Lane Map Making Device

Abstract

The present invention relates to a lane map making device (1) for a vehicle (3). The lane map making device (1) has a camera unit (20) for detecting image data and a control unit (10). The control unit (10) is configured to determine a lane area (12) in the image data detected by the camera unit (20). In addition, the control unit (10) is also configured to convert the image data including the determined lane area (12) into a bird's-eye view by means of inverse perspective imaging and project it onto a layer surface. The control unit (10) is also configured to add the determined lane area (12) to a grid map within the layer. In addition, the control unit (10) is also configured to calibrate and combine at least two grid maps from two different driving processes to generate a final grid map (11) of the lane area.

Description

Technical Field

[0001] The invention relates to a lane map making device for a vehicle, a back end for lane map making, a vehicle having such a lane map making device, a method for lane map making, a computer program unit and a computer-readable storage medium. Background Art

[0002] For several years now, the development of driver assistance systems, right up to automated driving, has been gaining increasing attention. A core component of such systems is the accurate perception of the surrounding environment. This includes, in particular, the perception of other traffic participants, infrastructure and the direction of the lane. In addition to processing sensor-based signals directly in the vehicle, accurate digital maps are absolutely necessary for the navigation of autonomous systems. The creation of such maps using conventional mapping methods is complex, time-consuming and costly.

[0003] To map complex road networks, one or more mobile survey vehicles must be used to drive over all relevant roads. This means that the production, preparation and operation of survey vehicles with special surveying configurations require high personnel costs and considerable investment. Summary of the invention

[0004] One object of the present invention is to provide accurate road maps.

[0005] A first aspect of the present invention relates to a lane map making device for a vehicle. The lane map making device has a camera unit for detecting image data and a control unit. The control unit is configured to determine a lane area, such as an asphalt area, an asphalt area or a concrete area, in the image data detected by the camera unit. In addition, the control unit is further configured to convert the image data including the determined lane area into a bird's-eye view by means of inverse perspective imaging (inverse perspective transformation, inversen perspektivischenAbbildung), and project it onto a plane that preferably corresponds sufficiently approximately to a road surface. The control unit is configured to add the determined lane area to a grid map within the plane. In addition, the control unit is further configured to calibrate and combine at least two grid maps originating from two different driving processes with each other to generate a final grid map of the lane area.

[0006] In this way, an accurate digital map of the lane area can be accurately, simply and continuously generated, from which it can be seen where there are lane areas and where there are no lane areas. In other words, where vehicles can be driven and where they cannot be driven. In addition, changes such as construction sites or parking lots can be recognized and mapped. In other words, the map can be continuously supplemented and changed. In addition, the identification and positioning of the determined lane areas are averaged by multiple (at least two) grid maps from which the final grid map of the lane area is generated, whereby inaccuracies can be compensated. In addition, the probability of determining all lane areas at least once using multiple drives increases. In addition, the lane map production device can draw the lanes independently of any road markings that may be present.

[0007] For mapping, the lane map making device may include a camera unit and a control unit. The camera unit may detect or generate image data of the vehicle's surroundings. It should be noted that the camera unit may also be a stereo camera device or a panoramic surround camera device. In addition, the camera unit may have depth and / or height information for objects in the image data and for areas in the image data. This depth or height information may be provided, for example, by an additional sensor such as a lidar sensor, a radar sensor or an ultrasonic sensor, but may also be provided by a stereo camera device. In addition, the detected image data may be classified by the control unit, for example, as to whether it is an urban surrounding, a rural surrounding, a highway, daytime, nighttime, summer or winter. In addition, the control unit may analyze and evaluate the image data and determine the lane area therein. For example, the control unit may determine a dark area close to the ground as a lane area, and a curb or a kerb may be used as a boundary between a sidewalk and a lane. It should be noted that for the above determination, the control unit may be configured with an artificial neural network that has been trained in lane area determination. In order to use the image data as a map, the data is converted into a bird's-eye view or a top view. This may be done, for example, with the aid of inverse perspective imaging. The converted image data can then be projected onto a plane which is advantageously arranged at the height of the lane surface and parallel to the lane surface. This plane can be, for example, a grid map to which the determined lane area is added. In order to achieve a more accurate position determination and positioning and to obtain an improved grid map, several grid maps originating from different driving processes can be calibrated and combined with one another so that a final grid map of the lane area is generated by a control unit. The accuracy and reliability of the digital map can be increased by calibrating and combining several grid maps. In particular, several grid maps can originate from different driving directions, so that traffic islands and access roads can also be mapped in detail from different angles. The final grid map of the lane area can then be converted into a digital map or digital road map so that it can then be used as the basis for vehicle navigation.

[0008] A roadway area is understood to be an area along which a vehicle can travel, for example an area on an asphalt or concrete surface. In contrast, a sidewalk is identifiable as a non-roadway area, in particular a curbstone serving as a boundary.

[0009] It should be noted that, in addition to the camera data, the image data can also be lidar data, radar data or ultrasound data, or can be supplemented by these data, thereby in particular being able to obtain depth and height information.

[0010] A bird's eye view primarily describes a top view, but the top view may also be at an angle of, for example, ±20 degrees relative to the vertical direction.

[0011] It should be noted that a lane region may also be referred to as a road region or a road segment.

[0012] It should be noted that the control unit may also be a control switch, a switch circuit, a processor or a computing unit.

[0013] The terms “digital map” or “digital map data” are also to be understood as meaning maps which are used for advanced driver assistance systems (ADAS) or automated or autonomous driving without navigation.

[0014] According to one embodiment, the projection plane is at the height of the lane surface. In addition, the control unit is further configured to correct the lane area in the image data with the aid of a height information. In order to correct or distinguish the road from a building or a wall, the control unit may take into account height information or depth information in the image data or from an object contained in the image data. Thus, when determining the lane area, buildings or walls may be simply but reliably ignored, so that only the lane area is determined. In particular, it is conceivable that the lane area is usually flat in front of or beside the vehicle. In addition, errors in lane areas or non-lane areas may be corrected, for example if a relatively small area has been determined as a lane area, but this area is surrounded by non-lane areas, this may be considered to be an error or can be corrected as an error. As an alternative or additional measure, this may also be done for a non-lane area surrounded by a lane area.

[0015] According to one specific embodiment, the control unit is configured to classify individual regions in the image data into lanes and non-lane areas by means of an artificial neural network and to thereby determine lane areas therein.

[0016] The neural network can evaluate or analyze the image data in a simple, fast and reliable manner and determine lane areas therein or distinguish lane areas from other areas in the image data. For this purpose, the neural network can be trained accordingly beforehand / in an early stage. It should be noted that the neural network can be a recurrent neural network (RNN) or a convolutional neural network (CNN), but is not limited thereto. In this case, networks can also be used which have been pre-trained for image processing and can also be subjected to additional post-training.

[0017] According to another embodiment, the control unit is configured to calibrate and combine at least two grid maps with each other according to features contained in the grid maps to generate a final grid map of the lane area.

[0018] In order to calibrate and combine several grid maps from several driving processes, unique features such as roundabouts, intersections or traffic islands can be used. In particular, the calibration can be done by local displacements of individual grid cells of the grid map. The local displacement of a single point can be determined as a suitable combination of the displacements of neighboring points detected.

[0019] According to one embodiment, the control unit is configured to determine a probability value of correct determination / identification for each determined lane region and to mark the corresponding lane region in the grid map as a lane if the probability value exceeds or is above a predetermined threshold.

[0020] It can thus be ensured that only lane regions which have been reliably determined as lane regions enter the final grid map. This allows the error rate and the false recognition rate to be reduced. The probability value can depend in particular on the number of determinations or recognitions in relation to the number of trips. The more times a lane region is determined in this way, the higher the probability that it is indeed a lane region. Only when the probability value exceeds a predefined threshold value, such as 80%, is the corresponding lane region entered as a lane in the final grid map.

[0021] According to one embodiment, the lane map generating device further comprises a positioning unit, which is configured to determine the position of the lane map generating device, the position of the determined lane area, or the position of a point in the grid map.

[0022] For this purpose, the positioning unit can be used to classify the respective final grid maps of the lane area in a global coordinate system and to add the grid maps to the digital road map. The positioning unit can determine the position, for example, by means of satellite navigation (Global Positioning System (GPS) or Galileo Positioning System (GALILEO)) or by means of mobile radio positioning tracking. In addition, errors in the single positioning in the final grid map can be compensated by features and a large number of specific single positions of the lane area (per driving process). In addition, the positioning can be supported and further improved by odometry (distance counter in the car). In this way, an accurate map production of the lane area can be achieved.

[0023] Furthermore, it should be pointed out that in the context of the present invention, GPS stands for, for example, a global navigation satellite system (GNSS), such as the Global Positioning System (GPS), the Galileo Positioning System (Galileo), the Russian GLONASS Positioning System (GLONASS), the Chinese BeiDou Satellite Navigation System (Compass) or the Indian Indian Regional Navigation Satellite System (IRNSS).

[0024] It should be noted that the vehicle positioning can also be implemented by cell positioning. This is possible in particular when using a GSM network, a UMTS network or a LTE network. As an alternative or additional measure, positioning can also be performed based on a camera.

[0025] According to one embodiment, at least two grid maps used to generate the final grid map are derived from image data of at least two different driving directions.

[0026] In order to improve the representation of the lane area in the final grid map of the lane area, the grid maps used to generate the final grid map of the lane area can be acquired from different driving directions, such as driving from north to south and driving from south to north. This allows driving on both sides of the road and allows entry roads and intersections to be detected and determined from different angles.

[0027] According to another embodiment, the control unit is configured to determine an outer contour for the area in the grid map that is determined as a lane area. As an alternative or in addition, the control unit can also depict the outer contour by means of geometric curves, such as B-spline lines representing a smooth outer contour. By using an outer contour and / or a geometric curve, storage space can be saved, since only the boundary curves need to be depicted, and not the entire area.

[0028] According to one embodiment, the lane map generating device further comprises a transmitting / receiving unit, which is configured to transmit or send the grid map, the detected image data or the converted image data to a rear end or another vehicle.

[0029] For transmitting the detected image data, the converted image data or the grid map or for receiving the final grid map, the lane mapping device may include a transmitting / receiving unit. The transmitting / receiving unit may in particular transmit the generated grid map with the determined lane areas to a backend, which then performs further processing and generates the final grid map of the lane areas. As an alternative or in addition, the backend itself may also carry out the conversion from the detected image data and generate the grid map. Since the calculation is carried out at the backend, computing power can be saved on the lane mapping device. Advantageously, the transmission can be carried out wirelessly or by radio.

[0030] The wireless transmission or wireless reception of the grid map can be accomplished via Bluetooth, wireless local area network (WLAN) (e.g. WLAN 802.11a / b / g / n or WLAN 802.11p), wireless personal area network (ZigBee) or Worldwide Interoperability for Microwave Access (Wireless Metropolitan Area Network, WiMax), or cellular radio systems such as General Packet Radio Service (GPRS), Universal Mobile Telecommunications System (UMTS), 3G, 4G or Long Term Evolution (LTE). Other transmission protocols may also be used. The mentioned protocols have the advantage of being standardized.

[0031] According to one embodiment, the camera unit has a stereo camera.

[0032] Another aspect of the invention relates to a backend for lane mapping. The backend has a transmitting / receiving unit configured to receive a grid map from the lane marking device. The backend is configured to receive at least two grid maps from at least two lane mapping devices of one vehicle or different vehicles and to calibrate and combine these grid maps with each other to generate a final grid map of the lane area.

[0033] In addition to generating the final grid map of the lane area on the control unit of the lane mapping device, the back end can also generate the final grid map. In this way, grid maps can be generated by multiple vehicles with lane mapping devices, which are then received by the back end, calibrated with each other and combined to generate the final grid map of the lane area. In other words, the final grid map of the lane area can be generated centrally by multiple vehicles. In this way, other error influences such as single sensor errors and driving style can be compensated. In addition, the back end can also make the final grid map of the lane area available to the vehicle again. In addition, the final grid map or digital road map of the lane area can be generated simply, quickly, continuously and cost-effectively. In addition, the back end can also send the final grid map back to the lane mapping device.

[0034] The backend is a centralized or decentralized computing unit that can exchange data with the individual lane mapping devices. This data exchange can take place in particular via the Internet or a similar network. The backend can be designed, for example, as a server that processes and stores the grid maps of multiple vehicles. This eliminates the need for an evaluation for each vehicle and allows the final grid map or digital road map to be created more quickly and reliably from multiple vehicles.

[0035] A further aspect of the invention relates to a vehicle with a lane mapping device as described above and below.

[0036] The vehicle is, for example, a motor vehicle, such as a car, bus or truck, or also an air vehicle, such as a helicopter or airplane, or, for example, a bicycle.

[0037] Another aspect of the present invention relates to a method for producing a lane map. The method comprises the following steps:

[0038] - detecting image data by means of a camera unit;

[0039] - determining, by means of a control unit, a lane area in the detected image data;

[0040] - converting the detected image data including the determined lane area into a bird's eye view and projecting the converted image data onto a plane at the height of the lane surface;

[0041] - adding lane areas to a grid map of said plane;

[0042] - calibrating and combining at least two grid maps originating from two different driving processes with each other; and - generating a final grid map based on the calibrated and combined grid maps.

[0043] It should be noted that the step of the method can also be implemented in another order or simultaneously. In addition, longer time intervals may also be arranged between each single step.

[0044] A further aspect of the invention relates to a program unit which, if executed on a control unit of a lane mapping device, instructs the lane mapping device to carry out the method described above and below.

[0045] Another aspect of the invention relates to a computer-readable medium on which a program element is stored which, if executed on a control unit of a lane mapping device, instructs the lane mapping device to execute the method described above and below.

[0046] Further features, advantages and application possibilities of the invention can be gathered from the following description of exemplary embodiments and the drawings.

[0047] The accompanying drawings are schematic diagrams and are not drawn to scale. If the same reference numerals are used in the following description of the drawings, it is indicated that these reference numerals refer to the same or similar elements. BRIEF DESCRIPTION OF THE DRAWINGS

[0048] Figure 1 A block diagram of a lane map generating device according to an embodiment of the present invention is shown.

[0049] Figure 2 A block diagram of a lane map making device and a backend for lane map making according to an embodiment of the present invention is shown.

[0050] Figure 3 Displays image data detected by the camera unit.

[0051] Figure 4 The final grid map of the lane area according to one embodiment of the present invention is shown.

[0052] Figure 5 A mapped lane area is shown with closed areas not mapped as lane areas according to an embodiment of the present invention.

[0053] Figure 6 The outer contour of a mapped lane area according to an embodiment of the present invention is shown.

[0054] Figure 7 A vehicle with a lane map making device according to an embodiment of the present invention is shown.

[0055] Figure 8 A flow chart of a method for lane map generation according to an embodiment of the present invention is shown.

[0056] Fig. 9 The final grid map of the lane area according to one embodiment of the present invention is shown.

[0057] Fig.10a A mapped lane area is shown with closed areas not mapped as lane areas according to an embodiment of the present invention.

[0058] Fig.10b The outer contour of a mapped lane area according to an embodiment of the present invention is shown.

[0059] Fig.11 A vehicle with a lane map making device according to an embodiment of the present invention is shown.

[0060] Fig.12 A flow chart of a method for lane map generation according to an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0061] Figure 1A lane map making device 1 is shown, which comprises a control unit 10 and a camera unit 20. Furthermore, the lane map making device 1 may optionally comprise a transmitting / receiving unit 30 and a positioning unit 40. Here, the camera unit 20 detects image data from the vehicle surroundings. The image data may be evaluated by the control unit 10 to determine the lane areas therein. The control unit 10 may use an artificial neural network to determine the lane areas. In other words, the control unit 10 may determine which image parts in the image data are lane areas and which are not lane areas. Furthermore, the control unit 10 may convert the image data, in particular the determined lane areas, into a bird's-eye view by means of inverse perspective imaging and project it onto a plane. It should be noted that the bird's-eye view may be a top view, but the top view may also be at an angle of, for example, ±20 degrees relative to the vertical direction. Furthermore, it should also be noted that the plane onto which the image data is projected should preferably correspond to the road surface in a sufficiently approximate manner. Furthermore, the control unit 10 may correct the image data by means of depth information or height information in the image data, so that the lane areas can be distinguished from buildings and walls. For the depth information or height information, the control unit 10 can take into account image data from a stereo camera, a lidar sensor, a radar sensor or an ultrasonic sensor. The control unit 10 can add the determined lane area to the grid map of the plane after conversion, thereby mapping the lane. The lane map generating device 1 can repeat this process in several driving processes and along several driving directions. Finally, the control unit 10 can calibrate and combine the grid maps generated from several different driving processes to generate a final grid map of the lane area. In particular, the calibration and combination can be performed with the help of several unique features in the grid map, such as traffic islands, intersections, turning curves and roundabouts. In addition, the control unit 10 can determine a probability value for the correct identification of a lane area, and if the probability value exceeds a predetermined threshold, the corresponding lane area can be marked as a lane in the grid map. As an alternative or additional measure, the control unit 10 can also determine the outer contour of the determined lane area in the final grid map of the lane area. In addition, the outer contour can also be represented and smoothed by a geometric curve. Since only the boundary between the lane area and the non-lane area needs to be saved, the file size of the final grid map can be reduced.

[0062] The positioning unit 40 can determine the position of the lane map making device 1 (i.e. its own position), the position of the determined lane area or the position of a point on the grid map relative to the global coordinate system. The positioning unit 40 can, for example, have a GPS module or a similar satellite navigation module. As an alternative or additional measure, the positioning unit 40 can also be positioned by mobile radio. Positioning can also be based on a camera device.

[0063] Furthermore, the lane map generating device 1 can transmit or send the generated grid map with the determined lane areas to a backend by means of a transmitting / receiving unit 30. The backend can then align several grid maps of different vehicles with one another and combine them to generate a final grid map of the lane areas. For this purpose, the transmitting / receiving unit 30 can in particular have a wireless data transmission device, for example a wireless local area network (WLAN) or a mobile radio. Furthermore, the backend can be connected to the lane map generating device 1 via the Internet or a similar network.

[0064] Figure 2 A block diagram of a lane map making device 1 and a back end 2 is shown. The lane map making device 1 and the back end 2 are connected to each other via an at least partially wireless data connection device, for example via a mobile radio or a wireless local area network (WLAN). Here, the lane map making device 1 can send, transmit or pass the generated grid map to the back end 2. The back end 2 can receive grid maps from several different lane map making devices 1 of different vehicles. The back end 2 can then calibrate and combine these several grid maps with each other to generate a final grid map of the lane area. In other words, the back end 2 can generate a map with lane areas from the results of several lane map making devices 1. It should be noted that the back end 2 can also send the final grid map to the lane map making device 1 or a vehicle with a lane map making device 1, so that the vehicle always has the current and latest final grid map of the lane area.

[0065] Figure 3 The image data of the camera unit are shown as an example. Figure 3 In the example of , these image data show the lane direction in front of the vehicle. The control unit is configured to analyze the image data detected by the camera unit and determine or recognize the lane area 12. The recognized lane area 12 is Figure 3 . Other image components, such as another vehicle, a traffic sign or the edge of the road, are identified as non-lane areas. The determined lane area 12 can then be converted into a bird's-eye view by means of inverse perspective imaging and projected onto a plane so that it can be added to a grid map.

[0066] Figure 4 The final grid map of the lane area is shown. Fig.11 The final grid area of ​​the lane Fig.11 A path traveled by the lane mapping device is shown in a bird's eye view or a top view and has the determined lane areas. The determined lane areas are shown in dark colors, while bright areas are not lane areas. In addition, the final grid map of the lane areas can be Fig.11Unique areas such as a roundabout or a traffic island are identified in the grid. In addition, it can be seen that some lane areas appear darker than other lane areas because they are already included in several grid maps or have a higher correct recognition probability value. It can be seen that multiple final grid maps of lane areas can be generated. Fig.11 , these multiple final grid maps are then merged to form a road map or a map with lane areas.

[0067] Figure 5 A mapped area or a grid map of an area is shown. In addition, determined lane areas 12 and non-lane areas 13 can also be identified. The non-lane area 13 can be, for example, a traffic island where vehicles cannot travel. If the determined lane areas 12 exceed a predetermined threshold for correct identification, the determined lane areas can be adopted into the final grid map of the lane areas. In addition, the determined lane areas 12 can also be subsequently processed so that even smaller interruptions determined as non-lane areas are also adopted into the lane areas. In other words, if the determination of the lane areas 12 involves a sufficiently small area, error correction can be performed for the determination of these lane areas. For example, the lane map making device can identify an area of ​​10*10 cm as a non-lane area 13, but it can be considered that if this area is in the middle of a lane area 12, then this area is also part of the lane area 12.

[0068] Figure 6 A cutout of a mapped area or a final grid map is shown, wherein an outer contour 110 has been determined. The outer contour 110 depicts the boundary between lane areas and non-lane areas. By determining the outer contour 110, storage space can be saved because less data must be stored. In addition, the calculation unit can also calculate geometric curves, such as B-splines, which draw the smoothed outer contour. Since the contour is drawn using geometric curves instead of a large number of single points, more storage space can be saved.

[0069] Figure 7 A vehicle 3 is shown with a lane mapping device 1. The lane mapping device 1 can also use sensors already present in the vehicle 3, such as a camera, a GPS module, a gyroscope, a lidar sensor, a radar sensor or an ultrasonic sensor.

[0070] Figure 8A flow chart of a method for lane map production is shown. In step S1, image data are detected by a camera unit, for example, of a stereo camera device. In step S2, the lane area is determined in the detected image data, which can be implemented, for example, with the aid of an artificial neural network. In addition, the lane area can also be corrected by height information. In step S3, the detected image data, in particular the determined lane area, is converted into a bird's-eye view (top view) and projected onto a plane at the height of the lane surface. As a result, image data detected at a specific angle are converted into a two-dimensional plane (2D) top view. In step S4, the determined lane area is added to a grid map in the plane. In step S5, the control unit calibrates and combines at least two, in particular more, grid maps that have been generated during different driving processes, thereby generating a final grid map in step S6.

[0071] Fig. 9 The final grid map of the lane area is shown. Fig.11 The final grid area of ​​the lane Fig.11 A route traveled by the lane mapping device is shown in a bird's eye view or a top view and includes the determined lane areas. The determined lane areas are shown in dark colors, while light areas are not lane areas. In addition, the lane areas can be marked in the final grid map. Fig.11 Unique areas such as a roundabout or a traffic island are identified in the grid. In addition, it can be seen that some lane areas are represented darker than other lane areas because they are already included in several grid maps or have a higher correct recognition probability value. It can be seen that multiple final grid maps of lane areas can be generated. Fig.11 , these multiple final grid maps are then merged to form a road map or a map with lane areas.

[0072] Fig.10a A mapped area or a grid map of an area is shown. In addition, determined lane areas 12 and non-lane areas 13 can also be identified. The non-lane area 13 can be, for example, a traffic island where vehicles cannot travel. If the determined lane area 12 exceeds a predetermined threshold for correct identification, the determined lane area can be adopted into the final grid map of the lane area. In addition, the determined lane area 12 can also be subsequently processed so that even smaller interruptions determined as non-lane areas are also adopted into the lane area. In other words, if the determination of the lane area 12 involves a sufficiently small area, error correction can be performed for the determination of these lane areas. For example, the lane map making device can identify an area of ​​10*10 cm as a non-lane area 13, but it can be considered that if this area is in the middle of a lane area 12, then this area is also part of the lane area 12.

[0073] Fig.10b A cutout of a mapped area or a final grid map is shown, wherein an outer contour 110 has been determined. The outer contour 110 depicts the boundary between lane areas and non-lane areas. By determining the outer contour 110, storage space can be saved because less data must be stored. In addition, the calculation unit can also calculate geometric curves, such as B-splines, which give a smoothed outer contour. Since the contour is drawn using geometric curves instead of a large number of single points, more storage space can be saved.

[0074] Fig.11 A vehicle 3 is shown with a lane mapping device 1. The lane mapping device 1 can also use sensors already present in the vehicle 3, such as a camera, a GPS module, a gyroscope, a lidar sensor, a radar sensor or an ultrasonic sensor.

[0075] Fig.12 A flow chart of a method for lane map production is shown. In step S1, image data are detected by a camera unit, for example, of a stereo camera device. In step S2, the lane area is determined in the detected image data, which can be implemented, for example, with the aid of an artificial neural network. In addition, the lane area can be corrected by height information. In step S3, the detected image data, in particular the determined lane area, is converted into a bird's-eye view (top view) and projected onto a plane at the height of the lane surface. As a result, image data detected at a specific angle are converted into a two-dimensional plane (2D) top view. In step S4, the determined lane area is added to the grid map of the plane. In step S5, the control unit calibrates and combines at least two, in particular more, grid maps that have been generated during different driving processes, thereby generating a final grid map in step S6.

Claims

1. A lane map making device (1) for a vehicle (3), the vehicle map making device comprising: - a camera unit (20) for detecting image data; as well as - a control unit (10), The control unit (10) is configured to determine a lane area (12) in image data detected by the camera unit (20), The control unit (10) is configured to convert image data including the determined lane area (12) into a bird's-eye view by means of reverse perspective imaging and project the converted lane area onto a plane; The control unit (10) is further configured to add the projected and transformed lane area (12) to the grid map in the plane, and The control unit (10) is configured to calibrate and combine at least two grid maps originating from two different driving processes with each other according to features contained in the grid maps to generate a final grid map (11) of the lane area; wherein the at least two grid maps used to generate the final grid map (11) are derived from image data of at least two different driving directions; wherein the calibration and combination of the grid maps, including deformation, is performed, and the image data is converted into a bird's-eye view before constructing the grid map; Therein, the final grid map includes projected, transformed, calibrated and combined lane areas, which are used as lane representations in the final grid map.

2. The lane map making device (1) according to claim 1, in, The projection plane is at the height of the lane surface. The control unit (10) is further configured to correct a lane region (12) determined in the image data using the height information.

3. The lane map making device (1) according to claim 1 or 2, in, The control unit (10) is configured to determine a lane region (12) in the image data using an artificial neural network.

4. The lane map making device (1) according to claim 1 or 2, in, The control unit (10) is configured to determine a probability value of correct identification for each determined lane area (12), and if the probability value exceeds a predetermined threshold, mark the corresponding lane area (12) as a lane.

5. The lane map making device (1) according to claim 1 or 2, further comprising: A positioning unit (40) is configured to determine the position of the lane map making device (1), the position of the determined lane area (12), or the position of a point in a grid map.

6. The lane map making device (1) according to claim 1 or 2, in, The control unit (10) is configured to determine an outer contour (110) of an area in a grid map that is determined as a lane area (12) or to convert the outer contour into a geometric curve.

7. The lane map making device (1) according to claim 1 or 2, further comprising: A sending / receiving unit (30) is configured to transmit the grid map to the back end (2).

8. The lane map making device (1) according to claim 1 or 2, in, The camera unit (20) comprises a stereo camera device.

9. A backend (2) for lane map production, the backend comprising: A transmitting / receiving unit (21) configured to receive a grid map from a lane mapping device according to any one of the preceding claims, The back end (2) is configured to receive at least two grid maps from at least two vehicles (3), and calibrate and combine the grid maps to generate a final grid map (11) of the lane area.

10. A vehicle (3) having a lane map generating device (1) according to any one of claims 1 to 8.

11. A method for lane map production: - detecting (S1) image data by means of a camera unit; - determining (S2) a lane area in the detected image data by a control unit; - converting (S3) the detected image data including the determined lane area into a bird's eye view and projecting the converted image data onto a plane at the height of the lane surface; - adding (S4) the projected and transformed lane areas to the grid map in said plane; - aligning and combining at least two grid maps originating from two different driving processes with each other (S5); and - generating (S6) a final grid map based on the calibrated and combined grid maps; wherein at least two grid maps used to generate the final grid map (11) are derived from image data of at least two different driving directions; calibration and combination of the grid maps, including deformation, are performed, and the image data are converted into a bird's-eye view before constructing the grid maps; The final grid map includes converted, projected, calibrated and combined lane areas, which are used as lane representations in the final grid map. 12 . A computer program element which, if it is run on a control unit of a lane mapping device, instructs the lane mapping device to carry out the method according to claim 11 .

13. A computer-readable storage medium on which a computer program element according to claim 12 is stored.

Images