Method and assistance system for an information filter-based driving tube estimate for a motor vehicle

Abstract

The invention relates to a method and to an assistance system (7) for the repeated estimation of a current driving tube (14) of a motor vehicle (5), wherein the method is able to be performed repeatedly at regular intervals during operation of the motor vehicle (5), and the invention furthermore relates to a correspondingly configured motor vehicle (5). In the method, multiple measured data (11), which characterize a respective traffic situation of the motor vehicle (5), are recorded with associated uncertainties. An updated information matrix and an updated information vector of the information filter-based formalism are thus each estimated in an update step. An updated estimate of the driving tube (14) is then determined from the updated information matrix and the updated information vector.

Description

[0001] Method and assistance system for an information filter-based driving path estimation of a motor vehicle

[0002] The present invention relates to a method and an assistance system for estimating a driving path of a motor vehicle during its operation. The invention further relates to a correspondingly configured motor vehicle.

[0003] Today, increasing automation of vehicle guidance and further improvement of assistance functions and systems in motor vehicles are being pursued. A useful basis for this can be a precise estimate, i.e., prediction, of the driving path of the respective motor vehicle. Such a driving path, which can also be referred to as a driving corridor, indicates an area ahead of the motor vehicle in the direction of travel, through or along which the motor vehicle is at least likely to move. Such a driving path can be delimited, for example, by roadsides or the like and / or by defined edges.

[0004] A method for detecting road edges on a roadway to determine a clear path is described, for example, in DE 1020132005 950 B4. In this method, line candidates are detected based on input images, and an initial vanishing point in an image is identified as a function of the line candidates. For each line candidate, a region of interest is identified in the image, each region of interest comprising a line candidate and a surrounding region.

[0005] Features related to the line candidate extracted from the region of interest are input to a classifier to identify whether the line candidate is a potential road edge. The classifier assigns a confidence value to the line candidate, which identifies a probability regarding whether the line candidate is a road edge. The potential line candidate is identified as a reliable road edge based on the confidence value being greater than a predetermined value.

[0006] As described, for example, in DE 102006 040 334 A1, the current speed of the respective vehicle and its yaw rate can be used to estimate the driving path. However, corresponding known methods are not always sufficient for more complex driver assistance systems. Therefore, a method for lane detection is proposed there. In this method, a sensor system of the vehicle detects lane markings in an area in front of the vehicle. Support points with coordinates from a first coordinate system are assigned to the lane markings, which are then converted into a second coordinate system. The course of the lane markings and / or lanes is reconstructed from the position of the support points in the second coordinate system.

[0007] DE 102016 003 935 A1 discloses a method for determining edge development information describing edge development along a route traveled by a motor vehicle. Trajectory data describing the past trajectory of the motor vehicle and / or a predicted trajectory of the motor vehicle are used as input data. In the method, a determination area covering the described trajectory is divided into interval areas perpendicular to the vehicle's longitudinal direction. For each interval area, obstacle positions within the interval area are sorted as belonging to a left or right edge development depending on the position of the trajectory. The edge development information is then determined from the obstacle positions closest to the trajectory for each interval area.This should enable the determination of boundary development information from sensor data provided via a generic data interface, independent of a model and the specific sensor types used. A problem with previous approaches to estimating a driving path is that these approaches are often comparatively computationally intensive, while in motor vehicles, only limited hardware and computing resources are available, for example, control units designed as embedded systems or similar. Therefore, there is a need not only for improved environment detection and driving path estimation, but also for a reduction in the computing power required for this.

[0008] The object of the present invention is to enable a particularly efficient estimation of a driving path of a motor vehicle during its operation.

[0009] This object is achieved according to the invention by the subject matter of the independent patent claims. Possible embodiments of the present invention are disclosed in the dependent patent claims, in the description, and in the figure.

[0010] The method according to the invention can be used to repeatedly estimate a current driving path of a motor vehicle that lies ahead of the motor vehicle in the direction of travel. The method can be applied while the respective motor vehicle is in operation. By means of the method according to the invention, the respective current driving path can be estimated continuously or regularly while driving, or the estimated driving path can be updated continuously or regularly. In one method step of the method according to the invention, measurement data that indicate or characterize a respective traffic situation of the motor vehicle are recorded with the associated uncertainties. The measurement data can, for example, be specified or organized in a vector. The uncertainties can, for example, be specified or organized in an uncertainty matrix, which can, in particular, contain only the uncertainties of the measurement data.The uncertainties can, for example, indicate measurement errors, confidence intervals, or correspondingly limited accuracies of the measured data. The uncertainties can be provided and / or calculated by a sensor together with the respective measurement data or measured values, for example, using a predefined heuristic or the like.

[0011] The measurement data and their associated uncertainties can together provide a

[0012] They form or span a measurement space, or they can be specified within a measurement space. The measurement space is therefore an abstract mathematical space. If multiple measurement data, for example of different types and / or for different positions, are used to estimate or update the travel path, the measurement space can be correspondingly high-dimensional. In practice, the uncertainty matrix can, for example, have on the order of 100 rows and columns or more.

[0013] The measurement data can describe or depict the respective surroundings of the motor vehicle, in particular those ahead in the direction of travel of the motor vehicle. Likewise, the measurement data can be or include, for example, operating or status data of the motor vehicle, movement data of other road users, which can, for example, indicate trajectories driven by other vehicles at earlier times in the respective surroundings, and / or more. In particular, the individual measurement data can be of several different types, originate from several different sensors, originate from different measuring points, i.e. measuring locations in the surroundings, and / or more. This can enable a particularly robust and accurate estimation or updating of the driving path. However, with previous approaches, this can also involve a correspondingly large data processing or computing effort.However, this problem is circumvented or avoided by the method according to the invention.

[0014] For this purpose, in a further method step of the method according to the invention, for example, as soon as new measurement data is available or has been acquired, an updated information matrix and an updated information vector are estimated in an update step within the framework of the information filter formalism or mechanism. In other words, the information filter or an information filter is used here to consider the new or updated measurement data, including their uncertainties, in the estimation or update of the travel path. The information filter is a type of Kalman filter, which is also technically referred to as an inverse covariance filter.

[0015] In a further process step, for example, in each run of the process or after each update step, an updated estimate of the travel path is determined from the updated information matrix and the updated information vector. In other words, a state or state vector can be determined that describes the travel path according to the current estimate or prediction.

[0016] The application of the information filter formalism proposed here enables the estimate of the travel path to be updated without inverting the uncertainty matrix, as is typically necessary with previous approaches. Instead, individual measurements can be considered or processed independently of one another in the update step. For this purpose, correspondingly smaller matrices can be inverted for each individual measurement, for example a single sensor or a single measuring point or position or the like. The information matrix, which may need to be inverted, can be smaller, in particular much smaller, i.e. have a smaller dimension than the uncertainty matrix for the respective entirety of the recorded or newly recorded measurement data. Thus, the present invention enables a travel path estimate orDriving path update, which in its results is equivalent to previous approaches, in which updating the driving path estimate by taking into account the new measurement data in each case requires the inversion of the comparatively high-dimensional uncertainty matrix, which represents the uncertainties of all new measurement data or measurements. In comparison to such previous approaches, the method according to the invention can be carried out with much less computational effort. This can enable a driving path estimation during operation of a motor vehicle that is equivalent to previous methods or even improved, for example by using more measurement data and / or more frequent or high-frequency updates, particularly when using limited hardware or computing resources, such as control units or embedded systems commonly used today.

[0017] The driving path or its estimation can be described by a small number of parameters or values ​​compared to the quantity or number of measured data and their uncertainties, so that even calculation steps or process steps that are carried out in a corresponding state space spanned by the parameters used to describe the driving path or its state can be carried out with particularly low computational effort. Since a correspondingly accurate and reliable driving path estimation can form the basis for a wide variety of different assistance systems and / or at least partially automated driving of the vehicle, effort and energy can be saved without disadvantage and / or improved safety and / or comfort can be achieved particularly easily.For example, even with the computing hardware commonly found in today's motor vehicles, an update of the driving path estimate can be performed in less than 1 ms on average. This allows for a correspondingly rapid response to new measurement data or a change in the vehicle's traffic situation.

[0018] The traffic situation of the motor vehicle within the meaning of the present invention can be given by an environmental situation or the surroundings of the motor vehicle. The traffic situation can also include the current state of the motor vehicle, for example its position, speed, steering angle and / or the like. The surroundings can be given or described, for example, by the respective local traffic infrastructure, the presence and, if applicable, positions of environmental features, objects, obstacles, other road users or the like, as well as their sensory detectability. Corresponding features or objects in the environment can be or include, for example, road or lane markings, curbs, a road or roadway course, elements of road equipment, road or roadway edges, guard rails, strips of vegetation and / or the like.

[0019] The problem addressed and the idea proposed here for its solution can be summarized as follows, which will be explained in more detail below: Matrix inversion for large matrices is fundamentally very complex. This applies to the measurement uncertainty matrix, which in previous methods must be inverted for the update, which is much larger than the state uncertainty matrix. The estimation problem can be formulated on the basis of an information matrix and an information vector. The information filter can thus be applied at least as part of the solution. This allows an independent consideration of all individual measurements. This means that instead of inverting the comparatively large entire measurement uncertainty matrix for the update, only an inversion of several comparatively much smaller measurement uncertainty matrices can be performed.The final inversion of the likewise comparatively small information matrix for transferring the updated estimate into the state space is associated with significantly less effort than inverting the entire measurement uncertainty matrix. In one possible embodiment of the present invention, the measurement data describe at least the course of the road ahead of the motor vehicle in its direction of travel at different distances from the motor vehicle, i.e. from its current position. The measurement data can therefore characterize, map or scan the respective environment or road, for example, with a resolution in the range of centimeters or decimeters or on the order of 1 m, for example up to a distance of several tens of meters.The road ahead can be recognized with particular precision, which ultimately enables a particularly accurate, robust, and reliable estimation of the driving path. The use of such detailed measurement data as a basis for estimating the driving path is made possible by particularly efficient data processing, i.e., particularly efficient driving path estimation, without significant disadvantages compared to previous approaches, which, given the given computing capacity or hardware, may only be able to process less or less detailed measurement data in real time.

[0020] In a further possible embodiment of the present invention, the measurement data are organized in a measurement data vector, and the associated uncertainties are organized in an uncertainty matrix. In the updating step, the individual measurement data and their associated uncertainties are considered independently of one another without inverting the entire uncertainty matrix. For example, individual submatrices of the overall uncertainty matrix corresponding to the individual measurement data can be individually inverted, i.e., individually, instead of inverting the entire uncertainty matrix. These submatrices can, for example, have a dimension of 3 x 3 or 4 x 4, while the dimension of the overall uncertainty matrix can be on the order of 100 x 100 or more. In particular, the dimension of the submatrices can be an integer divisor of the dimension of the overall uncertainty matrix.The advantage of the information filter formalism can be exploited here: measurements or individual measurement data can be filtered in each run or period of the method by summing the corresponding information matrices and information vectors. Individual measurement data, in the present sense, can be individual measurement points or data points from one or more data sources. A single sensor can provide a single measurement datum or multiple individual measurement data points for each new measurement. The embodiment of the present invention proposed here offers an easy-to-implement approach for estimating the travel path without loss of accuracy, while avoiding the computational or data processing effort required for inverting the entire uncertainty matrix.

[0021] In a further possible embodiment of the present invention, a corresponding updated state or state vector for the travel path is first determined from the updated information matrix and the updated information vector. This state or state vector characterizes or describes the travel path using predetermined, in particular geometric, parameters. The number of these predetermined parameters is smaller, in particular much smaller, than the dimension of the measurement data or the measurement space in which the measurement data and their uncertainties are specified. When the measurement data is organized in vector or matrix form, there may be dozens or hundreds of rows or columns, for example. The number of predetermined parameters for characterizing, describing, or specifying the travel path orThe number of parameters for the state or state vector for the travel path, however, can be a maximum of 20 or a maximum of 10, for example, 4. The number of parameters can therefore be a maximum of one-fifth or a maximum of one-tenth of the number of rows of the measurement data vector or the associated uncertainty matrix. In this way, the travel path, i.e., its current estimate, can be specified or handled with particularly low data processing effort.

[0022] In a possible further development of the present invention, the predetermined parameters for characterizing the driving path comprise its curvature, in particular at several predetermined distances from the motor vehicle or from its respective current position, i.e. at several points spatially spaced from one another in the direction of travel of the motor vehicle or in the longitudinal direction or along the longitudinal extent of the driving path. This allows the driving path or its course to be described or specified particularly simply, efficiently, and with minimal data. Such a characterization of the driving path can be sufficient for practical purposes, since certain properties of the driving path or corresponding boundary conditions for the course of the driving path must always be present or fulfilled, i.e., they can be predetermined accordingly.This may, for example, include the driving path always running on the local road surface, being able to have a predetermined width, having a maximum curvature that is predetermined or determined, for example, by the technical characteristics of the motor vehicle, being continuous, i.e., uninterrupted, not having any jumps, steps, or transverse offsets, and / or the like. Likewise, the predetermined parameters for characterizing the driving path may include, for example, its radius, length, width, orientation, and / or the like. The respective values ​​of these parameters may, for example, be specified or entered in the state vector mentioned elsewhere. This may enable particularly simple and efficient data processing.

[0023] In a possible further development of the present invention, the most recently determined updated information matrix is ​​inverted to determine the updated state of the driving path. This inverted information matrix is ​​then multiplied by the most recently determined updated information vector. This represents a particularly simple and low-effort option for describing the driving path or its state or estimate using real, clearly understandable, for example, geometric quantities, i.e., specifying it in a correspondingly clear and practical state space. This can enable particularly simple and efficient reuse of the current driving path estimate by other assistance systems and assistance functions. For example, the latter can then be implemented independently of the formalism used for the driving path estimation.

[0024] In a possible further development of the present invention, a corresponding updated estimated path of the driving path in the respective environment is determined from the updated state of the driving path using a predefined model. The predefined model models or considers at least one predefined condition for a permissible real driving path path. The model can therefore, for example, model the properties or boundary conditions for the driving path path mentioned elsewhere.From the combination of the state, which can be specified, for example, by basic or point-based parameter values, and the at least one predefined condition, the model can determine the complete driving path course, for example as a two- or three-dimensional, elongated geometric object that can be located, for example, relative to the current position of the motor vehicle and / or in a predefined, for example, world-fixed, coordinate system. The model can therefore specify or model how an actual course of the driving path results from the respective measurement data or the state or state vector of the driving path estimated therefrom. The model can, for example, be stored or implemented in an estimation device that is set up to estimate the state of the driving path.The model can construct the real route path based on the current state, for example, from circular segments or curved lines or similar. A condition can be specified, for example, that transitions between consecutive circular segments or curved lines must be seamless, meaning that these circular segments or curved lines must connect or merge into one another without any jumps, offsets, or gaps. This can enable particularly efficient modeling of the real route path.

[0025] In a further possible embodiment of the present invention, environmental data is recorded as part of the measurement data. This environmental data describes the respective surroundings of the motor vehicle, in particular those ahead in the direction of travel of the motor vehicle. Furthermore, steering angle data of the motor vehicle is recorded as part of the measurement data. This steering angle data can, for example, indicate or include a current steering angle or a current change in the steering angle, i.e., for example, its direction of change and / or speed of change. It is then further determined whether a lane change maneuver or a turning maneuver of the motor vehicle is likely to occur in a range from the respective current position of the motor vehicle up to a predetermined distance from the motor vehicle in its direction of travel.This can be done, for example, on the basis of navigation data, a driving history of the motor vehicle and / or the respective driver, a corresponding predefined movement model and / or the like.

[0026] If it is determined that such a maneuver is likely imminent, the updated estimate of the driving path is determined up to the specified distance based on the steering angle data, and from the specified distance onwards based on the environmental data without taking the steering angle data into account. Up to the specified distance, the driving path can therefore be estimated, for example, exclusively based on the steering angle data or based on the steering angle data and the environmental data. This can enable a particularly accurate and reliable estimate of the driving path up to the specified distance, since the steering angle primarily determines and limits the direction of movement of the vehicle in the corresponding close range.

[0027] At the same time, because the steering angle data are disregarded from the specified distance onwards, the subsequent, more distant driving path can also be estimated particularly reliably, since unforeseen steering angle changes or steering maneuvers can occur between the current position of the vehicle and the specified distance, so that the current steering angle data at the current position of the vehicle from the specified distance onwards do not form a reliable basis for estimating the driving path.

[0028] Thus, the embodiment of the present invention proposed here enables particularly accurate and reliable estimation of the travel path with particularly low computational or data processing effort both in the near range up to the specified distance and in the far range beyond that.

[0029] The specified distance can, for example, depend on the respective motor vehicle or its technical characteristics and / or the respective environment and / or the current speed of the motor vehicle and / or similar factors. The specified distance can therefore be dynamically specified, meaning it can be automatically adjusted accordingly during operation of the motor vehicle. The specified distance can also be specified as a fixed value. The specified distance can, for example, be between 10 m and 30 m, for example, 15 m. Other values ​​may also be possible depending on the application.

[0030] For example, a larger distance can be specified or set for a smaller maximum possible steering angle and / or a smaller maximum possible steering angle change speed of the vehicle. Likewise, a larger distance can be specified or set for a higher vehicle speed. Likewise, for example, a larger distance can be specified or set on a motorway or a motorway-like road than in urban areas. This enables a particularly situation-adapted, optimized driving path estimation for different situations or environments.

[0031] The present invention further relates to an assistance system for a motor vehicle. The assistance system according to the invention has an interface - implemented in hardware and / or software - for recording measurement data that characterizes a traffic situation of the motor vehicle. Furthermore, the assistance system according to the invention has a processing device, for example a microchip, microprocessor or microcontroller or the like, and a computer-readable data memory coupled thereto. The assistance system according to the invention is designed to carry out the method according to the invention, in particular automatically. The assistance system according to the invention can therefore be used for a regular, continuous or quasi-continuous estimation or prediction of the driving path of the motor vehicle orAn update of the estimated or predicted driving path of the motor vehicle can be configured. For this purpose, a corresponding operating or computer program can be stored in the data memory, which encodes or implements the method steps, measures, or sequences mentioned in connection with the method according to the invention, or corresponding control instructions. This operating or computer program can then be executable by means of the processing device in order to execute the corresponding method or to initiate its execution.

[0032] The invention further relates to a motor vehicle having an environmental sensor system for recording environmental data and an assistance system according to the invention. The environmental data can be data that describe or characterize a respective environment of the motor vehicle, in particular that ahead of the motor vehicle in the direction of travel. The environmental sensor system can be part of the assistance system or connected to it, for example via an on-board electrical system of the motor vehicle. The motor vehicle according to the invention can in particular be the motor vehicle mentioned in connection with the method according to the invention and / or in connection with the assistance system according to the invention or correspond to it. Further features of the invention can arise from the claims, the figures and the description of the figures.The features and combinations of features mentioned above in the description as well as the features and combinations of features shown below in the description of the figures and / or in the figures alone can be used not only in the respective combination specified, but also in other combinations or on their own, without departing from the scope of the invention.

[0033] The single figure in the drawing shows an exemplary schematic overview of a traffic situation to illustrate an information filter-based traffic flow estimation for a vehicle. For the sake of clarity, only a representative selection of elements that occur repeatedly is explicitly marked in the figure.

[0034] The identification of objects relevant to the longitudinal and lateral control of a vehicle may require the detection of the vehicle's potential driving path, to which all detected objects can then be related. To ensure high availability of the driving path or a current estimate of the driving path in a wide variety of traffic situations and road conditions, a predefined fusion algorithm can be used to merge various measurement or input data from different data sources, such as smart, heterogeneous sensors or the like. This allows different aspects of the respective environment, such as data on lane markings, verge structures, turf, curbs, and / or the like, to be captured and combined to create a coherent overall picture of the driving path.Objects detected in the respective environment, particularly moving ones, can then be assigned to the driving path. Moving objects in the vehicle's surroundings, especially those assigned to the driving path, can be used or taken into account, for example, for longitudinal control functions such as ACC, intersection assist, adaptive recuperation, and / or the like, as well as for lateral control functions such as steering and lane guidance assistants.

[0035] Previous approaches are often based on the assumption that the vehicle's path, unless already known from a planning component of an automated vehicle guidance system, corresponds to the course of the road or lane currently being traveled where the vehicle is located and must be determined using sensors that monitor the respective environment. This path can be determined, for example, in a recursive estimation process that compares a current idea or estimate of the course of the road or lane with lane markings detected in sensor data, such as camera images.

[0036] The problem to be solved is to determine or estimate the upcoming lane or its course. To date, corresponding estimation processes have typically worked in the corresponding state space of the solutions, where the uncertain state is updated with each uncertain measurement, i.e. with each recording or capture of uncertain measurement data or, for example, with each uncertain detection of lane markings or the road or lane course or the like. This state can be specified, for example, by parameters or parameter values ​​of a parametric representation of the geometry of the road or the lane being traveled on. With such an update, the estimated state can be better aligned with the respective measurement or adapted to the measurement, while at the same time reducing the uncertainty of the state estimate.In previous approaches, the uncertainties of the measurements or the measured data are typically represented using high-dimensional matrices of the covariances of the measured quantities.

[0037] A disadvantage of the previous approach becomes all the more apparent the more different measurements or measurement data, for example measurement data from different sensors or finely resolved measurement data of the road course at different distances from the vehicle, have to be taken into account in the estimation process. Updating the estimate by taking into account the new measurements or measurement data can then require the inversion of a correspondingly high-dimensional matrix in previous approaches, which represents the uncertainties of the measurements. Such a matrix can, for example, have a size of 500 x 500. Inverting or inverting such matrices can require considerable computational effort, while at the same time, embedded systems, such as those often underlying today's driver assistance systems, can only offer very limited computing time or computing power. Against this background, Fig.Figure 1 shows an exemplary schematic and partial overview of a traffic situation. Specifically, a section of a road 1 is shown, which is bordered by road edges 2. On road 1, there are, for example, two lanes separated from each other by lane markings 3. Furthermore, a guardrail 4 runs along one of the road edges 2.

[0038] In this case, a motor vehicle 5 is moving along road 1. This motor vehicle 5 can be equipped with assistance functions that use a current driving path of the motor vehicle 5 as a data basis. To meet this requirement, the motor vehicle 5 is configured for a continuous or regular estimation of the driving path or a corresponding update of the estimation of the driving path. For this purpose, the motor vehicle 5 has an environmental sensor system 6 and an assistance system 7. The assistance system 7 can capture environmental data recorded by the environmental sensor system 6 as well as other data, for example steering angle data of the motor vehicle 5, via an interface 8. To process this data to estimate the driving path, the assistance system 7 can have a processor 9 and a computer-readable data memory 10 - as indicated schematically here as an example.

[0039] The recorded data can be of different types, originate from different sources, and characterize or describe the surroundings of the motor vehicle 5 at different locations. By way of example, several corresponding data or measurement points 11 are represented here, of which, for the sake of clarity, only a selection is explicitly marked. The data or measurement points 11 can, for example, originate from a third-party vehicle 12 also traveling on road 1, describe its trajectory 13, comprise one or more points on a path 17 of the motor vehicle 5 derived from the current steering angle or a current position of steered wheels 16 of the motor vehicle 5, i.e., projected into the future based on the steering angle, be or represent measurements or detections of the road edge 2, the lane markings 3, the guardrail 4, and / or the like.Based on this, the assistance system 7 continuously and repeatedly determines a current driving path estimate 14 during operation of the motor vehicle 5, which is also illustrated schematically here. The respective driving path estimate 14 can be described by comparatively few parameters compared to the number or dimension of the data recorded as the basis for the driving path estimate 14. In the specific case, a state to be estimated for the driving path can, for example, consist of four parameters, each of which indicates the curvature of the driving path at different distances, which are represented here as exemplary support points 15.The assistance system 7 can then describe, i.e., determine, the complete course of the driving path, i.e., the complete driving path estimate 14 for the motor vehicle 5, using a predefined model, stored, for example, in the data storage 10, which uses this respective estimated state as input. The respective current driving path estimate 14 can then be stored, for example, in the data storage 10 and / or provided or transmitted via the interface 8.

[0040] Based on the observation that the dimension of the corresponding state space spanned by the parameters for describing the state or course of the travel path is smaller, in particular much smaller, than the dimension of the measurement space in which the recorded data or their uncertainties are specified or represented, a determination or updating of the travel path estimate 14 can be carried out which is simpler, less computationally intensive and thus more efficient than previous approaches. For this purpose, the assistance system 7 estimates the information vector and the associated information matrix based on the respectively recorded data within the framework of the information filter formalism, instead of directly estimating the state of the travel path and the associated uncertainty. The application of the information filter is therefore provided here, with the respective updating of the travel path estimate 14 taking place without a complex inversion orInversion of the entire uncertainty matrix of the acquired data is possible, i.e., can be performed. Instead, the individual acquired data or measurements or measurement data, for example, of the individual measurement points 11, are considered independently of one another in the respective update step. In this case, the inversion of the entire uncertainty matrix can effectively be replaced by easier-to-calculate inversions or inversions of correspondingly smaller matrices. The estimate of the respective state can then be obtained from the information vector after inversion or inversion of the comparatively low-dimensional information matrix.

[0041] For a first starting point, i.e. a first driving path estimate 14 during or after commissioning of the motor vehicle 5, for example, a simplified driving path estimate 14 can be determined based on the current steering angle of the motor vehicle 5, which can then be continuously updated as described here, in particular taking into account environmental data or an estimate based thereon of the course of the road 1 ahead of the motor vehicle 5 in the direction of travel. Thus, starting from a driving path originally based on the current steering angle of the motor vehicle 5, the estimate of the parameters indicating the state of the driving path, i.e. for example the aforementioned curvatures at the multiple support points 15, can be adapted in a continuous process to the most recently recorded data or the most recent measurement or estimate of the actual road course of the road 1.

[0042] For this purpose, a multitude of different measurements from different sensors can be available at any time, all of which can be combined into a uniform format for data fusion. Such measurements can be, include, or indicate, for example, radio waves reflected from structural elements or facilities along the road or carriageway edge 2, such as the guardrail 4, back to a radar sensor of the environmental sensor system 6; markings, such as the lane markings 3, and / or strips of vegetation or turf or the like, which are detected in camera images from a camera of the environmental sensor system 6; the trajectory 13 of the other vehicle 12 traveling ahead; a collective movement of other road users in the respective environment; the steering angle or change in steering angle of the motor vehicle 5, its position, orientation, direction of movement, and speed; and / or the like.

[0043] This means that, for example, several hundred uncertain measurements, for example in the form of a specific angle specification of the course of road 1 at a specific distance from several different distances, can be available for fusion in each update step. By transferring the problem to the information space, i.e., expressing it using the information vector and the information matrix, and applying the information filter, the respective update can be performed with particularly little computational effort and thus particularly quickly or with correspondingly fewer computing resources, especially in real time. This can, for example, enable such an update step to be executed on a conventional embedded system or control unit on average, especially in unfavorable cases, in less than 1 ms.

[0044] Here, a combination of estimation of the travel path, the description of the travel path's condition using parameters—in comparison to the currently acquired data—particularly geometric parameters, for example, a maximum of 20 parameters, and the information filter formalism is proposed. This allows a large number of different input data to be processed particularly easily, with little effort, and quickly into the current travel path estimate 14.

[0045] Overall, the described examples show how a runtime-optimized data fusion for determining the driving path can be realized, especially on embedded systems.

[0046] List of reference symbols

[0047] 1 Street

[0048] 2 Roadside

[0049] 3 lane markings

[0050] 4 guardrail

[0051] 5 Motor vehicle

[0052] 6 Environmental sensors

[0053] 7 Assistance system

[0054] 8 Interface

[0055] 9 processor

[0056] 10 data storage

[0057] 11 measuring points

[0058] 12 third-party vehicle

[0059] 13 Trajectory

[0060] 14 Driving path estimation

[0061] 15 support point

[0062] 16 steered wheels

[0063] 17 Driveway

Claims

Patent claims 1. Method for repeatedly estimating a current driving tube (14) of a motor vehicle (5), wherein during the operation of the motor vehicle (5) regularly repeatedly - several measurement data (11) that describe a particular traffic situation of the motor vehicle (5) characterize, with associated uncertainties captured, - so that in each update step an updated information matrix and an updated information vector of the information filter formalism are estimated, and -an updated estimate of the driving tube (14) is determined from the updated information matrix and the updated information vector.

2. Method according to claim 1, characterized in that the measurement data (11) describe at least the road course ahead of the motor vehicle (5) in the direction of travel at different distances from the motor vehicle (5).

3. Method according to one of the preceding claims, characterized in that the measurement data (11) are organized in a measurement data vector and the associated uncertainties in an uncertainty matrix, and in the update step the individual measurement data (11) and their associated uncertainty are taken into account independently of one another without inverting the entire uncertainty matrix.

4. Method according to one of the preceding claims, characterized in that an updated state for the drive tube (14) is first determined from the updated information matrix and the updated information vector, which determines the drive tube (14) by predetermined, in particular geometric parameters are characterized, wherein the number of these parameters is smaller than the dimension of the measurement data (11). A method according to claim 4, characterized in that the predetermined parameters for characterizing the driving tube (14) comprise its curvature, in particular at several predetermined distances (15) from the motor vehicle (5). A method according to claim 4 or 5, characterized in that, to determine the updated state of the driving tube (14), the last determined updated information matrix is ​​inverted and this inverted information matrix is ​​multiplied by the last determined updated information vector.Method according to one of claims 4 to 6, characterized in that a corresponding updated course of the driving hose (14) in the respective environment is determined from the updated state by means of a predetermined model, wherein the model models at least one predetermined condition for a permissible real driving hose course (14).Method according to one of the preceding claims, characterized in that, as part of the measurement data (11), environmental data describing the environment of the motor vehicle (5) and steering angle data of the motor vehicle (5) are recorded and it is determined whether, in an area from the respective current position of the motor vehicle (5) up to a predetermined distance from the motor vehicle (5) in its direction of travel, a lane change maneuver or a turning maneuver of the motor vehicle (5) is likely to take place and, if this is the case, the updated estimate of the driving hose (14) is determined up to the predetermined distance based on the steering angle data and from the predetermined distance based on the environmental data without taking the steering angle data into account. Assistance system (5) for a motor vehicle (5), comprising an interface for acquiring measurement data (11) that characterizes a traffic situation of the motor vehicle (5), a processor unit (9) and a computer-readable data storage device (10) coupled thereto, wherein the assistance system (5) is for The motor vehicle (5) is equipped to perform a method according to one of the preceding claims, comprising an environmental sensor system (6) for recording environmental data and an assistance system (5) according to claim 9.

Images