Method for operating a radar device, radar device, vehicle comprising a radar device, program code and computer-readable storage medium

Abstract

The invention relates to a method for operating a radar device (1), comprising the following steps to be carried out by the radar device (1): detecting data sets of a scanning area (3) of the radar device (1) which are offset from one another in time, wherein data points (q k ,p) of the data sets radar device-related position coordinates (q k ,p) and radar-related Doppler velocity values; assigning the data points (q k ,p) of a target data set (Q) of the data sets for the data points (q k ,p) of a source data set (P) of the data sets; determining parameters of a transformation T for transforming the source data set (P) into the target data set (Q) by applying a given optimization method to solve a multi-criteria overall optimization problem, wherein the overall optimization problem involves minimizing a position coordinate error function (E t (T)) and a minimization of a velocity error function (E v (T)); and determining a movement and / or a change in position of the radar device (1) based on the determined parameters.

Description

The invention relates to a method for operating a radar device, a radar device, a vehicle which has at least one radar device, a program code, and a computer-readable storage medium.According to the present state of the art, a vehicle is located primarily via satellite-assisted navigation systems. Additionally or alternatively, sensor-based odometry methods are used. The sensor-based odometry methods include, for example, lidar-based, camera-based or radar-based odometry methods. In this case, in detection processes which are respectively offset with respect to one another in time, detection ranges of a surrounding area of the vehicle are detected in each case. In the acquisition processes, data sets, also called data point clouds, are acquired. The data point clouds include data points having coordinates with respect to the sensor device. In the case of lidar or radar devices, these are reflection points of the lidar or radar beams, which are emitted by the sensor device, reflected by objects in the detection range of the sensor device and detected by the sensor device. Camera-based approaches, such as Structure from Motion, are data points that are tracked in the images of the camera device. When the vehicle moves, the vehicle environment changes, so that the data records of the detection processes change over time. By evaluating the change in the coordinates of the data points between the measurements, a change in the position of the vehicle associated therewith can be determined.Lidar devices and camera devices generate data sets with relatively many data points, but may be affected by environmental influences in the environment. Radar devices are less susceptible to environmental influences, but generate data sets with relatively few data points compared to other sensor devices.The publication by Rapp, M. et al. 2017 (Rapp, M. et al.: Probabilistic egomtion estimation using multiple automotive radar sensors. In: Robotics and Autonomous Systems, Vol. 89, 2017, pp. 136-146. ISSN 0921-8890) describes a method for estimating ego motion, which comprises two components. The first component is the spatial registration of successive scans. A reference scan is represented by a sparse Gaussian mixture model, which is improved by means of clustering algorithms. For spatial matching of successive scans, normal distribution transformation based optimization is used. The second component is a likelihood model for a Doppler velocity. Using a hypothesis for the ego motion state, an expected radial velocity is calculated and compared to an actually measured Doppler velocity. The estimation of the ego motion is carried out by a common spatial and Doppler-based optimization function.It is an object of the invention to determine a position of a vehicle based on data points of a radar device. Another object of the invention is to store the acquired data points in a map of the environment.This object is achieved by the independent patent claims. Advantageous refinements of the invention are disclosed in the dependent patent claims, the following description and the figures.A first aspect of the invention relates to a method for operating a radar device. The radar device may be provided for arrangement in a vehicle.The method comprises acquiring mutually time-offset data sets of a scanning range of the radar device. Data points of the data sets include radar-device-related position coordinates and radar-device-related Doppler velocity values. In other words, it is provided that data sets are detected by the radar device, which data sets detect a scanning range of the radar device. The data sets are acquired at different times. The data points can describe reflection points at which radar signals transmitted by the radar device can be reflected to the radar device. Position coordinates are present for each of the data points, which position coordinates describe the position of the data point with respect to the radar device. The data points also have respective Doppler velocity values that describe a velocity of the data point with respect to the radar device. In the radar device, a plurality of data sets can thus be present, which were detected by the scanning region at respective points in time.In the method, it is provided that the data points of a target data set of the data sets are assigned to the data points of a source data set of the data sets. In other words, one of the data records is defined as a source data record and another of the data records is defined as a target data record. The source data record can describe, for example, a preceding data record to which reference is to be made in the further method. The target data record can in particular be a current data record of the data records and describe the data record for which parameters of a transformation T are determined which transfer the source data record into the target data record.The target data set and the source data set have the respective corresponding data points. Respective data points of the target data set and the source data set may relate to a same feature in the scan range. The same feature can be, for example, the same data point of an object that is recorded both in the target data set and in the source data set. In the method, the data points corresponding to each other are determined by the radar device. The mutually corresponding data points can be described in a tuple m k.The data points corresponding to one another in total are described in an assignment which can comprise all tuple m k.In the method, it is provided that parameters of a transformation T ∈ SE( 3) are determined, which transform the source data set into the target data set. The transformation parameters are determined by applying a predetermined optimization method to solve an overall multi-critical optimization problem. In other words, the parameters of the transformation are determined by the predefined optimization method. The overall optimization problem includes minimizing a position coordinate error function E t( T) and minimizing a velocity error function E v( T).The position coordinate error function describes a sum of deviations between the radar device-related position coordinates of the data points of the target data set and the radar device-related position coordinates of the associated data points of the source data set transformed with the transformation as a function of the parameters of the transformation. The error functions may be adaptively weighted with the Geman-McClure kernel ρ t.In other words, the radar device-related position coordinates of corresponding data points have a deviation ||q k- Tp k|| from one another, which depends on the parameters of the transformation T. The deviations between the radar device-related position coordinates of corresponding data points result in a total deviation between the source data set and the target data set, which is indicated by the position coordinate error E t. The dependence of the position coordinate error E t on the parameters is described by the position coordinate error function E t( T). Within the scope of the optimization method, it is provided that the position coordinate error E t must be minimized.The overall optimization problem also includes minimizing the velocity error function which describes a sum of deviations between the radar device-related Doppler velocity values v Dk of the data points of the target data set and estimated radial velocity values v estk for the data points of the target data set as a function of the parameters of the transformation.The Geman-McClure kernel ρ v may be empirically tuned based on a distribution of velocityresidues in a radar scan.In other words, an estimated radial velocity value is present for each of the data points. The estimated radial velocity value has a deviation from the radar device-related Doppler velocity value of the data point ||v Dk- v estk( T)||. The deviations between the estimated radial velocity values and the radar-device-related Doppler velocity values of the data points result in a total deviation which is indicated by the velocity error E v. The dependence of the velocity error on the parameters is described by the velocity error function E v( T). Within the scope of the optimization method, it is provided that the speed error must be minimized.The minimization of the position coordinate error function and the minimization of the velocity error function may be weighted with respective weights γ.In a further step of the method, it is provided that a movement and / or a change in position of the radar device is determined based on the determined parameters of the transformation.The invention enables the Doppler velocity values provided by the radar device to be used to extend the data point-to-data point iterative closest point algorithm (ICP). Thereby, accuracy of parameter determination can be increased.The invention also comprises further developments, by means of which further advantages result.A further development of the invention provides that the method comprises determining the estimated radial speed values of the data points of the target data set from an estimated radar device speed value v S of the target data set and directions d k of the respective data points of the target data set relative to the radar device. In other words, there is an estimated speed value for the radar device. The estimated radar device speed value may be determined from the movement of the vehicle to which the radar device may be attached, for example. The directions of the respective data points of the target data set with respect to the radar device can be determined from the respective position coordinates of the respective data points with respect to the radar device. From the estimated velocity values of the radar device and the directions of the respective data points with respect to the radar device, the respective estimated radial velocity values of the data points can be determined.A further development of the invention provides that the method comprises determining the estimated radar device speed value of the target data set from the Doppler speed values of at least some of the data points of the target data set. In other words, the estimated radar device speed value of the target data set is determined from the Doppler speed values of the at least some data points of the target data set taking into account the radar device-related position coordinates of the data points of the respective data points with respect to the radar device.A development of the invention provides that the method comprises determining the estimated radar device speed value of the target data set from an estimated radar device speed value of the source data set. In other words, the estimated radar device speed value of the target dataset may depend on the estimated radar device speed value of the source dataset. For example, it can be provided that a radar device speed value has been determined by the radar device for the source data set. The radar device speed value estimated for the source data set can be adopted, for example, as an estimated radar device speed value of the target data set.A further development of the invention provides that the method comprises ascertaining a reference radar device speed value of a data set from the respective parameters of at least two of the transformations for transforming preceding data sets. In other words, the reference radar device speed value of the respective data set can be determined from the parameters of previous transformations.The method further includes determining reference radial velocity values for the data points of the respective data set. The reference radial velocity values for the data points are determined from the directions of the data points of the respective data set with respect to the radar device and the reference radar device velocity value of the respective data set. In other words, it is provided that reference radial velocity values for the data points are respectively determined from the reference radar device velocity value and the respective directions of the data points with respect to the radar device.In a further step, it is provided that filtering is effected of those data points of the respective data set whose Doppler velocity values meet a predefined prefilter criterion with respect to their reference radial velocity values. In other words, the data points are prefiltered based on the deviations of their Doppler velocity values from the reference velocity values determined for them. For example, it can be provided that it is assumed on the basis of the preceding transformations that the radar system has the determined reference speed value. Using the reference radar device speed value, the respective Doppler-based reference speed values of the data points can be determined, within the range of which the Doppler-based reference speed values of the data points must lie if these are assumed to be static. If the respective Doppler velocity values of some of the data points deviate from the respective reference velocity values, it can be assumed that these are measurement errors, noise values or values which can be attributed to dynamic data points. The latter are unsuitable for determining the transformation parameters, since they contain velocity components which are not attributable to the movement of the radar device. For this reason, the corresponding data points are removed.A further development of the invention provides that the method comprises a comparison of the estimated radial velocity values of the data points of the target data set with the Doppler velocity values of the data points of the data set. In other words, it can be provided that an estimated speed of the radar device is determined from at least some of the Doppler speed values and, based thereon, an estimated radial speed value of the respective data points. The estimated radial velocity values of the data points may be compared with the Doppler velocity values of the data points in order to be able to determine possible deviations.In a further step, filtering of the data points of the target data set takes place, the Doppler velocity values of which meet a predefined post-filter criterion with respect to their estimated radial velocity values. In other words, the estimated radar device speed value of the target dataset may be determined based on the acquired Doppler speed values of the data points of the target dataset. This may be done based on at least some of the data points. Based on the geometric relationships between the data points and the radar device, the estimated radial velocity that each data point should have can be calculated. Due to noise or dynamic environment, some Doppler velocity values may deviate from the estimated radial velocity values. To improve a dataset, such data points may be filtered out.A development of the invention provides that the method comprises generating the source data record from a predetermined number of the data records. In other words, it is provided that a plurality of preceding data sets are combined by the radar device. This can provide a combined data set which can have more data points than a single one of the preceding data sets.A further development of the invention provides that the method comprises a transformation of the position coordinates of the data points of the target data set, which position coordinates are obtained from the radar device, by means of the determined parameters into a local coordinate system of the source data set. In other words, the transformation T of the source data record into the target data record is described by the parameters. By back-transforming the target data set, it is possible to transform the position coordinates of the data points of the target data set into the coordinate system of the source data set.In a further step of the development, it is provided that a deviation between the position coordinates of the data points of the source data set obtained from the radar device and the position coordinates of the assigned data points of the target data set obtained from the back-transformed radar device is determined. The deviation is determined between the data points assigned to one another.If the deviation satisfies a predefined deviation criterion, the relevant data point of the target data set is added to a global mapping which is described in a global coordinate system. The global coordinate system can be related to a reference data set.The development results in the advantage that the acquired data points can be checked for a deviation and suitable data points can be added to the global mapping for describing an environment.For use cases or application situations which can arise in the method and which are not explicitly described here, provision can be made for an error message and / or a request for inputting a user feedback to be output and / or for a default setting and / or a predetermined initial state to be set according to the method.A second aspect of the invention relates to a radar apparatus. The radar device is configured to acquire data sets of a scanning range of the radar device that are offset with respect to one another in time. The data sets include data points. The data points have position coordinates related to the radar device and doppler velocity values related to the radar device. The radar device is configured to assign data points of a target data set of the data sets to the data points of a source data set of the data sets.The radar device is configured to ascertain parameters of a transformation T for transforming the source data set into the target data set using a predefined optimization method. The predefined optimization method is provided for solving a multi-critical overall optimization problem. The overall optimization problem includes minimizing a position coordinate error function and minimizing a velocity error function.The position coordinate error function describes a sum of deviations between the radar device-related position coordinates of the data points of the target data set and the radar device-related position coordinates of the associated data points of the source data set calculated by the transformation T as a function of the transformation parameters.The velocity error function describes a sum of deviations between the radar device-related Doppler velocity values of the data points of the target data set and the estimated radial velocity values for the data points of the target data set depending on the parameters of the transformation T. The radar device is configured to determine a movement and / or a change in position of the radar device on the basis of the determined parameters.The invention also includes the control device for the motor vehicle. The control device can have a data processing device or a processor device which is configured to carry out an embodiment of the method according to the invention. For this purpose, the processor device can have at least one microprocessor and / or at least one microcontroller and / or at least one FPGA (field programmable gate array) and / or at least one DSP (digital signal processor). In particular, a CPU (central processing unit), a GPU (graphic processing unit) or an NPU (neural processing unit) can be used as the microprocessor in each case. Furthermore, the processor device can have program code which is configured to carry out the embodiment of the method according to the invention when executed by the processor device. The program code can be stored in a data memory of the processor device. The processor device can be based on at least one circuit board and / or on at least one SoC (system on chip), for example.A third aspect of the invention relates to a vehicle having at least one radar device. The motor vehicle according to the invention is preferably designed as a motor vehicle, in particular as a passenger car or truck, or as a passenger bus or motorcycle.A fourth aspect of the invention relates to a program code which, when executed by a computer or a computer group, causes the latter to execute an embodiment of the method according to the invention. The program code can be provided as binary code and / or as assembler code and / or as source code of a programming language (e.g. C) and / or as a program script (e.g. Python).A fifth aspect of the invention relates to a computer-readable storage medium comprising program code which, when executed by a computer or a computer group, causes the computer group to execute an embodiment of the method according to the invention. The storage medium may be provided at least partially as a non-volatile data memory (e.g. as a flash memory and / or as an SSD-solid state drive) and / or at least partially as a volatile data memory (e.g. as a RAM-random access memory). The storage medium can be arranged in the computer or computer network. However, the storage medium can also be operated, for example, as a so-called store server and / or cloud server on the Internet. The computer or computer network can provide a processor circuit with, for example, at least one microprocessor. The program code can be provided as binary code and / or as assembler code and / or as source code of a programming language (e.g. C) and / or as a program script (e.g. Python).The invention also includes developments of the radar device according to the invention, of the vehicle according to the invention, of the program code according to the invention, and of the computer-readable storage medium according to the invention, which have features as have already been described in connection with the developments of the method according to the invention. For this reason, the corresponding ones of the radar device according to the invention, the vehicle according to the invention, the program code according to the invention, and the computer-readable storage medium according to the invention are not described again here.The invention also includes the combinations of the features of the described embodiments. The invention therefore also comprises implementations which each have a combination of the features of a plurality of the described embodiments, provided that the embodiments have not been described as mutually exclusive.Exemplary embodiments of the invention are described below. The following shows: FIG. 1 shows a schematic illustration of a vehicle which has a radar device; FIG. 2 is a schematic illustration of the radar device of the vehicle; FIG. 3 shows a schematic representation of a source data set and a target data set; and FIG. 4 shows a schematic illustration of a sequence of a method for operating a radar device.The exemplary embodiments explained below are preferred embodiments of the invention. In the exemplary embodiments, the described components of the embodiments each represent individual features of the invention that are to be considered independently of one another and that also develop the invention independently of one another. Therefore, the disclosure is intended to include combinations of the features of the embodiments other than those illustrated. Furthermore, the described embodiments can also be supplemented by further features of the invention that have already been described.In the figures, identical reference numerals designate functionally identical elements.FIG. 1 shows a schematic illustration of a vehicle which has a radar apparatus.The radar device may be configured to acquire data sets of a scanning region 3 of the radar device 1 acquired in a time-shifted manner with respect to one another. The data sets include data points. The data points may be reflection points at which radar beams transmitted from the radar device 1 may be reflected. The radar device 1 may be configured to determine position coordinates of the respective data points that may relate to the radar device 1. The radar apparatus 1 can furthermore be configured to acquire Doppler velocity values of the respective data points. FIG. 1 shows the data points of a data set within the scan range 3. The radar device 1 can acquire Doppler velocity values v Dk of the data points which can be related in a radial direction d k with respect to the radar device 1. The data set may include static data points of the scan region 3 and dynamic data points p 1, p 2 of the scan region 3. The static data points p k may be data points p k that may be associated with a fixed object in the environment of the vehicle 2, such as a traffic sign or wall. The respective speed values v 1, v 2 of the static data points may only depend on the radar device speed value v S which depends on the movement of the vehicle 2 with respect to the environment. The dynamic data points p 1, p 2 may be associated with a non-stationary object in the environment of the vehicle 2, e.g. a moving other vehicle 4.The respective Doppler velocity value v Dk of the data point depends on the velocity value v k of the data point and a direction d k of the data point relative to the radar device 1. The radar device 1 can be configured such that it detects time-offset data sets of the scanning range 3 of the radar device 1. By comparing the data sets, it is possible to infer a movement of the radar device 1 through the vehicle environment. The radar device 1 can be configured to assign data points of a target data set to the data points of a source data set. The data points assigned to one another can be assigned, for example, to a reflection in the respective scanning region 3 that has taken place on a same object. The radar device 1 can be configured to determine parameters which are assigned to a transformation T in order to convert the source data set into the target data set. The parameters can be determined by applying a predefined optimization method for solving a multi-critical overall optimization problem. The overall optimization problem may include minimizing a position coordinate error function E t( T) and minimizing a velocity error function E v( T).The position coordinate error function E t( T) can describe a sum of position coordinate deviations between the radar device-related position coordinates of the data points of the target data set and the radar device-related position coordinates of the associated data points of the source data set transformed with the transformation T as a function of the parameters of the transformation. The velocity error function E v( T) may describe a sum of deviations between the radar device-related Doppler velocity values v Dk of the data points of the target dataset and estimated radial velocity values for the data points of the target dataset depending on the parameters of the transformation T. The radar device 1 can be configured to determine a movement or a change in position of the radar device 1 from the determined parameters of the transformation T.The estimated radial velocity values v estk for the data points of the target data set may be based on an estimated radar device velocity value v S. The estimated radar device speed value v S may be determined from at least some of the Doppler speed values v Dk. FIG. 1 shows a dependence of the Doppler velocity values v Dk of the data points on an angle θ of a direction d k of the data point relative to the radar device 1. It can be seen in FIG. 1 that the dynamic data points p 1, p 2, which can be assigned to the moving other vehicles 4, are outside the illustrated profile of the estimated radial speed values v estk with respect to θ. These data points are not suitable for the parameter determination, since their speed values v 1, v 2 are not based solely on the movement of the radar device 1. For this reason, it may be provided that the dynamic data points p 1, p 2 are filtered out by the radar device 1.FIG. 2 shows a schematic illustration of the radar device of the vehicle.The radar device 1 can be mounted at a defined location on the vehicle 2. In order to be able to ascertain a movement of the vehicle 2, it is therefore necessary to transform the speed values of the vehicle v V by means of a coordinate transformation from a coordinate system V of the vehicle 2 into a coordinate system S of the radar apparatus 1. This can be done by a calibration matrix. A direction vector d k, which points in the direction of the data point p k which is located at its position coordinates p k is shown. The data point p k has a velocity value v k. The radial component of the velocity value v Dk is a projection of the velocity value v k. of the data point in the direction d k, of the radar device 1.The figure shows a rigid body geometry between the vehicle 2, the radar device 1 and the data point p k. The radar device 1 supplies only the radial component of the speed value v k of the data point. v Dk is the projection in the direction d k, of the radar device 1. a deviation can be determined from this speed v Dk and the estimated speed of the radar device v S. The estimated speed of the vehicle v est can be determined from the parameters of the transformation T. With the aid of a calibration matrix, the transformation from the reference system V of the vehicle 2 into the reference system S of the radar device 1 is possible. If, for example, the vehicle 2 rotates about its center of gravity, the speed of the vehicle 2 in the x / y direction is zero, while the radar device 1 has a speed since it is not at the center of gravity of the vehicle 2.FIG. 3 shows a schematic representation of a source data set and a target data set.The source data set can be composed of a plurality of data sets by the radar device 1 and thus contain the data points of the relevant data sets. The data points may be transformed to lie in an identical coordinate system. Data points of a target data set, which may have been acquired temporally after the data points of the source data set, are likewise shown. As a result of the movement of the radar device 1, the data points can be offset with respect to the associated data points. The displacement may be caused by the movement of the vehicle. In the case of dynamic data points, the displacement can also be caused by the movement of the object in question. Moreover, the displacement may also be caused by measurement errors. In order to be able to take only static data points into account, the data points of the target data set can be transformed back with the aid of the parameters of the transformation T, so that they are located in the coordinate system of the source data set. Some data points of the target data set are superimposed on data points of the source data set. Other data points may deviate from each other, wherein the deviation may exceed a predetermined threshold. This deviation may be caused by dynamic movements and measurement errors. It can be provided to filter out the relevant data points.The radar device 1 may be configured to add the data points to a global map that can map the environment.FIG. 4 shows a schematic illustration of a sequence of a method for operating a radar device.A first step S 1 of the method can comprise a source data set of a scanning region 3 being acquired by the radar device 1, wherein data points of the source data set comprise radar device-related position coordinates and radar device-related Doppler velocity values.A second step S 2 can comprise the acquisition of a target data record which can be acquired in a temporally offset manner, for example at a later point in time than the source data record. The target dataset may include data points of the target dataset, which may include radar-device-related position coordinates and radar-device-related doppler velocity values. Due to a movement of the radar device 1 with respect to an environment between the time of acquisition of the source data set and the time of acquisition of the target data set, the radar device-related position coordinates of the data points of the target data set may deviate from the radar device-related position coordinates of the data points of the source data set.In order to be able to determine the movement of the radar device 1 that occurred between the time of the acquisition of the source data set and the time of the acquisition of the target data set, it can be provided in a third step S 3 to assign the data points of the target data set to the data points of the source data set.A fourth step S 4 may comprise ascertaining parameters of a transformation T for transforming the source dataset into the target dataset by applying a predefined optimization method for solving a multi-critical overall optimization problem. An overall optimization problem to be solved by the predetermined optimization method may include minimizing a position coordinate error function and minimizing a velocity error function.The position coordinate error function can describe a sum of deviations between the radar device-related position coordinates of the data points of the target data set and the radar device-related position coordinates of the assigned data points of the source data set transformed by means of the transformation T as a function of the parameters of the transformation T. The velocity error function may describe a sum of deviations between the Doppler velocity values of the data points of the target data set and estimated radial velocity values for the data points of the target data set depending on the parameters of the transformation T.In a fifth step S 5, the radar device 1 can determine, based on the determined parameters, a movement and / or a change in position of the radar device 1, which can have occurred between the time of acquisition of the source data set and the time of acquisition of the target data set. On the basis of the ascertained movement and / or change in position of the radar device 1, the radar device 1 can ascertain a movement and / or change in position of the vehicle 2 on which the radar device 1 is located.A core of the method comprises a determination of the parameters in order to obtain the transformation T, which minimizes a deviation between two data sets, also known as data point sets or data point clouds, a source data set on the one hand and a target data set on the other hand. Each of the data sets is derived from a respective radar measurement performed by the radar device of the vehicle, wherein the vehicle has performed a certain movement between the times of the measurements. Based on the parameters of the transformation T, the movement / current position of the vehicle can be estimated.A second kernel of the method is filtering the data sets, taking into account only data points within a certain velocity range derived from the Doppler measurements associated with each data point for the above-mentioned method for deriving the transformation parameters.A third core of the method is to provide a global map of an environment, wherein the environment may include an area along a route. In doing so, the data points of each dataset may be stored to obtain a representation of the environment in the global map that may be used by localization and detection algorithms. In order to avoid storing outliers in the global map, prefiltering can take place depending on the speed of the vehicle.The aim is to know where the car is in space. Sometimes, GPS is not available, so other sensors in the car must be used. In this case, radar sensors. The radars are automatic radars which provide a data set in the form of a data point cloud per measurement. The measurement has a range value, a directional component, e.g., 20 degrees left / right, and the velocity of the data point derived from the Doppler measurement.Once this has occurred, it is necessary to determine how the car has moved between two different measurements. For this purpose, the data points of the current data set are matched to the data points of the previous data set. The target may be given its parameters for a transformation T that allows to get from the first data set to the second data set. This can be done by data point-to-data point comparison and using the Doppler velocity value.The parameters of the transformation T do not allow an absolute position of the vehicle to be determined in map coordinates. Instead, the parameters of the transformation T allow a determination of a relative location or the variation of the position of the vehicle between the two successive measurements.The method aims to estimate the pose of the vehicle and store the sensor readings in a global map of the environment. For this purpose, two new odometry methods are presented for 3D and 2D vehicle wheels.The goal of the data point-to-data point ICP is to obtain the transformation T ∈ SE(3) that minimizes the distance between a source point set and a target point set. In our method we deal with the small number of radar point clouds by creating Q as a submap which summarizes the ten previous scans. Each iteration comprises two steps.We obtain the set of data point correspondences based on the Euclidean distance between the matches, represented as tuple m k= ( p i, p j).This approach then minimizes the position coordinate error function E t between the mutually corresponding data points, the index t designating the position. The error terms may be adaptively weighted with the Geman-McClure kernel ρ tWe can derive the Jacobian of the position coordinate error for a data point k using its Lie-Algebra formulation, where ^Λ denotes the skew symmetric operator:Originally, the position coordinate error function was used only for laser sensors and only takes into account position coordinates. In the method, the position coordinate error function is extended to the radar device by assuming a rigid body geometry between the vehicle, the radar device and the data point measurement. The expected speed of the data point Vv estk is given by the projection of the estimated sensor speed Vv S in the vehicle frame, denoted by V, in the radial measurement direction Vd k for a data point k in the vehicle frame:The velocity error E v is the difference between the estimated velocity of the data point from the currently calculated update and the measured Doppler velocity, where the index v denotes the velocity:The Geman-McClure kernel ρ v is empirically tuned based on the distribution of velocity residues in a radar scan.The Jacobian of the velocity error depends on the vehicle-to-sensor calibration matrix, which is represented by a rotation matrix of a translation vector. It is given as:The optimization problem provides the optimal transformation between the previous pose and the current pose T ∈ SE(3). The final error function depends on the position coordinate errors and the velocity errors of the data points. With the weighting parameter γ ∈ [0,1], each type of error can be individually weighted. A value of γ=0 would result in a simple data point-to-data point ICP without considering the Doppler velocity values.As a first estimate for the optimization, a constant velocity from the previous pose can be assumed, as is the case with DICP and KISS ICP. However, one of the advantages of radar devices is that they provide the velocity measurement for each observed data point. With this information, the estimated speed of the radar device can be determined directly from the current data set by following a least squares approach. This also allows dynamic data point outliers to be removed with RANSAC prior to ICP optimization if they are not consistent with the estimated speed of the radar device.The final jakobi for a data point k is obtained by combining the data point-to-data point jakobi with the velocity jakobi:Overall, the examples show how a purely radar-based odometry and mapping method can be provided for autonomous vehicles.

Claims

Method for operating a radar device (1), comprising the following steps to be carried out by the radar device (1): - acquisition of mutually time-offset data sets of a scanning region (3) of the radar device (1), wherein data points (q k,p) of the data sets comprise radar device-related position coordinates (q k,p) and radar device-related Doppler velocity values; - assignment of the data points (q k,p) of a target data set (Q) of the data sets to the data points (q k,p) of a source data set (P) of the data sets; determining parameters of a transformation T for transforming the source data set (P) into the target data set (Q) by applying a predefined optimization method for solving a multi-critical overall optimization problem, wherein the overall optimization problem comprises a minimization of a position coordinate error function (E t( T)) and a minimization of a velocity error function (E v( T)); wherein - the position coordinate error function (E t( T)) describes a sum of deviations between the radar device-related position coordinates (q k, p k) of the data points (q k, p k) of the target data set (Q) and the radar device-related position coordinates (q k,p) of the associated data points (q k,p) of the source data set transformed with the transformation T as a function of the parameters of the transformation T; the velocity error function (E v( T)) describes a sum of deviations between the Doppler velocity values (v Dk) of the data points (q k, p k) of the target data set (Q) and estimated radial velocity values for the data points (q k, p k) of the target data set (Q) depending on the parameters of the transformation T, and - determining a movement and / or a change in position of the radar device (1) based on the determined parameters.Method for operating a radar device (1) according to claim 1, comprising the following steps to be carried out by the radar device (1): - determining the estimated radial speed values of the data points (q k, p k) of the target data set (Q) from an estimated radar device speed value of the target data set (Q) and directions (Vd k) of the data points (q k, p k) of the target data set (Q) with respect to the radar device (1).Method for operating a radar device (1) according to one of the preceding claims, comprising the following steps to be carried out by the radar device (1): - determining the estimated radar device speed value of the target data set (Q) from the Doppler speed values (v Dk) of at least some of the data points (q k, p k) of the target data set ( Q ). Method for operating a radar device (1) according to one of the preceding claims, comprising the following steps to be carried out by the radar device (1): determining the estimated radar device speed value of the target data set (Q) from a radar device speed value (1) of the source data set ( P ). Method for operating a radar device (1) according to one of the preceding claims, comprising the following steps to be carried out by the radar device (1): - determining a reference radar device speed value (1) of a data set from the parameters of the at least two transformations T for transforming respective previous data sets of the respective data set; - determining reference radial speed values for the data points (q k,p) of the respective data set from the directions (d k) of the data points (q k, p) of the respective data set with respect to the radar device (1) and the reference radar device speed value (1) of the respective data set; and filtering out the data points (q k, p k) of the respective data set, the Doppler velocity values (v Dk) of which meet a predetermined prefilter criterion with respect to their reference radial velocity values.Method for operating a radar device (1) according to one of the preceding claims, comprising the following steps to be carried out by the radar device (1): - comparing the estimated radial speed values of the data points (q k, p k) of the target data set (Q) with the Doppler speed values (v Dk) of the data points (q k, p) of the target data set (Q) - filtering out the data points (q k, p k) of the target data set (Q) the Doppler speed values (v Dk) of which meet a predefined post-filtering criterion with respect to their estimated radial speed values.Method for operating a radar device (1) according to one of the preceding claims, comprising the following steps to be carried out by the radar device (1): - generating the source data set ( P ) from a predetermined number of the data sets.Method for operating a radar device (1) according to one of the preceding claims, comprising the following steps to be carried out by the radar device (1): - retransforming the radar device-related position coordinates (q k, p k) of the data points (q k, p k) of the target data set ( Q ) by means of the determined parameters into a local coordinate system of the source data set ( P ), _ner63_ - determine a deviation between the radar device-related position coordinates (q k p k) of the data points (q k, p k) of the source data set (P) and the back-transformed radar device-related position coordinates (q k, p k) of the associated data points (q k, p k) of the target data set (Q), - add the data points (qk, pk) of the target data set (Q), which meet a predetermined deviation criterion with respect to the deviation in a global mapping in a global coordinate system.Radar device (1), characterized in that the radar device (1) is configured to - acquire time-offset data sets of a scanning range (3) of the radar device (1), wherein data points (q k, p k) of the data sets comprise radar device-related position coordinates (q k, p k) and radar device-related Doppler velocity values (v Dk) ; - assign data points (q k, p k) of a target data set ( Q ) of the data sets to the data points (q k, p k) of a source data set ( P ) of the data sets; parameters of a transformation T for transforming the source data set (P) into the target data set (Q) by applying a predetermined optimization method for solving a multi-critical overall optimization problem, wherein the overall optimization problem comprises a minimization of a position coordinate error function (E t( T)) and a minimization of a velocity error function (E v( T)); wherein - the position coordinate error function (E t( T)) describes a sum of deviations between the radar device-related position coordinates (q k, p k) of the data points (q k, p k) of the target data set (Q) and the radar device-related position coordinates (q k, p k) of the associated data points (q k, p k) of the source data set transformed with the transformation T as a function of the parameters of the transformation T; the velocity error function (E v( T)) describes a sum of deviations between the Doppler velocity values (v Dk) of the data points (q k, p k) of the target data set (Q) and estimated radial velocity values for the data points (q k, p k) of the target data set (Q) as a function of the parameters of the transformation T, and to determine a movement and / or a change in position of the radar device (1) on the basis of the determined parameters.Vehicle (2) comprising at least one radar device (1) according to claim 9.Program code which, when executed by a computer or a computer cluster, causes the latter to execute an embodiment of the method according to one of Claims 1 to 8.A computer readable storage medium comprising program code according to claim 11.

Images