (20), in which a clothoid segment (21) is represented by a parameter representation of the clothoid function, wherein in the parameter representation each sample point (63) of the clothoid segment (21) is represented in a vector representation, if an error is detected, storing (33b) the error.

Description

[0001] Embodiments of the present invention relate to a computer-implemented method, a vehicle, a computer program, and a device for estimating a road course. The road course is described by a state function comprising a clothoid spline.

[0002] When controlling some automated or autonomous vehicles, methods are used to estimate the course of the road, on the basis of which the control system of such a vehicle makes driving decisions.

[0003] The road course can be approximately described by determining a road model corresponding to the state function. This allows the road course to be determined, in particular, relative to a position and / or orientation of the vehicle. Properties of the road course can include, for example, the width, direction, curvature, or number of adjacent lanes, or the position of the vehicle relative to the lanes (for example, the vehicle is in the right, middle, or left of three lanes).

[0004] The vehicle can then be controlled in such a way that it can, for example, follow the course of the road, avoid obstacles or change lanes. For highly available, safe and comfortable operation of the vehicle, environmental measurement data from several data sources can be combined (fused) when determining the course of the road or road model. Combining data sources, also known as data fusion, means combining data from several data sources. Typical data sources can be, for example, sensor data sources such as a camera or waypoints on a digital map. Such a digital map can, in particular, be a low-resolution (SD) map or a high-resolution (HD) map. There are various options for this type of data fusion.

[0005] Document WO 2018 / 019 454 A1 describes a method for determining a road model. A plurality of hypotheses for a road model are generated using different parameterizations of map data and observation data. A hypothesis for a road model is selected based on the confidence values ​​of the hypotheses. A method for fusing the observation data and camera data, which ensures highly available determination of the road course, is not considered here.

[0006] DE 10 2016 214 045 A1 relates to a method and a device for determining a road model by means of recursive estimation, as well as to a vehicle. The recursion comprises the following steps: sensory acquisition of observation data characterizing the surroundings of the vehicle or its movement; acquisition of map data characterizing the surroundings of the vehicle; generation of a plurality of different hypotheses for the road model to be determined according to a parameterized state function, wherein each of the hypotheses represents a possible road model for the current recursion step and is characterized by a different association, generated by means of appropriate parameterization, of the observation data and map data serving as input variables of the state function;Estimating a confidence value with respect to a predetermined confidence measure for each of the hypotheses of the current recursion step using a recursive estimation method, including at least one hypothesis from the previous recursion step as an input variable; Selecting, based on the estimated confidence values, one of the hypotheses as a road model for the environment of a vehicle at the current recursion step;

[0007] DE 10 2013 003 944 A1 relates to methods for providing driver assistance functionality in a vehicle. The method comprises loading a predefined catalog of road scenarios from a database, wherein the catalog comprises a plurality of road scenarios, each describing a road situation by at least one parameterized environmental feature. The method further comprises obtaining sensor data from at least one sensor of the vehicle, each describing at least one measured environmental feature in an environment of the vehicle. The method further comprises, for at least some of the road scenarios of the predefined catalog: evaluating the respective road scenario by assigning a probability value to the respective road scenario, which is obtained at least from a comparison of the at least one measured environmental feature with the at least one parameterized environmental feature of the respective road scenario.The method further comprises selecting a road scenario depending on the associated probability values ​​of the at least some road scenarios and providing the driver assistance functionality depending on the selected road scenario and / or depending on the measured environmental features.

[0008] US 6,751,547 B2 discloses a method and apparatus for accurately estimating the forward path geometry of a vehicle based on a two-track road model. Road data provided by a camera or radar system is acquired, and a complete measured value transfer function of the two-clothoid model is calculated. The near-range and far-range coefficients of the clothoids are estimated simultaneously; and the vehicle's forward path is estimated using the data provided by the two-clothoid model.

[0009] DE 10 2017 005 921 A1 discloses a method for the autonomous detection of a lane course of at least one lane, comprising at least a first method step in which at least two lane boundaries are detected by means of at least one environmental sensor, comprising at least a second method step in which lane boundary courses of the detected lane boundaries are described by means of a computing unit at least approximately by means of an arc model, and comprising at least a third method step in which the lane boundary courses are compared by means of the computing unit, wherein a parallelism of the detected lane boundaries is determined on the basis of curvature values ​​of the lane boundary courses.

[0010] DE 10 2017 212 254 A1 discloses a method for predicting the future course of the road for a vehicle traveling on a road, wherein the road has a first section, a curve and a second section, and the curve is arranged between the first and the second section, comprising the following steps: - Determining an approximate roadway course derived from digital map material by means of an original clothoid segment, wherein the clothoid segment has a starting point on the first section, an end point on the second section, and the clothoid segment covers the angle specified by the curve between the first and the second section, - Correcting the original clothoid segment by subsequent corrected clothoid segments, whereby the corrected clothoid segments take into account the curvature of the path actually traveled after the starting point.

[0011] Other conventional control systems typically use data exclusively from a single data source (camera, digital map) to create a road model. High-availability determination of the road course cannot be guaranteed, as if the data source malfunctions, determining the road course is no longer possible.

[0012] Furthermore, individual data sources can generate noisy or partially erroneous environmental measurement data, which can make an estimate of the road course inaccurate. Therefore, there is a need to create a computer-implemented method for estimating the road course that ensures highly available and precise determination of the road course. The computer-implemented method, the device, the computer program, and the vehicle according to the independent claims address this need. Embodiments of the present invention provide a computer-implemented method, a vehicle, a computer program, and a device that serve to estimate a road course in the environment of the vehicle.By fusing environmental measurement data from multiple data sources with a state function (road model) describing the road course, high availability and high precision of the estimation can be ensured.

[0013] The environmental measurement data from the respective data sources are output variables from measurements of the road surface and include at least one polynomial function. The state function includes a clothoid spline. The data fusion of the environmental data with the state function involves adapting the state function, or clothoid spline, to the environmental measurement data using a real-time filter. For estimating the road surface, a Kalman filter, for example, has proven to be an advantageous real-time filter.

[0014] Adapting the state function with the Kalman filter typically requires transforming the state function from a state space into a measurement space. In embodiments of the present invention, this may involve transforming the state function between different coordinate systems. In the measurement space, a state function resulting from a prediction by the Kalman filter can be corrected using environmental measurement data from multiple data sources or sensors. In the context of the Kalman filter, this is typically referred to as innovation. The state function updated by the adaptation or innovation corresponds to an approximate estimate of the road's course. By repeatedly adapting the state function to environmental measurement data over time, the estimate is continuously updated, so that, for example, the course of a road's course can be estimated precisely and reliably.Adapting the state function to environmental measurement data from multiple data sources ensures high availability and high accuracy in estimating the road course. A first aspect of the invention relates to a computer-implemented method for estimating a road course in the environment of a vehicle based on a state function describing the road course. The state function comprises a clothoid spline.

[0015] The computer-implemented method further comprises providing environmental measurement data that describe the road course in a specific vicinity of the vehicle's current position. The environmental measurement data comprises at least one polynomial function. The polynomial function can be generated, for example, from a polyline using a curve regression method. Such a polyline refers to a sequence of connected line segments. The computer-implemented method further comprises transforming the state function and the environmental measurement data into a common coordinate system and adapting the state function based on the environmental measurement data in the common coordinate system.

[0016] For example, the computer-implemented method is based on a recursive method, such as the Kalman filter, which can comprise successive recursion steps. Each recursion step can be used to estimate the roadway course, which is described by the state function. Since clothoids are often used in road construction when calculating the roadway course, the roadway course can be precisely estimated or approximately reconstructed using successive clothoid segments. Therefore, it can be advantageous to assume the clothoid spline for the state function when estimating the roadway course. The clothoid spline can be completely determined by parameters such as curvature, change in curvature, angle, and y-intercept.

[0017] The input variables for estimating the road course are the ambient measurement data. The ambient measurement data can be in the form of polynomial functions or polynomial splines, which can be derived from measurement data from external and / or internal sensors. The ambient measurement data, for example, approximately represent the observed road course in the vehicle's surroundings.

[0018] The environmental measurement data can then be compared with the state function. For this purpose, it is advantageous if the state function and the environmental measurement data are present in the same measurement space. Transforming the state function into another coordinate system is often suitable for this purpose. For example, the clothoid spline can be represented in a coordinate system with curvature values ​​and arc lengths, whereas the environmental measurement data is present in a coordinate system with spatial coordinates. For this reason, a measurement model is applied to the state function in order to at least approximately represent the state function in the coordinate system with spatial coordinates. The clothoid spline, which is typically represented as a function of an arc length, is approximately represented in spatial coordinates in a parametric representation. In the parametric representation, each sample point of a clothoid segment is represented in a vector representation.The state function captured during prediction can then be updated with the environmental measurement data in the Kalman filter.

[0019] The updated state function is typically obtained as the output of a recursion step. For a highly available and precise computer-implemented method for determining the road course, it can be advantageous to use environmental data from multiple, possibly different, data sources or sensors for the data fusion of the recursion step. This type of data fusion is called multi-sensor data fusion.

[0020] In each recursion step, a check can also be performed to determine whether there is an error in the underlying environmental data. Such an error could occur, for example, if lane markings have been altered due to construction. In such a case, the environmental data of a map and a camera differ. Furthermore, the quality of the lane markings could be poor, or the camera could produce an incorrect measurement.

[0021] The check can be performed by determining a deviation between a value of the state function determined in a prediction step of a Kalman filter and the current environmental data and comparing this deviation with a specified threshold. If the determined deviation is greater than a specified (upper) threshold or less than a specified (lower) threshold, an error has occurred.

[0022] This error can then be saved, and the current environmental data will no longer be used to update the status function. Saving may also include marking a faulty lane marker corresponding to the error.

[0023] Furthermore, the error and corresponding data, such as the highlighted faulty lane marking, can then be sent to a receiving unit located externally of the vehicle. Depending on the error, the transmitted data can be used in a variety of ways, for example, by a road authority to improve the faulty lane marking on a road and / or by a vehicle manufacturer to improve a camera / LIDAR lane marking detection function, particularly for manual labeling to improve learning algorithms. Furthermore, the transmitted data can be used to decide whether autonomous driving should be deactivated at this faulty lane marking. This data can thus be used to increase the reliability and safety of autonomous driving functions.Furthermore, the sent data can be checked to determine whether the error was caused by faulty data from a camera or faulty data from a digital card.

[0024] For example, the message can be sent automatically after the error is detected. Alternatively or additionally, the message can be sent in batches, e.g., always at night or always after the vehicle is parked, or similarly. Furthermore, the message can also be sent to other vehicles.

[0025] In some embodiments, the ambient measurement data can be captured by at least one camera. For example, a camera can be used as one of several sensors, or several cameras can be used for multi-sensor data fusion. Images of a roadway captured by the at least one camera can be used to detect the road course using a suitable image processing application. From the images, for example, a course of road markings or road boundaries can be determined and approximately represented by at least one polynomial function. The at least one polynomial function corresponds to the ambient measurement data. For a more precise estimation, the road course can also be approximated using a plurality of connected polynomial functions.

[0026] Capturing the course of the road using multiple cameras can also contribute to a more accurate estimation of the road course.

[0027] Additionally or alternatively, environmental measurement data can be provided from a map. The map is available, for example, as a digital map with a specific spatial resolution. High-resolution maps (HD maps), for example, have a spatial resolution of 10 to 25 cm, which means that structures or objects whose dimensions correspond to the respective spatial resolution in at least one dimension can be displayed with the accuracy of the spatial resolution. For example, the roadway, road boundaries, or road markings can be detected in this way. Waypoints that indicate the course of the road can be taken from the map. By interpolating the waypoints, a polynomial or, as is common, a polynomial spline of several polynomials can be determined, which approximately determine the course of the road based on the waypoints on the map. The polynomial or polynomial spline is referred to as map data.In some embodiments of the present invention, first environmental measurement data can be acquired by at least one camera, and second environmental measurement data can be provided by a map. Based on the environmental measurement data acquired by the camera or determined from the map, a polynomial or a polynomial spline can be determined, each of which approximately describes the road course in the surroundings of the vehicle. The state function can be adapted to the first environmental measurement data and additionally to the second environmental measurement data. Adapting the state function to environmental measurement data from several different sensors or data sources (camera and map) can result in high availability and increased precision in determining the road course.High availability means that even if a data source fails, the road course can still be reliably estimated, so that in this case the vehicle can still be steered to a safe parking position or to a workshop, for example.

[0028] In further embodiments, the computer-implemented method may include transforming the state function and the environmental data into a common coordinate system with spatial coordinates. The innovation of the Kalman filter typically requires transforming the state function into the measurement space in order to update the state function with the environmental measurement data. Typically, the camera's environmental measurement data is already available in a coordinate system with spatial coordinates. Transforming the state function thus means converting it using a measurement model so that the state function can be represented in the measurement space, the coordinate system with spatial coordinates. The clothoid spline is described, for example, by one or more third-degree polynomials or, in the parametric representation, by Fresnel integrals.To reduce the numerical complexity of data fusion, the Fresnel integrals can be approximated by a Taylor series. The advantage of data fusion in the spatial coordinate system is that the orientation and absolute position of the vehicle on the road can be determined. For example, this can be used to determine which lane of a multi-lane roadway the vehicle is in and / or whether it is moving along the roadway or across it.

[0029] Further embodiments of the present invention further comprise a sampling method for determining sample points for adapting the state function. With multiple sample points, it can be advantageous if a constant curve length lies between adjacent sample points. The sample points can be determined, for example, using a suitable mathematical method. The constant curve length can serve as an input variable for the mathematical method and can, for example, be predetermined or determined depending on a property of the road course, such as a road curvature.

[0030] For adaptation using the Kalman filter, it can be advantageous if the state function and / or the environmental measurement data are present in a point space of the measurement space. This means that, for example, both the state function and the environmental measurement data are each represented by a plurality of sample points, which is advantageous for adaptation. Using the sample points, the Kalman filter can adapt the state function to the environmental measurement data. Typically, the sample points are present in a coordinate system with spatial coordinates.

[0031] Optionally, the sample points can be determined exclusively for the state function. The sample points are then interpolated, for example, using a polynomial of the same order as the polynomial of the ambient measurement data. The Kalman filter can then be designed so that, in each recursion step, polynomial parameters of the transformed state function are approximated to further polynomial parameters of the ambient measurement data. Using a measurement model, parameters of the state function can be determined from the polynomial parameters, so that the state function with the parameters approximately corresponds to the ambient measurement data.

[0032] With a plurality of sample points of the state function and the ambient measurement data, the sample points can be selected depending on a sampling method in such a way that it is advantageous for adapting the state function to the ambient measurement data. Sample points to be compared are selected so that they lie at the same arc length of the state function or the ambient measurement data. Furthermore, with a plurality of sample points, these can be selected so that a constant arc length of the curve lies between adjacent sample points of a curve (state function, ambient measurement data).

[0033] This allows a state function with pronounced curvatures to be precisely adapted to the camera's ambient measurement data. In some embodiments, this ensures reliable estimation of a potentially highly curved road course.

[0034] In further embodiments, the computer-implemented method can comprise transforming the state function and the environmental data into a common coordinate system with curvature values. The measurement space comprises a two-dimensional coordinate system determined by curvature values ​​and arc lengths. The clothoid spline is typically present in this measurement space. The map data is typically present as a polynomial spline in a coordinate system with spatial coordinates, which can result from an interpolation of the waypoints. The polynomial spline can be transformed such that the curvature values ​​of the polynomial spline can be determined depending on the arc length of the polynomial spline. The state function and the polynomial spline can thus be represented in the measurement space, which comprises the coordinate system with the curvature values ​​and the arc lengths.The curvature values ​​of the state function are thus comparable to the curvature values ​​of the polynomial spline for the Kalman filter.

[0035] In some embodiments of the present invention, adapting the state function to the environmental measurement data may involve adapting at least one curvature value of the state function to at least one other curvature value of the environmental measurement data. With the innovation of the Kalman filter, for example, the curvature values ​​of the state function and the polynomial spline can be used as input variables. Data fusion of the map data with the Kalman filter means adapting the curvature values ​​of one or more points of the state function to the curvature values ​​of a transform of the map data.

[0036] The computer-implemented method described herein can be designed, in particular, to determine the course of ordinary roads or lanes. Roads, especially on highways, highway-like roads, and in urban areas, are usually provided with lane markings to indicate the lane course, for example, to a driver. The lane markings can be used, for example, to estimate the lane course.

[0037] According to further embodiments, the state function can identify a course of the lane markings of the road. For this purpose, images of the lane markings can be taken by the at least one camera. Detection of the lane course by means of the at least one camera is usually carried out via an image processing application, which can determine a course of lane markings or lane boundaries from images taken by the at least one camera. Environmental measurement data from the at least one camera can thus identify the course of the lane markings or lane boundaries. This estimation can be advantageous when controlling the vehicle, for example to determine a lane width or a distance of the vehicle from the lane boundary.

[0038] In further embodiments, determining the course of a lane center by transforming the state function may be advantageous. In some cases, the course of the lane center is required for vehicle control. For example, ambient measurement data from at least one camera describes the course of lane boundaries or lane markings located to the side of the road. Therefore, by fusing this ambient measurement data, the course of the lane markings or lane boundaries can be estimated.

[0039] If the road width is known, this estimate can be used to determine the state function that describes the path of the lane center. By translating and changing curvature values, the state function that describes the path of the lane center can be deduced from the path of the lane markings and lane boundaries.

[0040] Optionally, in further embodiments, the computer-implemented method can further comprise providing validation measurement data and validating the estimation of the roadway course using the validation data. Additional sensors, such as lidar, radar, and / or ultrasonic sensors, can provide the validation measurement data, which, for example, identify the position and / or course of roadside structures based on location coordinates. Such structures can be, for example, guardrails, noise barriers, or guide posts.

[0041] The validation measurement data thus identify areas accessible to the vehicle. When validating the road course estimate, it can be determined whether the state function lies within the accessible area, thus allowing a statement to be made as to whether the estimated road course can be realistic. Validation increases the robustness of the computer-implemented method against errors in the road course estimation. For example, an incorrect estimate, i.e., a state function that does not lie within the accessible area, can be disregarded when controlling the vehicle. A second aspect of the present invention relates to a device for estimating the road course in the surroundings of the vehicle. The device comprises one or more interfaces for acquiring environmental measurement data.Furthermore, the device comprises a processing unit which is designed to carry out the computer-implemented method for estimating the road course in the surroundings of the vehicle.

[0042] The environmental measurement data can be, for example, map data or environmental measurement data from the camera. Accordingly, an interface of the device is, for example, a connection to a storage medium on which the map data is stored and from which it can be retrieved or transmitted. Further interfaces could be connections to one or more cameras that provide the environmental measurement data.

[0043] The processing unit can be, for example, a computer, a microcontroller, a processor, or a programmable hardware component configured to execute the computer-implemented method. The environmental measurement data acquired by the interfaces are transmitted from the interfaces to the processing unit, so that the environmental measurement data serves as an input for estimating the road course.

[0044] A third aspect relates to a vehicle with the device for estimating the road course in the vehicle's surroundings. The vehicle can be any ground-based vehicle, in particular a passenger car, truck, motorcycle, or agricultural machine.

[0045] Exemplary embodiments are explained in more detail below with reference to the accompanying figures. They show: Fig. 1 a schematic representation of a multi-lane road with a vehicle on it. Fig. 2 an example of a clothoid spline with n clothoid segments. Fig. 3 a schematic flow diagram of the computer-implemented method for estimating the road course in the environment of the vehicle. Fig. 4 a flowchart of a multi-sensor data fusion. Fig. 5 a schematic representation of a sampling procedure of the clothoid spline and the environmental measurement data. Fig. 6 a graphical comparison of two approximation methods for approximating the clothoid spline. Fig. 7 a graphical comparison of two sampling methods. Fig. 8 a schematic illustration of a roadway marked by road markings. Fig. 9 a representation of the map data and the vehicle on the road. Fig. 10 a schematic illustration of the vehicle with a device for estimating the road course.

[0046] Various embodiments will now be described in more detail with reference to the accompanying drawings, in which some embodiments are illustrated. In the figures, the thickness dimensions of lines, layers, and / or regions may be exaggerated for clarity.

[0047] In the area of ​​autonomous driving concepts, estimating the environment and the road course can be of great importance for the control of, for example, autonomously driving vehicles. For example, the vehicle can be controlled in such a way that it autonomously, without driver intervention, regulates steering, acceleration, speed, and other vehicle or driving parameters so that the vehicle can, for example, safely follow the estimated road course and / or avoid obstacles.

[0048] In Fig. Figure 1 shows, by way of example, a schematic roadway profile of the roadway 14 on which the vehicle 11 is located and / or moving. Embodiments of the invention described herein can be configured to estimate the roadway profile in two translational directions, i.e., in two dimensions. Fig. 1 shows the roadway from a top view in two dimensions. Vehicle 11 is located in a middle of three lanes of roadway 14, which may be driven in one or both directions, for example. In advantageous embodiments of the invention described herein, for example, a course of roadway 14 or of the lane can be estimated. In a broader sense, in the present disclosure, the course of the roadway can refer to both the course of the roadway and the course of the lane.

[0049] A lane center and / or the course of the lane center may be marked by a center line 12. However, roads typically do not have a center line 12, so the lane center is not visible. The individual lanes are typically delimited by road markings, such as a boundary line 13.

[0050] Conventional methods from the state of the art, for example, determine the course of the road using a road model based on polynomials or polynomial splines, which can, for example, result directly from camera images or from the digital map.

[0051] However, during construction planning, the roadway 14 is typically constructed according to a spline (clothoid spline) of connected clothoid segments for driving comfort and safety reasons. The advantage of such a design of the roadway 14 is a linear change in the curvature of the roadway, so that curves can be negotiated with high driving comfort and can be easily estimated by the driver of non-autonomous vehicles. Due to a design based on the clothoid spline, it can prove advantageous to assume a roadway model based on a clothoid spline when estimating the roadway course. For example, this can increase the accuracy and reliability of the estimation.

[0052] The roadway model of roadway 14, which is based on a clothoid spline 20, is shown as an example in Fig. 2. The clothoid spline 20 comprises the clothoid segments 21 and the transition points 22 between two clothoid segments 21. The clothoid segments 21 correspond to parts of clothoid functions.

[0053] The clothoid functions, or curvature values ​​c(s) of the clothoid functions, can be mapped depending on the arc length s according to: c(s)+c0+c1,m∗s

[0054] Where c0 represents an initial curvature and c 1,m for a constant change in curvature of the m-th clothoid segment 41 with the arc length s. From (1) it can be seen that the curvature value c(s) of the clothoid function varies linearly with c 1,m changes.

[0055] In the Fig. For the clothoid spline 20 shown in Figure 2, an orientation θ0 and a y-axis intercept y0 of a clothoid segment 21 can also be specified. In general, each of the clothoid segments 21 of the clothoid spline 20 can be completely characterized by the parameters c0, c1, θ0, and y0. For the clothoid spline 20 with n (n ∈ ℕ) clothoid segments 21, this would mean that the clothoid spline 20 can be described by 4*n parameters.

[0056] The Fig. The clothoid spline 20 shown in Figure 2 satisfies connection conditions between the clothoid segments 21, which can reduce the number of parameters required to describe the clothoid spline 20, so that the computational effort for determining the clothoid spline 20 is reduced. For example, the clothoid spline 20 is continuous and differentiable at the transition points 22, and the curvature values ​​of two consecutive clothoid segments 21 are equivalent at the transition points 22. Under the connection conditions mentioned, the number of parameters required to describe the clothoid spline 20 can be reduced to n + 3. Therefore, the clothoid spline 20 can be fully characterized by a reduced set of parameters (c).

[0057] The clothoid spline 20 can represent the course of the roadway 14, for example, both in front of and behind the vehicle 11, to a good approximation. Environmental measurement data from several, possibly different, data sources or sensors are used to determine the clothoid spline 20. Particularly when using different types of sensors, it may be necessary to determine the clothoid spline 20 in such a way that the environmental measurement data from the different types of sensors are included in the determination of the clothoid spline 20. This need can be met, for example, by embodiments of the Fig. 3 shown computer-implemented method 30 for estimating the road course.

[0058] The computer-implemented method 30 for estimating the road course in the vehicle's surroundings is based on determining the state function describing the road course. The state function includes the clothoid spline 20.

[0059] The computer-implemented method 30 comprises providing 31 environmental measurement data that describe the road course at a current position of the vehicle 11. The environmental measurement data comprise the at least one polynomial function. The computer-implemented method 30 further comprises transforming 32 the state function and the environmental measurement data into the common coordinate system and adapting 33 the state function based on the environmental measurement data in the common coordinate system.

[0060] The ambient measurement data from the vehicle's own sensors, such as the camera or several cameras, can, as in a present exemplary embodiment, describe a course of the boundary lines 13 or the course of the lane center 12 of the roadway using a polynomial function or a polynomial spline.

[0061] An HD map from a map manufacturer can also be used to estimate the road's course. Environmental measurement data can be determined based on waypoints on the road that can be taken from the HD map. These waypoints can, for example, mark the center of a lane. The resulting map data typically describes the road's course using another polynomial spline.

[0062] Real-time filters, such as a Kalman filter, may, for example, make it possible to update the state function, or rather the clothoid spline 20, with the environmental measurement data, provided that the state function and the environmental measurement data are located in the common coordinate system. For example, the state function, as in illustration (1), is present in a state space that includes curvature values. The environmental measurement data is present, for example, in a measurement space with spatial coordinates. The state function can be mapped by transformation 32, for example, in the common coordinate system, which can include curvature values ​​or spatial coordinates.

[0063] In the common coordinate system, for example, the Kalman filter can use the environmental measurement data to adjust 33 the clothoid spline 20 to the environmental measurement data.

[0064] The computer-implemented method 30 can, for example, describe a single recursion step of the recursive method. The recursive method comprises, for example, several consecutive recursion steps, each of which serves to estimate the roadway path. The recursive method can be implemented, for example, using the Kalman filter or other real-time filters. In some embodiments, the Kalman filter has proven advantageous.

[0065] Input variables of the individual recursion steps are the state function from one of the previous recursion steps and the environmental measurement data, which can be captured by providing 31 the environmental measurement data by means of the at least one camera and the HD card.

[0066] Capturing and transforming 32 the state function of a previous recursion step characterizes the first phase of the Kalman filter. This phase is called prediction.

[0067] During the prediction of the Kalman filter, an estimate for a current state of the clothoid spline 20 can be determined. For this purpose, the clothoid spline can be subjected to state dynamics. The state dynamics are determined, for example, by a movement of the vehicle 11. For example, if the vehicle 11 is moving along the roadway 14, the clothoid spline 20 can be extended in front of the vehicle 11 and shortened behind the vehicle 11.

[0068] Since the clothoid spline 20 is usually composed of individual segments 21, the clothoid spline 20 cannot, for example, be continuously extended or shortened at each recursion step, but only in the event that the vehicle 11 passes over one of the transition points 22.

[0069] In each recursion step, a check can also be performed to determine whether there is an error in the underlying environmental data 33a. The check can be performed by determining a deviation between a value of the state function determined in a prediction step of a Kalman filter and the current environmental data and comparing this deviation with a predefined threshold. If the determined deviation is greater than a predefined (upper) threshold or less than a predefined (lower) threshold, an error has occurred. This error can then be stored 33b and not used further for the state function, i.e., not in a phase referred to as innovation. Storing can also include marking an erroneous lane marking corresponding to the error.Furthermore, the error and corresponding data, such as the highlighted faulty lane marking, can then be sent to a receiving unit located externally of the vehicle. Depending on the error, the transmitted data can be used in a variety of ways, for example, by a road authority to improve the faulty lane marking on a road and / or by a vehicle manufacturer to improve a camera / LIDAR lane marking detection function, particularly for manual labeling to improve learning algorithms. Furthermore, the transmitted data can be used to decide whether autonomous driving should be deactivated at this faulty lane marking. This data can thus be used to increase the reliability and safety of autonomous driving functions.Furthermore, the sent data can be checked to determine whether the error was caused by faulty data from a camera or faulty data from a digital card.

[0070] If there is no error, the following process is carried out.

[0071] The state function of the previous recursion step is updated by adaptation 33 to the environmental measurement data in each recursion step. The adaptation 33 corresponds to a second phase of the Kalman filter, referred to as innovation. For adaptation 33, for example, the state function is compared with the environmental measurement data. For this purpose, individual values ​​of the state function are compared with further values ​​of the environmental measurement data. Depending on the measurement space, these values ​​can be, for example, spatial coordinates or curvature values ​​of the state function and the environmental measurement data. Typically, the values ​​of the state function have an uncertainty, or the values ​​of the environmental measurement data have a measurement uncertainty, which can be represented by a probability distribution, such as a Gaussian distribution.In the second phase of the Kalman filter, the state function can be adapted 33 using the state function and the ambient measurement data as input variables. The weighting of the input variables during the adaptation 33 of the state function can depend on the measurement uncertainty or the uncertainty. The lower the measurement uncertainty of the ambient measurement data, the more closely the state function determined during prediction can be approximated to the ambient measurement data. The greater the measurement uncertainty of the ambient measurement data 14, the lower the weighting of the ambient measurement data can be during the adaptation 33. The uncertainty of the state function present as an input variable is based, for example, on mean values ​​and information on the dispersion (e.g., covariances) of the parameters of the clothoid spline 20. The mean values ​​and covariances can be determined from the parameters of previous recursion steps.The uncertainty of the state function can, for example, be determined by the covariances. If the covariance is high, the state function captured during the prediction may be lightly weighted.

[0072] With low covariance, however, the state function is heavily weighted. Depending on the degree of uncertainty and measurement uncertainty, an updated state function or clothoid spline 20 results from the ambient measurement data and the state function serving as the input variable. In a subsequent recursion step, the most recently updated state function can again be used for the prediction.

[0073] The environmental data from individual sensors (sensors from the map manufacturer and the vehicle's own sensors) cannot accurately determine the road course. The environmental measurement data from the sensors can, for example, be partially noisy or inaccurate. However, the fusion of the environmental data from several different sensors (data sources) can ensure a robust, highly available, and low-error estimation of the road course. This concept of multi-sensor data fusion is used in Fig. 3 illustrates.

[0074] In multi-sensor data fusion, data, especially environmental measurement data, from a number of sensors can be fused to form the road model. Fig. In the embodiment shown in Figure 4, the environmental measurement data can be provided by means of the camera 41 and the HD card 43.

[0075] The camera 41 can, for example, be mounted on the vehicle and directed in the direction of travel. As already mentioned, the images from the camera 41 can be used to approximately represent the lane markings, lane boundaries, and their course in the form of one or more consecutive polynomial functions using an image processing application. Typically, the range or measurement radius of the camera 41 is limited by obstacles or the optical range of the camera 41.

[0076] The HD map 43 is typically based on output variables from a sensor system of a map manufacturer 42. By interpolating the waypoints that can be taken from the HD map, one or more connected polynomial functions can be determined to approximately describe the roadway. The HD map 43 or corresponding map data can be present, for example, on a storage medium attached to the vehicle 11.

[0077] Alternatively or additionally, the HD map 43 or the map data could be transmitted from a transmitter to a receiver module of the vehicle 11 or a device for estimating the course of the road.

[0078] By adapting 33 the state function in the second phase of the Kalman filter, for example, according to the multi-sensor data fusion 40, the environmental measurement data from the HD map 43 and the camera 41 are incorporated into the estimation of the road course or the road model 45, which can be represented by the state function or the clothoid spline 20. Although the embodiment described here provides for the use of a single camera, further embodiments can include a plurality of cameras 41, which can optionally be oriented in different directions.

[0079] With the Kalman filter, the clothoid spline 20 can be adapted, for example, to the environmental data of the camera 41. In the following, the Fig. The data fusion 44 of the ambient measurement data of the camera 41 will be discussed in more detail in the curves 20 and 50 shown in Figure 5.

[0080] For the adaptation 33 of the state function 20 in the second phase of the Kalman filter, the state function 20, i.e., the clothoid spline 20, and the environmental measurement data 50 are transformed into a point space. This means that both the environmental measurement data 50 and the clothoid spline 20 can each be described by a plurality of points in the point space. The points can be determined using a sampling method from the clothoid spline 20 and the environmental measurement data 50. Therefore, the points are also called sample points 51 and 52.

[0081] Using the usual representation (1) for the clothoid spline 20, it is not possible to determine sample points 51 or spatial coordinates of the sample points 51. Therefore, the transformation 32 of the clothoid spline 20 is necessary. For the transformation 32 of the clothoid spline 20, a measurement model can be determined, with the aid of which the clothoid spline 20 can be transformed into the measurement space of the ambient measurement data 50, so that the clothoid spline 20 can be represented by spatial coordinates.

[0082] There are several measurement models for transforming 32 the clothoid spline 20 to represent it by spatial coordinates instead of arc lengths and curvature values ​​as in (1). In the following, an example will be given using the Fig. 6 shown Fig. Two measurement models will be discussed that can be used to transform 32 the clothoid spline 20.

[0083] A first measurement model for the transformation 32 of the clothoid spline 20 provides an approximate representation of the clothoid spline 20 by approximating the clothoid segments 21 by third degree polynomials 62. This representation is shown in the Fig. The third-degree polynomial 62 can be represented as follows: y(x)=ax2+bx2+cx+d a, b, c, and d correspond to parameters that determine a form of the polynomial 62. For the approximate representation of the clothoid segment 21, the parameters a, b, c, and d can be replaced by the parameters c0, c 1,m , θ0 and y0 of the clothoid segment 21 are replaced as follows: a=c1,m, b=c0, c=atan(θ0), d=y0

[0084] By inserting the parameters according to (3) one obtains, for example, the polynomial 62, which approximately represents a course of the clothoid or, for example, the clothoid segment 21.

[0085] Thus, sample points 61 approximately correspond to sample points 51. By approximately representing the clothoid segment 21 using (2) and (3), spatial coordinates can be assigned to each sample point 61. Using the first measurement model described here, the clothoid spline 20 can be represented in the Kalman filter in the point space. Using sample points 61 and sample points 52 of the point space as input variables, the Kalman filter can approximate the polynomial 62 to the ambient measurement data 50. Values ​​for the parameters c0, θ0, y0, and c 1,1 ...c 1,n of the clothoid spline 20. By inserting the values, the estimate of the road course can be determined, which results from the adaptation 33 of the clothoid spline 20 to the environmental measurement data 50 of the camera 41.

[0086] For small curvatures, the clothoid segments 21 can be well approximated by the measurement model described in (2) and (3). For large curvatures, the accuracy of the approximate representation of the clothoid spline 20 according to (2) and (3) may be insufficient to ensure high accuracy for the adaptation 33 of the clothoid spline 20 to the ambient measurement data 50.

[0087] For a better approximation in the case of strong curvatures, a second measurement model can be applied to the clothoid spline 20 instead of the first measurement model. For example, the clothoid segment 21 can be represented by the parametric representation of the clothoid function.

[0088] This is in Fig. illustrated. In the parametric representation of, for example, the clothoid segment 21, each sample point 63 is represented in a vector representation. A vector of such a sample point 63 comprises, for example, two components, which can be expressed by Fresnel integrals. In order to reduce the numerical computational effort of the Fresnel integrals, a fifth-order Taylor expansion of the Fresnel integrals can be used instead of the Fresnel integrals. From this approximation of the clothoid segment 21, for example, a function 64 can result, where the function 64 corresponds to a Taylor polynomial 64 due to the fifth-order Taylor expansion. As can be seen in Fig. 6, this polynomial 64 describes the course of the clothoid segment 21 relatively accurately. The sample points 63 determined using the second measurement model describe the clothoid segment 21 and / or the clothoid spline 20 with good approximation, even in the case of strong curvatures. The sample points 63 generated by the second measurement model can represent the clothoid spline 20 more approximately than the sample points 61 generated by the first measurement model, particularly in the case of large curvatures. Therefore, the road course can usually be determined with greater accuracy using the second measurement model than with the first measurement model. In general, both measurement models can be used to estimate the road course.

[0089] Different sampling methods can be used to determine the sample points 61 and 63 of the respective measurement models and the sample points 52 of the environmental measurement data 50. In Fig. Figure 7 illustrates two possible sampling methods. In the first, conventional sampling method (left in Fig. 7) the sample points 61, 63 and 52 can be selected such that the distance between the sample points in the x-direction is constant. It has been shown that this sampling method is not optimal for a strongly curved road course when fitting 33 the clothoid spline 20, or the polynomials 62 and 64, with the Kalman filter. With this sampling method, due to systematic errors based on geometric conditions in the case of strong curvatures of the road course, the distances between the sample points 61 or 63 and the sample points 52 can be relatively large, even though the polynomial 62 or 64 approximately corresponds to the ambient measurement data 50. As a result, the Kalman filter would shift the clothoid spline 20 “too much” in the innovation. For example, the clothoid spline 20 would shift Fig. 7, the sample points 61 or 63 are shifted too far downwards, since the distance d of the sample points 61 or 63 to the sample points 52 is relatively large.

[0090] Such a systematic error can be reduced by using a second sampling procedure (right in Fig. 7) in the sense of further examples. In the diagram on the right, the Fig. 7, the sample points 61, 63 and 52 are selected such that a constant arc length Δs lies between neighboring sample points 61, 63 and 52 of the respective polynomials 62, 64 or the environmental measurement data 50. In this way, even with strong curvatures of the roadway, the sample points 61, 63 and 52 are located such that the distance d between the sample points 61 or 63 and the sample points 52 is, for example, smaller than in the left diagram of the Fig. 7. Accordingly, the clothoid spline 20 is shifted downwards less significantly by the Kalman filter, for example. The adjustment 33 of the clothoid spline 20 with the Kalman filter using the second sampling method can achieve greater accuracy than the adjustment 33 using the first sampling method. The second sampling method also has a stabilizing effect on a function of the Kalman filter and thus also on lateral guidance behavior when steering the vehicle, since the clothoid spline 20 is slightly corrected during adjustment 33 using the second sampling method. Thus, the state function may be adjusted less "strongly" in the individual recursion steps and thus be described as "more stable."

[0091] In some embodiments, the camera 41 can capture the course of lane markings 12 or 13, and the image processing application can be configured to detect such lane markings 12 or 13 and to approximately describe them using a polynomial or a polynomial spline. The lane markings 13 indicate lane boundaries of the roadway. The lane marking 12 indicates the center line of the roadway, which is not visible in some cases. A schematic illustration of a roadway marked by lane markings 12 and 13 is shown in Fig. 8. Based on the course of the lane markings 12 and / or 13, a roadway or lane width can be determined, for example, using the appropriate image processing application from images taken by camera 41. In some cases, the estimate of the lane course can correspond to the course of the lane markings 13. This can be advantageous for vehicle control. From the courses of two lane markings 13, which delimit the lane to the right and left, it is possible, for example, to determine the lane width or to derive the course of the center line 12 from it.

[0092] When controlling the vehicle, it may be necessary under certain circumstances to determine the course of the lane center. If the center line 12 is missing, the course of the lane center cannot be directly determined using data fusion 44 of the camera's ambient measurement data 50. In such a case, the course of the lane markings or lane boundaries 13 can be determined using data fusion 44, and based on this, the course of the lane center or the center line 12 can be derived, taking the lane width into account. The lane width of the lane to be determined can either be taken from a data set available to the vehicle or determined based on the course of the lane markings 13.To determine the course of the center line 12, the clothoid spline 20, which, for example, describes the course of the roadway center, can be transformed such that a transformed clothoid spline describes the course of one of the road markings 13, in order to enable adaptation 33 of the clothoid spline 20 to the ambient measurement data of the camera. After adaptation 33 of the clothoid spline 20, the clothoid spline can be determined by inverse transformation, which in turn indicates the course of the roadway center. For geometric reasons, it may not be sufficient to translate the clothoid spline 20, i.e., in the x- and y-directions, for the transformation and inverse transformation.

[0093] For this purpose, it may be necessary, for example, to use a mathematical procedure to also calculate the parameters c0, c 1,mto be adjusted so that the transformed clothoid spline corresponds to the course of the center line 12 and thus runs parallel to the roadway boundaries 13.

[0094] Alternatively, for the adaptation 33 of the clothoid spline 20, the camera's ambient measurement data can be transformed, taking into account a lane width, so that they approximately characterize the course of the lane center. The Kalman filter can then adapt 33 the clothoid spline 20 to the transformed measurement data. For the Fig. 4 schematically illustrated multi-sensor data fusion, ambient measurement data from another data source other than the camera is usually used. In the Fig. In the embodiment shown in Figure 4, the map data of the HD card 43 is used for this purpose. An example of such map data is shown in Fig. 9. The sensors of the map producer 42 provide waypoints 91 located on the roadway of the vehicle 11. The waypoints can be connected by interpolation, taking into account additional connection conditions. During interpolation, the waypoints can be connected, for example, by polynomials 92. Thus, a polynomial spline 90 can be generated from several polynomial functions 92. The polynomial spline 90, for example, can correspond to the map data.

[0095] In some embodiments of the invention described herein, the polynomial spline 90 can be represented as a function in spatial coordinates. The clothoid spline 20, which can be adapted to the polynomial spline 90 by the Kalman filter during data fusion 44, is also typically present here as a parameter set of the parameters y0, θ0, c0, and c1,1 ... c 1,n which can map the clothoid spline 20 using the curvature values ​​c(s) resulting from (1).

[0096] Compared to the ambient measurement data 50 from camera 41, the map data can be determined long before camera 41 can capture the roadway. For example, the course of roadway 14 can be estimated after sharply curved and / or obscured bends. For example, the map data can be used to accurately determine the course of the roadway well in advance of vehicle 11.

[0097] The orientation of the vehicle can sometimes not be determined, or not accurately, based on the map data. The orientation of the vehicle 11 relative to the roadway 14 typically cannot be determined from the position data of the vehicle 11 and the polynomial spline 90, since this cannot provide any indication of the orientation of the vehicle 11. Therefore, the Kalman filter typically does not use location coordinates from the map data to adapt the clothoid spline 20 to the map data. A further measurement model can be applied here, in which the map data is represented in a measurement space that includes a coordinate system with curvature values.

[0098] For the aforementioned data fusion 44 in the measurement space with the coordinate system with curvature values, it is advantageous, for example, to represent the clothoid spline 20 according to (1). Thus, the state space of the clothoid spline 20 can already correspond to the measurement space.

[0099] The polynomial spline 90 provided by the HD map 43 is typically present as a mapping in a coordinate system with spatial coordinates. In advantageous embodiments of the invention disclosed here, a transformation 32 of the map data into the measurement space with the coordinate system with curvature values ​​may therefore be necessary. A suitable mapping of the map data can, for example, be a function of the arc length s of the map data, with a curvature value k of the map data resulting depending on the arc length s.

[0100] For the adaptation 33 of the clothoid spline 20 to the polynomial spline 90, for example, the innovation of the Kalman filter approximates the curvature values ​​of the clothoid spline 20 to the curvature values ​​k of the polynomial spline 90. For example, the curvature values ​​k at the waypoints of the map data are taken into account. The adaptation 33 can be achieved by adjusting the parameters y0, θ0, c0, and c 1,1 ...c1,n take place.

[0101] An advantage of the described data fusion 44 based on the curvature values ​​k is that this data fusion 44 is robust against angle and offset errors. Assuming that a localization determines a lane in which the vehicle 11 is actually located, the road course can be accurately estimated based on the curvature values ​​k, for example, even if the vehicle 11 is not exactly in the center of the lane or is not aligned parallel to the center line 12.

[0102] In advantageous embodiments of the invention disclosed here, the computer-implemented method may include multi-sensor data fusion 40. Advantages may arise from the multi-sensor data fusion 40 of environmental measurement data 50 and map data.

[0103] During the data fusion 44 of the environmental measurement data, for example, the position and orientation of the vehicle 11 relative to the roadway 14 can be determined. When controlling the vehicle 11, for example, a departure of the vehicle 11 from the roadway 14 can be prevented.

[0104] With the help of the data fusion 44 of the map data, the course of the road can be determined well in advance, so that the control of the vehicle 11 can, for example, reduce the speed of the vehicle early enough to safely negotiate a sharply curved curve that cannot be seen.

[0105] Furthermore, the data fusion 44 provides robustness against angle and offset errors in the estimation of the road course, thereby reducing measurement errors in the estimation. The computer-implemented method 30 can be used for the control of any ground-based vehicle 11. An example of such a vehicle 11 is shown in Fig.10. The vehicle 11 can be equipped with a device 100, which serves to estimate the course of the road. The device 100 comprises a processing unit (not shown here) and several interfaces for acquiring environmental measurement data. Such an interface can, for example, be a connection of the processing unit to one or more cameras 41. The cameras 41 can, for example, be directed in the direction of travel, against a direction of travel, or to the side and enable provision 31 of the environmental measurement data 50. Further interfaces can connect the processing unit to a storage medium 103, such as a hard disk drive or an optical storage device, and additionally or alternatively to a receiving unit 102.The provision 31 of map data 90 can be enabled by the map data 90 being present on the storage medium 103 or by the map data being received by the receiving unit 102 and provided to the processing unit. For example, the map data can be transmitted from a transmitter 110 to the receiving unit. Using the environmental measurement data 50 and the map data, the device 100 can determine an estimate of the road course by executing the computer-implemented method 10.

[0106] For this purpose, the processing unit is designed to perform the transformation 32 of the clothoid spline 20 and the environmental measurement data 50 and / or the map data. Furthermore, the processing unit is configured to adapt 33 the clothoid spline 20 to the environmental measurement data. The processing unit can be, for example, a processor, a microcontroller, a field-programmable gate array (FPGA), a computer, or a programmable hardware component.

[0107] The aspects and features described together with one or more of the previously detailed examples and figures may also be combined with one or more of the other examples to replace a like feature of the other example or to additionally introduce the feature into the other example.

[0108] Examples may further include, or relate to, a computer program having program code for performing one or more of the above methods when the computer program is executed on a computer or processor. Steps, operations, or processes of various methods described above may be performed by programmed computers or processors. Examples may also cover program storage devices, e.g., digital data storage media, that are machine-, processor-, or computer-readable and encode machine-executable, processor-executable, or computer-executable programs of instructions. The instructions perform or cause some or all of the steps of the above-described methods to be performed. The program storage devices may, for example,Digital storage devices may include or be magnetic storage media such as magnetic disks and magnetic tapes, hard disk drives, or optically readable digital data storage media. Further examples may also include computers, processors, or control units programmed to perform the steps of the methods described above, or (field) programmable logic arrays ((F)PLAs) or (field) programmable gate arrays ((F)PGAs) programmed to perform the steps of the methods described above.

[0109] The description and drawings only illustrate the principles of the disclosure. Furthermore, all examples provided herein are expressly intended to serve only illustrative purposes to assist the reader in understanding the principles of the disclosure and the concepts contributed by the inventor(s) to advance the art. All statements herein regarding principles, aspects, and examples of the disclosure, as well as specific examples thereof, include their equivalents.

[0110] A functional block referred to as "means for..." performing a particular function may refer to a circuit configured to perform a particular function. Thus, a "means for something" may be implemented as a "means configured for or suitable for something," e.g., a component or circuit configured for or suitable for the respective task. Functions of various elements shown in the figures, including any functional blocks referred to as "means," "means for providing a signal," "means for generating a signal," etc., may be implemented in the form of dedicated hardware, e.g., "a signal provider," "a signal processing unit," "a processor," "a controller," etc., as well as hardware capable of executing software in conjunction with associated software.When provided by a processor, the functions may be provided by a single dedicated processor, by a single shared processor, or by a plurality of individual processors, some or all of which may be shared.

[0111] However, the term "processor" or "controller" is far from limited to hardware capable of executing software only, but may include digital signal processor (DSP) hardware, network processor, application-specific integrated circuit (ASIC), field-programmable gate array (FPGA), read-only memory (ROM) for storing software, random access memory (RAM), and non-volatile storage devices. Other hardware, conventional and / or custom, may also be included.

[0112] For example, a block diagram may represent a high-level circuit diagram implementing the principles of the disclosure. Similarly, a flowchart, a sequence diagram, a state transition diagram, pseudocode, and the like may represent various processes, operations, or steps, for example, substantially embodied in computer-readable medium and thus performed by a computer or processor, regardless of whether such a computer or processor is explicitly shown. Methods disclosed in the specification or claims may be implemented by a device having a means for performing each of the respective steps of these methods.

[0113] It should be understood that the disclosure of multiple steps, processes, operations, or functions disclosed in the description or claims should not be construed as being in that particular order unless explicitly or implicitly stated otherwise, e.g., for technical reasons. Therefore, the disclosure of multiple steps or functions does not limit them to a particular order unless those steps or functions are not interchangeable for technical reasons. Furthermore, in some examples, a single step, function, process, or operation may include and / or be broken down into multiple sub-steps, functions, processes, or operations. Such sub-steps may be included and be part of the disclosure of that single step unless explicitly excluded.

[0114] Furthermore, the following claims are hereby incorporated into the Detailed Description, where each claim may stand on its own as a separate example. While each claim may stand on its own as a separate example, it should be noted that although a dependent claim may refer to a particular combination with one or more other claims in the claims, other examples may include a combination of the dependent claim with the subject matter of any other dependent or independent claim. Such combinations are explicitly suggested herein unless it is stated that a particular combination is not intended. Furthermore, features of a claim for any other independent claim are also intended to be included, even if that claim is not made directly dependent on the independent claim. List of reference symbols 11 vehicles 12 Center line 13 Road markings 14 Roadway 20 Clothoid splines 21 Clothoid segment 22 Transition point 30 Computer-implemented procedure 31 Providing environmental measurement data 32 Transforming the state function and the environmental measurement data 33 Adjusting the state function 33a Check the environmental measurement data for an error 33b Saving the error 33c Benefit of the error 40 Multi-sensor data fusion 41 Camera 42 Sensors of the card manufacturer 43 HD card 44 Data fusion 45 Road model 50 environmental measurement data of the camera 51 sample points of the clothoid spline 52 sample points of the camera's environmental measurement data 60-1 Illustration of the first sampling procedure 60-2 Illustration of the second sampling procedure 61 sample point of the polynomial 62 polynomial 63 sample point of the Taylor polynomial 64 Taylor polynomial 90 polynomial spline 91 Waypoint 92 polynomial function 100 device 102 Receiving unit 103 Storage medium 110 channels

Claims

[1] Computer-implemented method (30) for estimating a road course in an environment of a vehicle (11) based on a state function describing the road course, wherein the state function comprises a clothoid spline (20), the method (30) comprising: Providing (31) environmental measurement data describing the course of the road at a current position of the vehicle (11), wherein the environmental measurement data comprise at least one polynomial function; Transforming (32) the state function and the environmental measurement data into a common coordinate system; and Checking (33a) the environmental measurement data for an error; if no error was detected, adapting (33) the state function based on the environmental measurement data in the common coordinate system, wherein the adapting (33) of the state function comprises applying a measurement model to the clothoid spline (20), in which a clothoid segment (21) is represented by a parameter representation of the clothoid function, wherein in the parameter representation each sample point (63) of the clothoid segment (21) is represented in a vector representation, if an error is detected, storing (33b) the error. [2] Computer-implemented method (30) according to claim 1, wherein the checking comprises determining a deviation between a value determined in a prediction step of a Kalman filter and the environmental data and then comparing this deviation with a predetermined threshold value. [3] Computer-implemented method (30) according to claim 1 or 2, wherein the stored error is sent to a receiving unit arranged externally of the vehicle (11). [4] Computer-implemented method (30) according to claim 1, 2 or 3, wherein the stored error is used to improve a lane marking and / or to improve map data and / or to improve a lane marking recognition function and / or to improve an autonomous driving function (33c). [5] Computer-implemented method (30) according to one of the preceding claims, wherein the environmental measurement data (50) are acquired by at least one camera (41). [6] Computer-implemented method (30) according to one of the preceding claims, wherein the environmental measurement data is provided by a map (43). [7] Computer-implemented method (30) according to one of the preceding claims, wherein first environmental measurement data (50) are acquired by at least one camera (41) and second environmental measurement data are provided by a map (43). [8] Computer-implemented method (30) according to one of the preceding claims, wherein the common coordinate system comprises location coordinates. [9] Computer-implemented method (30) according to one of the preceding claims, wherein adapting (33) the state function comprises a sampling method for determining sample points (51), (52), (61) and (63), wherein for each plurality of sample points there is a constant curve length between adjacent sample points. [10] A computer-implemented method (30) according to any one of the preceding claims, wherein the common coordinate system comprises curvature values. [11] Computer-implemented method (30) according to one of the preceding claims, wherein the adaptation (33) of the state function to the environmental measurement data provides for adaptation of at least one curvature value of the state function to at least one further curvature value of the environmental measurement data. [12] Computer-implemented method (30) according to one of the preceding claims, wherein the state function characterizes a course of a road marking (12), (13) of a roadway. [13] Computer-implemented method (30) according to claim 12, further comprising determining a course of a lane center (12) by transforming the state function. [14] Computer-implemented method (30) according to one of the preceding claims, wherein the computer-implemented method (30) further comprises providing validation measurement data and validating the estimation of the road course using the validation data. [15] A device (100) for estimating the course of the road in the surroundings of the vehicle (11), comprising: one or more interfaces configured to acquire environmental measurement data; a processing unit configured to execute the computer-implemented method according to claims 1 to 14. [16] A vehicle (11) comprising the device (100) according to claim 15.

Images