Environment model with adaptive grid

Abstract

Method for providing an environment model for a vehicle (1), comprising the steps: Receiving data from at least one sensor system (10) for detecting the environment; Creating a grid map as the environment model based on the received data, wherein the grid (2) of the grid map consists of a plurality of cells (3), each representing an area of the environment of the vehicle (1); and Adapting the representation of the environment by the grid map depending on a driving situation of the vehicle (1) by changing the size and / or geometry of the areas of the environment, represented by the cells (3), and / or by changing the arrangement of the cells (3) in the grid (2) of the grid map, wherein the cells (3) of the grid (2) are arranged in blocks (B) and are managed block by block, the method further comprising a step of transmitting the information of the grid map to at least one downstream application (50, 60), wherein the transmission takes place block by block, and wherein preferably the blocks (B) are transmitted in an order wherein first blocks (B5 to B8) close to the vehicle (1), then more distant blocks (B1 to B4) in front of the vehicle (1) and finally blocks (B9 to B10) behind the vehicle (1) are transmitted, or whereby first more distant blocks (B1 to B4) in front of the vehicle (1), then blocks (B5 to B8) close to the vehicle (1) and finally blocks (B9 to B10) behind the vehicle (1) are transmitted, and / or where a flag is stored for each block (B), which indicates whether or not at least one cell (3) of the block (B) has been modified by new sensor measurements compared to the last update cycle.

Description

The present invention relates to a method and a unit for providing a surroundings model for a vehicle.Vehicles are increasingly being equipped with modern driver assistance and safety functions, which assist a driver in driving a vehicle, and which can increasingly take over the driving of the vehicle itself in various situations. In order to be able to implement these tasks, the vehicles are equipped with increasingly more and better environment sensors, which make possible an ever better detection of the environment of the vehicle.One task here is the detection and tracking of static, generic objects which are detected with the aid of the various surroundings sensors in the environment of the vehicle. For this purpose, grid-based approaches are increasingly used, wherein the environment around the vehicle is divided in a grid or "grid" in a flat plane into square cells, which are filled with busy or idle information depending on the sensor signal. This creates an occupancy map around the vehicle, which can be used during orientation and object recognition. Corresponding grid-based environmental models are known, for example, from the publications DE 10 2007 013 023 A1, DE 10 2009 007 395 A1, DE 10 2010 006 828 A1, DE 10 2011 081 740 A1 and WO 2013 / 060323 A1.The grid maps are selected to be world-fixed, that is to say with a coordinate system which has a fixed relationship to the real world, such that the vehicle moves relative to the grid map. This has the advantage that the own movement of the vehicle relative to the world does not result in the grid map having to be converted and transformed during each update cycle in accordance with the movement carried out in the meantime by the vehicle. Thus, the execution of unnecessary transformation operations is avoided, and more precise results can be obtained.The grid maps are constructed as square grid maps of square cells. In this way, account is taken of the fact that the vehicle moves relative to the grid map and, in particular, the orientation of the vehicle can change arbitrarily over time. Even if the vehicle therefore changes its orientation by any angle relative to the world and thus also relative to the grid map, it can therefore be ensured that the grid map always covers a constant minimum area in front of the vehicle, corresponding to half the lateral length of the square grid map.To enable efficient implementation, the grid map is constructed similar to a two-dimensional circular buffer, with the rows or columns of the map that run behind the vehicle from the field of view deleted and appended on the opposite side of the grid as corresponding new rows or columns.This approach has proven itself in particular in complex urban scenarios. For example, with a grid of 500×500 cells and a size of the individual cells of 8 cm×8 cm, a grid map can be used to map a range of 40 m×40 m, corresponding to a range of 20 m to each side of the vehicle.For high speeds, such as, for example, journeys on a freeway at speeds of, for example, more than 200 km / h, a substantially greater forward-view distance is required, on the other hand, which must be covered accordingly by the grid map, so that actuators and warnings can be controlled in good time.This requirement can be met by selecting a correspondingly larger grid with a correspondingly larger number of cells per column and row in order to achieve the desired look-ahead width. However, this has the disadvantage that the quadratic increase in the number of cells of the grid, which is associated with the enlargement of the grid map, leads to a correspondingly quadratic increase in storage requirement and computing time, which can represent a problem even for current, powerful computer platforms.If, on the other hand, the number of cells of the grid map is retained in order to achieve a greater preview width and the size of the individual cells is respectively increased accordingly, for example to 16 cm×16 cm or 20 cm×20 cm, the required storage requirement and the computing time can be retained, but by the price of a reduced resolution of the information which can be reproduced in the grid map and, associated therewith, a reduced suitability for use in complex urban scenarios.The publication DE 10 2013 210 263 A1 discloses an occupancy map for a vehicle, comprising a plurality of cells arranged in a grid-like manner, wherein the cells of the occupancy map are adapted to the driving situation depending on a driving situation of the vehicle.The publication DE 10 2011 113 016 A1 discloses a method for the environmental representation of a vehicle, in which environmental data are captured and stored in hierarchical data structures and objects are identified in the environment, wherein a level of a level of detail of the hierarchical data structures in different regions is set depending on properties of objects identified there. In this case, an uncertainty of the object identification is determined and the level of detail is increased in those regions in which objects are detected with a high uncertainty of the object identification.From the publication DE 10 2012 005 851 A1, a method is known for warning the driver of a motor vehicle of the presence of an object in the environment of the motor vehicle by displaying images of the environment on a display device of the motor vehicle, having the steps: capturing a camera image of the environment by means of a camera of the motor vehicle and identifying the object in the camera image, ascertaining an expected trajectory of the motor vehicle, generating a display image on the basis of the camera image, wherein the display image shows a superimposed optical representation of the expected trajectory, and displaying the display image on the display device, ascertaining a current position of the object relative to the expected trajectory and changing the optical representation of the trajectory in the display image as a function of the relative position.It is therefore the object of the present invention to specify a method and a unit for providing a surroundings model for a vehicle, which overcomes the above disadvantages.It is a further object of the present invention to specify a method and a unit for providing a surroundings model for a vehicle, which are suitable for use in different application situations.It is a further object of the present invention to specify a method and a unit for providing a surroundings model for a vehicle, which allow a large forward-view range with a high information quality and with low outlay on computing requirements and storage space.These and other objects of the present invention are achieved by methods, a model environment unit, a driver assistance system and a vehicle as defined in claims 1, 2, 12, 13 and 14. Further preferred embodiments are set out in the dependent claims.As a solution according to claim 1, a method is specified for providing an environmental model for a vehicle, comprising the steps of: receiving data of at least one sensor system for detecting the environment; creating a grid map, as the environmental model, based on the received data, wherein the grid of the grid map consists of a plurality of cells, which each represent a region of the environment of the vehicle; and adapting the representation of the environment by the grid map depending on a driving situation of the vehicle by changing the size and / or geometry of the regions of the environment, which are represented by the cells, and / or by changing the arrangement of the cells in the grid of the grid map.With the present invention it has been recognized that it is not necessary to operate continuously with a constantly constant grid configuration. Rather, this can be deviated from and the grid configuration can be adapted to these in different ways depending on a given driving situation. This opens up a wide variety of possibilities for arranging and configuring a given number of cells in different ways, in order to thus allow the most optimum possible modelling in the range and resolution of the section of the environment relevant to the vehicle, or to driver assistance functions and other applications, without requiring unnecessary additional outlay in storage space and computing time for this purpose.As an alternative solution according to claim 2, a method for providing an environmental model for a vehicle by an environmental model unit is set up for creating an environmental model in the form of a grid map, as described in claim 2.In the methods according to claims 1 and 2, the cells of the grid are arranged in blocks and managed block by block.The said methods according to claims 1 and 2 further comprise a step of transmitting information of the grid map to at least one downstream application, wherein the transmission is performed block by block, and wherein the blocks are transmitted in an order wherein first blocks close to the vehicle, then more remote blocks in front of the vehicle and finally blocks behind the vehicle are transmitted. Alternatively, the blocks may be transmitted in an order of first transmitting more distant blocks in front of the vehicle, then blocks near the vehicle, and finally blocks behind the vehicle.The transmission of the information of the environment model can be adapted in this way such that it is advantageous from the point of view of the downstream application consuming the information. Thus, a transmission in which information about the immediate environment of the vehicle is first sent allows a downstream application to respond quickly to changes in the immediate environment, such as emergency braking, for example, if an object is detected that suddenly crosses the roadway of the vehicle. Conversely, early detection of a far-ahead obstacle on the roadway, for example, may allow a corresponding avoidance reaction to be planned and carried out, in order to avoid emergency braking or an accident, for example.For each block, in the aforementioned methods according to claims 1 and 2, a flag can be further or alternatively preferably stored, which indicates whether or not at least one cell of the block has been modified by new sensor measurements compared to the last update cycle.The driving situation may include a speed of the vehicle, a road type, a course of the driving lane, a traffic density and / or a width of a roadway on which the vehicle is moving.The grid map is preferably constructed substantially along the lane of travel of the vehicle. In this way, only the area of the roadway on which the vehicle is moving or the driving path in which the vehicle is moving may be mapped in the grid map. The representation of the environment by the grid map can thus be concentrated on the area of the environment that has the highest relevance for the movement of the vehicle. Other regions of the environment, such as regions beyond the roadway, on the other hand, may be "hidden" in this way and not mapped as an environment model in the grid map. This has the advantage that no attention needs to be given to objects or events in those areas from which no effect or at most only a very slight effect or other relevance for the guidance of the vehicle can be expected. Computation time and working memory can thus be saved.Preferably, the cells of at least one row of the grid can represent a region which is bounded on two opposite sides in each case by a curve which substantially corresponds to the course of the lane.The curvature of the lane is used in this way as a specification to create a correspondingly curved grid map. The grid map can thus be constructed in a manner conforming to the actual course of the lane. In this way, it is also achieved that the cells are arranged corresponding to the direction of movement of the vehicle. In other words, the vehicle travels substantially parallel to a longitudinal axis of the grid across the grid map, wherein the longitudinal axis is defined by the curve profile of the cell boundaries, as defined by the lane profile. In this way, it is achieved that the grid map is always substantially optimally oriented relative to the orientation of the vehicle.It is also possible that the cells of at least one row of the grid represent a region whose extension in the direction of the vehicle movement is adapted depending on the speed of the vehicle. Alternatively or additionally, it is also possible for the cells of at least one row of the grid to represent a region whose extent in the direction of the vehicle movement is an integer multiple of the extent in the direction transverse to the vehicle movement.The cells of the lattice are thus selected to be correspondingly longer than wide, whereby the dissolution in the longitudinal direction is reduced, while the dissolution in the transverse direction can be maintained. This is possible because the vehicle always moves substantially along the longitudinal direction of the cells, and not in the transverse direction. In this way, with the number of cells remaining the same, a larger look-ahead width can be achieved accordingly, or alternatively, for a given desired look-ahead width, the number of cells in the grid can be reduced. Computation time and working memory can thus be saved.If the length adjustment of the cells takes place depending on the speed, for example by selecting a greater length at a higher speed, the forward-look width of the speed can also be adjusted.The adaptation is preferably carried out in discrete steps, preferably in steps of doubling or bisecting.If the adaptation is carried out only in discrete steps, it is correspondingly less necessary to carry out an adaptation. Therefore, the grid map has to be converted more rarely, as a result of which the errors arising from conversion can be correspondingly reduced and the outlay, in particular computing time, for the conversion can be reduced. The adaptation can be realized particularly easily if a change is selected as a bisection or doubling of the length or the width of the cells. Thus, for example, in the case of a bisection of the cell length, each cell of the starting grid can be divided into two cells of the adapted grid, and in the case of a doubling of the cell length, in each case two cells of the starting grid can be combined and united to form one cell of the adapted grid.In a further preferred embodiment, the cells of the grid can be arranged in blocks and managed block by block. The blocks may each have an equal number of cells in n rows and m columns. Preferably, the number of rows n is equal to the number of columns m. Further preferably, the cells of a block can each represent environmental regions with the same geometry.In particular in the case of a large number of cells, the software management, in particular the storage, updating and also the transmission to downstream applications which evaluate the information stored in the grid map, can be made more efficient by the block-by-block arrangement and management of the cells.It is also possible that the blocks have different cell sizes depending on the distance from the vehicle. A distance-dependent resolution can thereby be realized.With the aid of the flag it is possible to check in a quick and simple manner whether a block has been modified. If the flag indicates that no modification is present, downstream applications can optionally dispense with reevaluation of the block or of the cells combined in the block and computing time can be saved.As a further solution, an environment model unit is specified for providing an environment model for a vehicle, wherein the environment model unit is configured to receive data from at least one sensor system for environment detection, and based on the received data to create a grid map as the environment model, wherein the grid of the grid map consists of a plurality of cells, which each represent a region of the environment of the vehicle; and wherein the environment model unit is further configured to adapt the representation of the environment by the grid map depending on a driving situation of the vehicle by changing the size and / or geometry of the regions of the environment, which are represented by the cells, and / or by changing the arrangement of the cells in the grid of the grid map.The environment model unit is preferably further configured to execute the methods described herein.The environment model unit may be part of a driver assistance system.As a further solution, a vehicle is specified which has at least one sensor system for detecting the environment of the vehicle, and also has the environment model unit and / or the driver assistance system.The present invention will be described below with reference to preferred embodiments with reference to the drawings: FIG. 1 schematically illustrates a vehicle according to an embodiment; FIG. 2 schematically shows an exemplary grid for a grid map for a first driving situation; FIG. 3 schematically shows an exemplary driving situation of a vehicle that is moving on a multi-lane roadway of a freeway; FIG. 4 schematically shows an exemplary grid for a grid map for the driving situation of FIG. 3 ; FIG. 5 schematically shows a further exemplary driving situation of a vehicle which is moving on a multi-lane roadway of a freeway; FIG. 6 schematically shows an exemplary grid for a grid map for the driving situation of FIG. 5 ; and FIGS. 7A to 7C show exemplary grids for grid maps constructed from a plurality of blocks of cells.FIG. 1 shows a vehicle 1 according to an exemplary embodiment. The vehicle 1 is equipped with a sensor system 10 which is configured to detect at least one region of the environment of the vehicle 1 by sensor. The sensor system 10 can be, for example, a camera system which captures an area in front of the vehicle 1 and which can be configured, for example, to recognize lane markings, preceding vehicles or other objects on or next to the roadway. This example is not limiting, and other sensor systems, such as radar systems for the short range or the long range, ultrasonic systems or other known systems for detecting the environment of the vehicle 1, or any desired combination thereof, can also be used as sensor system 10 alternatively or additionally. The environment detection is also not limited to a detection of the area in front of the vehicle 1, and alternatively or additionally areas laterally and / or behind the vehicle 1 can also be detected, wherein particularly preferably a 360° detection can be realized around the vehicle 1.The sensor data acquired by the sensor system 10 are output, optionally after preprocessing such as plausibility checking, association and fusion of the outputs of a plurality of individual, different sensors of the sensor system 10, to a surroundings model unit 20.The environment model unit 20 is configured to generate, based on the received sensor data, an environment model in the form of a grid map which describes the environment of the vehicle 1. For this purpose, the environment of the vehicle 1 is divided in a grid or "grid" in a flat plane into a multiplicity of cells adjoining one another, wherein each cell represents a corresponding region of the environment of the vehicle 1. Based on the received sensor data, corresponding information values are assigned to the cells of the grid in each case in order to generate a grid map as a representation of the environment of the vehicle 1. The information values assigned to the cells can be occupancy information which indicates whether the cell, or the area of the environment represented by the cell, has been recognized as occupied or as free in order to create a grid map in the form of an occupancy map. The occupancy information may be stored as binary information, i.e. as 0 or 1, or preferably as probability values in the range between 0 and 1, to indicate the probability for an occupancy. However, this is only exemplary, and the environment model unit 20 can also be configured to generate other types of occupancy maps, or other grid-based map types, such as feature maps that contain height information or intensity information, for example.The grid map created by the environment model unit 20 can be output to various applications, in particular driver assistance function applications. FIG. 1 shows, for example, a lane recognition unit 50 which, based on the grid map produced and output by the environment model unit 20, can carry out object recognition for recognizing lane markings and, based on this, can carry out recognition of the course of the lane of the vehicle 1. The information about the recognized driving lane can in turn be made available to the environment model unit 20 by the driving lane recognition unit 50, for example in order to allow the environment model unit 20 to adapt the representation of the environment by the grid map to the course of the driving lane.As further illustrated in FIG. 1, the vehicle 1 can further have an odometry unit 30, which can be configured to provide information about the position, orientation and speed of the vehicle 1. Optionally, a navigation device 40 can also be provided, which can provide information on a road type, a roadway width, the number of lanes or a lane course based on data of a digital road map. This information can be used by the environment model unit 30 to determine whether and possibly how an adaptation of the representation of the environment by the grid map is to be or can be carried out.With reference to FIGS. 2 to 7C, it will now be described in greater detail how the environment of the vehicle 1 can be represented by the grid map, and the possibilities of suitably adapting the representation of the environment to different driving situations.FIG. 2 shows a grid 2 having a multiplicity of cells 3. Each of the cells 3 represents a representation of a square region in the environment of the vehicle 1. The cells 3 are arranged in regular rows and columns, wherein the number of cells 3 of a row corresponds to the number of cells 3 in a column. This results in a square grid 2 which represents a corresponding square area in the environment of the vehicle 1. The cells 3 can each represent, for example, a square area of the environment of 4 cm×4 cm. The square lattice 2 can be constructed, for example, from 1,000×1,000 cells. A square region of 40 m×40 m in the environment of the vehicle 1 can thus be represented and depicted as an environment model in the form of a grid map.The grid map is world-fixed, that is, with a coordinate system having a fixed relationship to the real world, and the vehicle 1 moves relative to the grid map. Cells 3 behind the vehicle 1 which travel out of the view of the vehicle 1 due to the vehicle movement are removed and reinserted as new cells 3 at respective opposing locations of the grid 2. This can be done row-wise or column-wise. It can likewise be provided that this takes place for individual cells 3 in each case. In this way, the grid map 2 can "move" along with the movement of the vehicle 1,The example of FIG. 2 provides a relatively high-resolution environment model which, starting from the vehicle 1 placed substantially centrally, maps the environment up to a distance from the vehicle 1 which is substantially the same in each direction. This type of environment representation can be suitable in particular for parking situations or in complex urban driving situations in which highly accurate localization of other objects in the nearby environment of the vehicle 1 is desirable.FIG. 3 illustrates, by way of example, a driving situation in which the vehicle 1 is moving on a three-lane roadway 8 of a freeway. The individual lanes of the roadway 8 are marked by lane markings 6. Lines 7 represent road boundaries.The driving situation of driving on a freeway is often associated, in particular in Germany, with high speeds of both the host vehicle 1 and other vehicles, it being possible for the speeds to exceed a value of, for example, more than 200 km / h. In this situation, it is therefore desirable, in particular for future developments in the field of highly autonomous driving, to detect the traffic even with a forward-view range over longer distances of 80 m, 100 m or more in front of the vehicle and to map it in the environmental model.If the principle of a high-resolution, square grid 2 shown and described with reference to FIG. 2 were applied accordingly to the driving situation of FIG. 3, the number of cells would have to be multiplied by the square of the corresponding magnification factor for an increased preview width.In order to avoid this, it is therefore proposed to adapt the representation of the environment of the vehicle 1 by the grid map to the changed driving situation, as illustrated by way of example in FIG. 4.As shown in FIG. 4, a grid 2 can be used for the driving situation of FIG. 4, which grid is not square, but is constructed substantially along the roadway 8, or a lane 5, in particular the lane 5 of the vehicle 1. In particular, as shown in FIG. 4, the grid 2 may be configured as a rectangular grid 2, wherein the grid 2 is substantially aligned along the course of the lane 5 such that the longitudinal edges of the cells 3 are substantially parallel to the course of the lane 5.The information about the course of the lane 5 can be provided by the navigation device 40, for example, or can be estimated by the lane recognition unit 50. In the case that the information about the course of the lane 5 is omitted, for example because the navigation device 40 does not contain map data about a travel area, or because the lane markings 6 on a travel route cannot be recognized or are so poor due to wear that the lane recognition unit 50 cannot recognize the course of the lane, a last current course of the lane can be retained and extrapolated into the future. Alternatively, the curvature can be reduced stepwise to zero starting from the last current lane course. If indicated by the circumstances, the preview width can be adjusted accordingly at the same time.In the longitudinal direction of the track 5, the resolution of the grid map can be reduced. Accordingly, the cells 3 or the area of the environment represented by each cell 3 can likewise be embodied as rectangular, such that the cells 3 each represent an area that is longer than wide. The cells 3 can each represent, for example, a region having a length of 16 cm. In this way, without having to increase the number of cells 3 of the grid 2 in the longitudinal direction, the look-ahead width can be extended substantially. If the length of the cells 3 is adapted, for example, depending on the speed of the vehicle 1, the forward-view range of the speed can be adapted.Since the vehicle 1 moves substantially parallel to the longitudinal direction of the grid 2, and thus parallel to the longitudinal direction of the cells 3, the lateral resolution is not appreciably impaired by the change in length of the cells 3. Such a resolution has proven to be sufficiently accurate for many current and future intended fields of application, in particular for highway driving at high speed in practical tests.Laterally of the vehicle 1, it may be sufficient that the grid 2 only covers the region of the roadway up to the first lateral obstacles. These are typically the road edges which may be marked, for example, by guardrails, guardrails, curbs, a grass scar, and so forth. This can be the case in particular if the environmental model provided by the environmental model unit 20 is provided only to applications 50, 60 that do not require information about events away from the roadway 8. For other applications, it may also be expedient for the grid 2 to cover only the region of a driving path in which the vehicle is moving.For other applications, such as a Simultaneous Localization and Mapping (SLAM) application, it may additionally be necessary to capture an additional area beyond the obstacles on both sides of the roadway 8, such as an additional strip of 3 m or 5 m width beyond the first lateral obstacles, in order to be able to map buildings or other landmarks that may serve for localization, for example, in the mesh map.In both cases, it is therefore sufficient to record and image a relatively limited width in the environmental model laterally of the vehicle 1, such as 20 m, 30 m, or 40 m. This makes it possible to choose a high resolution for the cells 3 in the direction transversely to the course of the roadway 8, for example 4 cm.The delimitation of the grid map in the lateral direction can furthermore have the advantage of improving the quality of the representation of the environment of the vehicle 1 by the grid map. This is based on the fact that often a sufficient number of measurement results cannot be obtained in the region of the lateral limits of the detection range of the sensor system 10 during travel in order to be able to derive reliable values for the grid map therefrom.In order to perform the above-described adaptation of the grid 2 by changing the length and / or the width of the cells 3 or the regions of the environment of the vehicle 1 represented by the cells 3, the environment model unit 20 can calculate a new grid map with a correspondingly changed and adapted second grid configuration, for example, starting from a current grid map with a first grid configuration. For this purpose, the second grid 2 can be superimposed on the current grid map and the corresponding values for the new grid map can be calculated on the basis of the information of the current grid map. In this way, it is possible to change from the current grid map to the new grid map in an update cycle.Alternatively, the adaptation can also take place successively, for example in such a way that cells 3 of the current grid map that are omitted from the field of view behind the vehicle 1 are removed and new cells 3 with a correspondingly changed geometry are added in the field of view in front of the vehicle 1 according to the adapted, second grid configuration.In both cases, it is preferred that changes in the length and / or the width of the cells 3 are carried out only in discrete steps in order to avoid or reduce the outlay on computing time for conversions necessary for this purpose, and also errors possibly caused by the conversions. It is particularly preferred in this case if only respective doublings and bisections of the length and / or of the width of the cells 3 are carried out as discrete steps, for example from 4 cm to 8 cm to 16 cm to 32 cm length or width. Thus, in the case of doubling the cell length, for example, in each case two cells 3 adjoining one another in the longitudinal direction can be combined to form a new cell 3, and in the case of bisecting the cell length, the cells 3 can be split into in each case two new cells 3.It is further preferred that angle changes, in particular as a rotation of the alignment or the longitudinal axis of the grating 2, are likewise carried out only in discrete steps. An angle change can be carried out in such a way that an angle offset by one full cell 3 is achieved between at least two rows of the grid 2. Thus, for a ratio of the length to the width of a cell of 1:1, this can correspond to an angle of 45°, for a ratio of 2:1, an angle of 22.5° and for a ratio of 4:1, an angle of 11.23°. This can be realized by shifting the contents of the cells to the right or left row by row according to the change in curvature.Furthermore, it can also be provided that two grid maps with respectively different grid configurations are calculated in the environment model unit 20. Thus, for example, a square, high-resolution grid map can be calculated parallel to an elongated grid map adapted to the lane 5. In this case, it is not necessary that "switching" and conversion must be carried out between different grid configurations, whereby the conversion can be excluded as a possible source of errors. This may be advantageous in particular for applications of highly autonomous driving in which the advantages of the improvement in reliability and functional reliability which can be achieved thereby account for the increase in storage and computing time requirements caused by the parallel provision of two different grid maps.While in FIG. 4 the grid 2 is shown as rectangular, this is not limiting and it is also possible that the cells 3 each represent regions of a different, non-rectangular geometry, as shown for example in FIGS. 5 and 6.FIG. 5 illustrates, by way of example, a section of a three-lane roadway 8 of a freeway in the region of a curve. In this case, as shown in FIG. 6, the geometry of the cells 3 or the geometry of the regions of the environment represented by the cells 3 can be selected such that they are each bounded in the longitudinal direction by two parallel curves extending in the longitudinal direction, wherein the curves run substantially parallel to the course of the lane 5. This substantially reduces the need for recalculations or conversions of the grid map 2 due to a changed orientation of the vehicle 1 relative to the grid map.Instead of the cells 3 bounded by the curves, as are shown in FIG. 6, cells 3 in the form of trapezoids can also be used to approach the course of the lane 5 in sections.In the examples of FIGS. 4 and 6, the grids 2 were each represented as grids 2 with a number of cells 3 that is constant in each of the rows and columns. However, this is not limiting, and it is likewise possible, for example, to adapt the number of cells 3 in the direction transverse to the course of the lane 5 in each case to the width to be represented. For example, the number of cells 5 in the width direction may be increased as the number of lanes increases, or in areas of entry or exit of the roadway 8, or the number of cells 3 in the width direction may be decreased when one lane is omitted. In addition, it may also be possible to correspondingly reduce or increase the number of cells 3 in the longitudinal direction of the lane course 5 in order to obtain a grid 2 overall having a constant number of cells 3.The number of cells 3 in the length and / or width of the grid 2 can also be varied to address situations where there are several possible hoses for the movement of the vehicle 1, for example when merging or dividing the freeway, or another type of road, such as at intersections, ramps or ramps. In particular, it can be provided here to provide an adapted grid 2 for each possible driving tube.In this way, by adapting the size of the regions of the environment represented by the cells 3 and / or by changing the arrangement of the cells 3 in the grid 2 of the grid map, the region represented by the grid map as environment model can be adapted suitably to a respective driving situation without the number of cells 3 having to be increased or substantially increased for this purpose.A further simplification of the management and handling of the cells 3 of the grid 2 can be achieved if the cells 3 are combined into groups of blocks B and managed together, as shown in Figures 7A to 7C.FIGS. 7A to 7C show different configurations of grids 2 each formed by a plurality of blocks B 1 to B 10, collectively denoted by B. Each block B in turn comprises a plurality of cells 3. Preferably, the blocks B can each have n rows and m columns of cells in the manner of a matrix, where n and m are integers. The cells 3 of a block B preferably each have the same geometry, that is to say they each describe a region of the environment of the vehicle 1 of the same geometry.Each block B has a world-fixed position. Blocks B falling from the area of the environment to be covered are deleted and inserted again at a position of the environment to be covered anew. Depending on the direction, change of direction and / or speed of the vehicle 1, and / or on the width of the roadway 8, the blocks B are removed not only at the rear boundary of the viewing area, but also laterally and are re-placed laterally or at the front boundary of the viewing area. For example, if FIGS. 7A through 7C are considered to be a sequence of grid configurations occurring at different times in the course of 90 degrees left rotation of the vehicle, blocks B are removed from the right edge of the view area in the course of rotation, for example, and are newly appended to the left edge of the view area. As is shown in particular in FIG. 7B, the blocks B can be arranged offset relative to one another, for example offset by one, two or more cell rows or columns.In order to realize a distance-dependent resolution, blocks B with different cell sizes can furthermore be used for different distances, wherein blocks B that are further away have a coarser resolution. In the example, blocks B 1 to B 4 could have a resolution of 16 cm, while blocks B 5 to B 10 can each have a resolution of 8 cm.Preferably, it is provided that the blocks B are arranged such that they approach the course of the lane 5 as well as possible. In particular, it can be provided that the blocks B are arranged in such a way as to achieve the most optimal possible coverage of the roadway course 5 or of the driving path. For example, it can be achieved that the largest possible area of the environment relevant for guiding the vehicle 1 is mapped by the grid map formed from the correspondingly arranged blocks B, while other areas of the environment, such as areas laterally to the first lateral obstacles, which are not relevant or are only of minor relevance, are not mapped, or are mapped only to a small extent.The arrangement of the cells 3 in blocks B can also be used advantageously to transmit a grid map as environment model to downstream applications 50, 60. Thus, the transmission sequence of the blocks B can be adapted, for example, such that this is advantageous from the viewpoint of the application 50, 60. For example, first the blocks B5 to B8 located close to the vehicle 1, then the remote blocks B1 to B4, and finally the blocks B9, B10 located behind the vehicle, could be transmitted.Depending on the application, it may also be advantageous to transmit data at a greater distance earlier in the cycle, so that the applications 50, 60 can adapt to a driving situation lying far ahead, it being possible to assume that only a few changes still occur near the vehicle 1, or can no longer be reacted to the latter. This could be achieved, for example, by the transmission sequence B5, B6, B1, B2, B3, B4, B7, B8, B9, B10.In both cases, it can further preferably be provided that for each block B, further information about the position and orientation of the block B in the world-fixed coordinate system is transmitted, and optionally additionally information about the number and / or arrangement of the cells 3 within the block B and / or about the size and geometry of the cells 3 of the block B. In this way, it can be made easier for the applications 50, 60 to evaluate the corresponding information of the block B already directly after reception, without waiting for the transmission of the complete grid map information.In addition, it may be advantageous in the case of the block-by-block display to store a flag for each block B, which flag indicates a modification by new sensor measurements compared with the last cycle. If a block B has not been modified, that is to say none of the cells 3 of the block has undergone a change, it is possible, for example, to dispense with fusion into a central grid map or renewed evaluation by an application 50, 60 and computation time can thereby be saved.As described above, the environment model unit 20, by a suitable selection of cell size, cell geometry and grid configuration, can provide a representation of the environment of the vehicle 1 adapted to the respective driving situation by means of a grid map. In this case, it is also possible, in particular, for the environment model unit 20 to be able to operate with a predefined number of cells 3, or a predefined upper limit for the number of cells 3 in the grid 2, in order to realize the different possible and desirable environment model representations, in particular different desirable look-ahead widths. Accordingly, the amount of computing time and the requirement for the required working memory that is required for storing the information of the grid map can be limited.In order to determine which form of the representation of the environment by a grid map is most suitable for a current driving situation, it can be provided that the environment model unit 20 is configured to solve an optimization problem in which the cell sizes, cell geometry and grid configuration are considered as input parameters in order to maximize an objective function, for example in order to maximize the imaged area of the environment weighted according to relevance and resolution.Alternatively, it may also be possible for suitable forms of the representation of the environment to be respectively specified for different typical driving situations and for the environment model unit 20 to select and use the corresponding form of the representation based on a detected driving situation. For example, for a parking process, driving in urban traffic, cross country driving and driving on freeways, specifications can be made in each case about the grid configuration to be used, cell size, cell geometry and so on. Alternatively or additionally, specifications depending on speeds can also be provided, such as, for example, according to whether the vehicle 1 is moving at less than 10 km / h, between 10 km / h and 60 km / h, between 60 and 120 km / h or over 120 km / h.

Claims

Method for providing an environmental model for a vehicle (1), comprising the steps of: receiving data of at least one sensor system (10) for detecting the environment; creating a grid map, as the environmental model, based on the received data, wherein the grid (2) of the grid map consists of a plurality of cells (3) which each represent a region of the environment of the vehicle (1); and adapting the representation of the environment by the grid map depending on a driving situation of the vehicle (1) by changing the size and / or geometry of the regions of the environment which are represented by the cells (3) and / or by changing the arrangement of the cells (3) in the grid (2) of the grid map, wherein the cells (3) of the grid (2) are arranged in blocks (B) and are managed in blocks, the method further comprising a step of transmitting the information of the grid map to at least one downstream application (50, 60), wherein the transmission takes place block by block, and wherein preferably the blocks (B) are transmitted in an order wherein firstly blocks (B5 to B8) are transmitted close to the vehicle (1), then further blocks (B1 to B4) are transmitted in front of the vehicle (1) and finally blocks (B9 to B10) are transmitted behind the vehicle (1), or wherein firstly further blocks (B1 to B4) are transmitted in front of the vehicle (1), then blocks (B5 to B8) are transmitted close to the vehicle (1) and finally blocks (B9 to B10) are transmitted behind the vehicle (1), and / or wherein a flag indicating, is further stored for each block (B), whether or not at least one cell (3) of the block (B) has been modified by new sensor measurements with respect to the last update cycle.Method for providing an environment model for a vehicle (1) by an environment model unit (20) set up for creating an environment model in the form of a grid map, comprising the steps of: receiving data of at least one sensor system (10) for environment detection; creating a grid map, as the environment model, based on the received data, wherein the grid (2) of the grid map consists of a multiplicity of cells (3) which each represent an area of the environment of the vehicle (1); wherein optionally an adaptation of the representation of the environment by the grid map can be carried out on the basis of provided information; wherein the cells (3) of the grid (2) are arranged in blocks (B) and are managed in blocks, the method further comprising a step of transmitting the information of the grid map to at least one downstream application (50, 60), wherein the transmission takes place block by block, and wherein preferably the blocks (B) are transmitted in an order wherein firstly blocks (B5 to B8) are transmitted close to the vehicle (1), then further blocks (B1 to B4) are transmitted in front of the vehicle (1) and finally blocks (B9 to B10) are transmitted behind the vehicle (1), or wherein firstly further blocks (B1 to B4) are transmitted in front of the vehicle (1), then blocks (B5 to B8) are transmitted close to the vehicle (1) and finally blocks (B9 to B10) are transmitted behind the vehicle (1), and / or wherein further for each block (B) a flag is stored which indicates whether or not at least one cell (3) of the block (B) has been modified by new sensor measurements in relation to the last update cycle.Method according to Claim 2, wherein the representation of the environment by the grid map is adapted as a function of a driving situation of the vehicle (1) by changing the size and / or geometry of the regions of the environment which are represented by the cells (3) and / or by changing the arrangement of the cells (3) in the grid (2) of the grid map.Method according to Claim 1 or 3, wherein the driving situation comprises at least one of a speed of the vehicle (1), a road type, a traffic density, a course of the driving lane, and a road width.Method according to Claim 4, wherein the grid map is constructed substantially along the course of the lane of travel of the vehicle (1).Method according to Claim 5, in which the cells (3) of at least one row of the grid (2) represent a region which is bounded on two opposite sides in each case by a curve which substantially corresponds to the course of the lane.Method according to one of the preceding claims, wherein the cells (3) of at least one row of the grid (2) represent an area, the extent of which in the direction of the vehicle movement is adapted as a function of the speed of the vehicle (1).Method according to one of the preceding claims, wherein the cells (3) of at least one row of the grid (2) represent a region whose extent in the direction of the vehicle movement is an integer multiple of the extent in the direction transverse to the vehicle movement.Method according to any of the preceding claims, wherein the adaptation is performed in discrete steps, preferably in steps of doubling or bisecting extents.Method according to any of the preceding claims, wherein the cells (3) of the grid (2) are arranged in blocks (B), wherein the blocks (B) each have an equal number of cells (3) in n rows and m columns, wherein n and m are preferably the same, and wherein more preferably all cells (3) of a block (B) each represent regions of the same geometry.Method according to claim 10, wherein depending on the distance from the vehicle (1) the blocks (B) have different cell sizes.Environment model unit (20) for providing an environment model for a vehicle (1), wherein the environment model unit (20) is configured to execute a method according to one of Claims 1 to 11, wherein the environment model unit (20) is further configured to receive data from at least one sensor system (10) for environment detection, and to create a grid map as the environment model on the basis of the received data, wherein the grid (2) of the grid map consists of a multiplicity of cells (3) which each represent an area of the environment of the vehicle (1); and wherein the environment model unit (20) is further configured to optionally adapt the representation of the environment by the grid map depending on provided information, in particular depending on a driving situation of the vehicle (1), by changing the size and / or geometry of the regions of the environment represented by the cells (3) and / or by changing the arrangement of the cells (3) in the grid (2) of the grid map.Driver assistance system, having a surroundings model unit (20) according to Claim 12.Vehicle (1), having at least one sensor system (10) for detecting the environment of the vehicle (1); and an environment model unit (20) according to Claim 12, and / or a driver assistance system according to Claim 13.

Images