Method for creating a three-dimensional road map

Abstract

The present invention provides a method for creating a three-dimensional road map (44), comprising the steps of: a) providing a vehicle (10) which comprises a measuring device (12), which has an inertial measuring system, a GNSS antenna (20) and a camera device (13) which is time-synchronous with the inertial measuring system and the GNSS antenna (20); b) recording the vehicle orientation using the inertial measurement system, the vehicle position using the GNSS antenna (20), and images using the camera device (13) while driving along a lane (60) such that at least some of the images show an outer contact point (32) of a tire (16, 18) and a lane edge (36); c) determining a distance (42) between the outer contact point (32) of the tire (16, 18) and a lane edge point (34) of the lane edge (36) from vehicle orientation values that are each recorded at the same time in step b) and one of the images; d) calculating an absolute position (52) of the lane edge point (34) defined in step c) from the vehicle position and the distance (42) determined in step c) for the point in time;) repeating steps c) and d) for a plurality of temporally successive absolute positions (52); f) determining a first lane edge line (46) of the lane by connecting or interpolating the plurality of temporally successive absolute positions (52); and g) repeating steps b) to e) for a second lane edge (36) opposite the first lane edge (36) and then h) determining a second lane edge line (48) of the lane (60) opposite the first lane edge (36) by connecting or interpolating the plurality of temporally successive absolute positions (54) of the second lane edge (36); and i) outputting the three-dimensional road map (44), comprising the first lane edge line (46) and the second lane edge line (48).

Description

[0001] Method for creating a three-dimensional road map

[0002] The present invention relates to a method for creating a three-dimensional road map, a three-dimensional road map created using such a method, and a method for using such a three-dimensional road map to control an autonomous vehicle.

[0003] The present invention is based on known techniques for surveying roads. These involve manually surveying roads, driving them with cameras, or scanning them with lasers.

[0004] The disadvantage of the known solutions is that they are either inaccurate, particularly complex to measure, or do not capture all relevant parameters for applications such as autonomous driving.

[0005] High-resolution three-dimensional road maps are regularly required, especially for applications in the field of autonomous driving. High-resolution maps can better consider important safety aspects in autonomous driving and contribute to the accurate detection of otherwise ambiguous driving situations. High-resolution three-dimensional maps are also advantageous for vehicle control and better route planning.

[0006] The object of the present invention is to at least partially remedy the disadvantages described above in a cost-effective and simple manner. In particular, the object of the present invention is to create three-dimensional road maps in a simple manner.

[0007] It is a further object of the invention to provide a three-dimensional road map that improves accuracy of control of an autonomous vehicle.

[0008] It is a further object of the invention to provide a three-dimensional road map that can be used to evaluate the driving behavior of a vehicle. It is a further object of the invention to provide a method that can guarantee the accuracy of a three-dimensional road map by means of validation.

[0009] It is a further object of the invention to provide a three-dimensional road map that enables the location of a vehicle in real time.

[0010] The above objects are achieved by a method having the features of claim 1, a road map having the features of claim 13, and methods having the features of claims 14 and 15. Further features and details of the invention emerge from the subclaims, the description, and the drawings. Features and details described in connection with the method according to the invention naturally also apply in connection with the three-dimensional road map according to the invention, and vice versa, so that with regard to the disclosure of the individual aspects of the invention, reference is always made to each other.

[0011] According to the invention, a method is intended to enable the creation of a three-dimensional road map. Such a method is characterized by the following steps: a) providing a vehicle which comprises a measuring device comprising an inertial measuring system, a GNSS antenna, and a camera device time-synchronized with the inertial measuring system and the GNSS antenna; b) recording the vehicle orientation with the inertial measuring system, the vehicle position with the GNSS antenna, and images with the camera device while driving along a lane, such that at least some of the images show an outer contact point of a tire and a lane edge; c) determining a distance between the outer contact point of the tire and a lane edge point of the lane edge from values ​​of the vehicle orientation recorded at the same time in step b) and one of the images;d) Calculating an absolute position of the lane edge point defined in step c) from the vehicle position and the distance for the time determined in step c); e) Repeating steps c) and d) for a plurality of temporally successive absolute positions; f) Determining a first lane edge line of the lane by connecting or interpolating the plurality of temporally successive absolute positions; and g) Repeating steps b) to e) for a second lane edge opposite the first lane edge and then h) Determining a second lane edge line of the lane opposite the first lane edge by connecting or interpolating the plurality of temporally successive absolute positions of the second lane edge; and i) Outputting the three-dimensional road map comprising the first lane edge line and the second lane edge line.

[0012] The core idea of ​​a method according to the invention is that it is easily possible to measure lanes and in particular lane edge lines with high accuracy and reliability using a vehicle that requires a camera device, an inertial measuring system and a GNSS antenna for measurement.

[0013] A vehicle equipped with an inertial measurement system and a GNSS antenna is used to record the map data. The vehicle is therefore a detection vehicle. The inertial measurement unit (IMU) comprises a combination of several inertial sensors such as acceleration sensors and angular rate sensors. To record the six possible kinematic degrees of freedom, the inertial measurement system comprises three orthogonally positioned acceleration sensors (translation sensors) for recording the translational movement in the x-, y-, and z-axes, and three orthogonally mounted gyroscopic sensors for recording rotating (circular) movements in the x-, y-, and z-axes. An inertial measurement unit provides three linear acceleration values ​​for the translational movement and three angular velocities for the angular rates.In an inertial navigation system (INS), the linear acceleration measurements of the inertial measurement system, after compensation for gravitational acceleration, are used to determine the linear velocity by integration, and the spatial position relative to a reference point is then integrated again. The integration of the three angular velocities relative to a reference point provides the spatial orientation. Additional magnetometers can be integrated to determine the integration constant, improve accuracy, and correct for the zero-point and long-term drift of the aforementioned sensors.

[0014] The GNSS antenna is used to determine the vehicle's three-dimensional position on Earth. GNSS is a collective term for the use of existing and future global satellite systems such as NAVSTAR GPS, GLONASS, Galileo, and Beidou.

[0015] The method can in particular be computer-implemented.

[0016] The measuring device can have one or more cameras, in particular three cameras, which each record videos and / or individual images taken at short intervals one after the other to carry out the method. The sampling rate is preferably at least 20 images per second, more preferably at least 30 images per second, and most preferably at least 30 images per second. However, it is also possible to carry out the method with significantly lower sampling rates. Preferably, the images taken in step b) depict the roadway at least every 2 meters. The cameras are each time synchronized with each other, with the inertial measurement system, and with the GNSS antenna. In principle, however, the different systems have different sampling rates. For example, the GNSS antenna can have a sampling rate of 100 Hz.The cameras can, for example, capture images at 20 fps (20 Hz) during the process. The inertial measurement system can, for example, capture data at a frequency of 30 Hz. Therefore, the times of the captured images and measured values ​​generally do not match exactly. To increase accuracy, a data synchronization step can be performed. For this purpose, data points from the GNSS antenna and the inertial measurement system can be calculated for each individual image, adjusted to the time of capture of the individual image. The calculation can be performed by interpolating the measured data, in particular by linear interpolation of the measured values ​​from the GNSS antenna and the inertial measurement system around this measurement time.The data synchronization step performed in this way allows the data from the GNSS antenna and the inertial measurement system to be synchronized with those from the images, i.e., data whose timestamps match the timestamps of the images is generated. For this purpose, each individual camera preferably has synchronization channels with the following parameters: image acquisition time, latitude of the image position, longitude of the image position, elevation of the image position, and orientation of the image position.

[0017] The cameras are preferably arranged on the left and / or right side of the vehicle, and in the case of two or three cameras, possibly also in the middle, with the image section of the camera(s) being selected such that the left and / or right front wheel and, if applicable, the front of the vehicle are located in the entire image field of the cameras. With the camera device in combination with the previous camera calibration, and using the developed distance calculation logic, it is particularly possible to determine relative, time-synchronized distances between the wheel contact points (left and right) and / or the outer contact points of the tires, as well as a known or determined point of intersection of the vehicle center plane in the vehicle's longitudinal direction with a plane through the vehicle's front axle and the road plane to the lane edges. The wheel contact points are each shifted by half a tire width compared to the outer contact points of the tires.The distance can also be zero, especially when crossing a lane edge line.

[0018] Preferably, in step c), the calibration zero point can also be used to determine a distance between the outer contact point of the tire and a lane edge point. The calibration zero point is a fixed point of known position, in particular on a calibration tape. The calibration zero point can be used to increase calibration accuracy. The method makes it possible to determine the exact geographical position of the vehicle at each image acquisition time. For this purpose, in addition to the acquisition time and the image number, the longitude, latitude, altitude, and the accuracy of the GNSS measurement can be saved. The method steps for recording the measurement data produce consistent and synchronized measurement data, which can be further processed automatically.

[0019] Steps c) and d) can be performed by determining the exact geographical position of the vehicle at each frame point in a survey video and, using a three-dimensional offset in Cartesian coordinates between the arrangement of the GNSS antenna and the outer tire contact point, determining the outer tire contact point. For this purpose, in addition to the recording time and the image number, the longitude, latitude, altitude, and the accuracy of the GNSS measurement are saved. This provides consistent and synchronized measurement data that can be further processed automatically. If multiple tires and / or multiple outer tire contact points are involved in the process, separate offsets must be considered for each tire.

[0020] Preferably, a synchronous measurement file of the image positions can be saved for each video file.

[0021] Determining a first and / or second lane edge by connecting or interpolating the temporally successive absolute positions can be achieved in particular by first performing a nearest neighbor search from a lane edge point to the next lane edge point starting from the time the image was taken, then determining time differences to the next lane edge point starting from the time the image was taken, subsequently selecting the minimum value and finally performing a second nearest neighbor search to the second nearest lane edge point starting from the time the image was taken. By connecting or interpolating the lane edge points, each of the lane edge lines is determined. The images of the first lane edge and the images of the second lane edge can be recorded one after the other or, if a suitable camera device is available, also simultaneously.The lane edge points are thus ordered according to the time of recording and then connected or interpolated in this order.

[0022] There are further advantages if the recording in step b) takes place at synchronous intervals.

[0023] This allows the recorded data to be checked for validity by easily identifying missing or additional data. Furthermore, it makes it possible to ensure consistent accuracy of the lane boundary lines on the three-dimensional road map, especially when the lane is driven at a substantially constant speed in step b).

[0024] Further advantages are achieved if the camera device has a first camera on the left side of the vehicle and a second camera on the right side of the vehicle, wherein the first camera is oriented such that its field of view includes an outer contact point of a left tire and the second camera is oriented such that its field of view includes an outer contact point of a right tire. In particular, the camera device can also have a third camera whose field of view shows the front of the vehicle and the lane.

[0025] With two cameras aimed at different tires, images of the first and second edges can be captured in parallel, i.e., simultaneously or essentially simultaneously. A third, central camera facilitates data analysis, as it can create a complete image of the road. Here, too, a calibration procedure can be applied to synchronize all relevant data.

[0026] Further advantages are achieved if the method further comprises step a1) after step a): a1) Calibrating the camera device with a calibration device arranged on the floor of the lane. The calibration of the area to be measured on the floor of the lane can be performed, for example, using a calibration tape with a centimeter scale, which is arranged on the lane. For this purpose, the following calibration steps are performed on an image taken by the camera device and showing the calibration tape:

[0027] • Right-angled alignment of the calibration tape, especially with a laser or laser measuring device

[0028] • Setting calibration points on the calibration tape

[0029] • Save the calibration file

[0030] • Validation of the calibration with a distance measurement

[0031] First, calibration points are set on the centimeter measurements. The calibration then takes place along a line in the image. The calibration points are placed on the calibration line in the image at the crosshairs on the calibration tape. The first calibration point is the zero point of the calibration, and the subsequent calibration points are set manually and / or automatically in ascending order. This allows the ratio of image pixels to actual distance measurement to be determined.

[0032] For example, if there are a total of 69 pixels in the outer edge area in a distance segment of 1 cm, the measurement can be accurate to 0.15 cm.

[0033] Calibration accuracy can preferably be performed before and after the survey, i.e., steps b) and c), to ensure that the distance measurement is correct. This step thus increases the accuracy of the road map.

[0034] Preferably, the calibration device can comprise a measuring device with calibration markings arranged on the measuring device. As an alternative to the simple form of a centimeter ruler as the calibration device, a measuring field can also be provided that arranges measuring devices in two dimensions on the lane surface.

[0035] Further advantages are achieved if the lane edge is defined in step c) using an edge detection algorithm. The edge detection algorithm is used to separate flat areas in the image from one another if they differ sufficiently in color or gray value, brightness, or texture along straight or curved lines. The edge detection algorithm is used to detect the transitions between these areas and mark them as edges. At the same time, however, a single, homogeneous area should be recognized as such and not divided into two areas by an edge. To do this, the color value gradient at each individual pixel of an image can be calculated by examining the area surrounding the point. This process is carried out by discretely convolving the image with a convolution matrix, the edge operator. The edge operator defines the size of the environment to be examined and the weighting of its individual pixels in the calculation.The edge operator determines an average gradient value for the central pixel from the surrounding area. If this operation is performed for all pixels in the image, an edge image can be compiled from the resulting gradient matrix. In this image, the edges between homogeneous areas stand out, since these areas have a comparatively large gradient of color values. Other edge detection methods are also possible.

[0036] Further advantages are achieved if the method further comprises the steps:

[0037] Calculating a center position as the geographical center of an absolute position of the first lane edge line and an absolute position of the second lane edge line opposite the absolute position of the first lane edge; and

[0038] Determining a lane centerline by connecting or interpolating the temporally successive center positions.

[0039] The lane centerline is another aspect of the three-dimensional road map that improves its use for autonomous driving. A deviation from an ideal driving line or the initiation of an overtaking maneuver can be more easily detected by autonomous vehicles if a lane centerline is present. The centerline is calculated from the two left and right lane edges. The centerline of each lane serves as a reference and is preferably interpolated at a constant distance. The resolution of the centerline can be varied as desired. This often also depends on the lane characteristics.Further advantages can be achieved if the method described last further comprises the step of calculating lane segments as a polygon from two consecutive center positions and those two absolute positions of the first lane edge line and the second lane edge line with which the two center positions were calculated, wherein a lane is defined as the sum of the lane segments.

[0040] A lane segment thus consists of a total of six control points. These control points each contain two consecutive GNSS points of the left and right lane margins, as well as the lane centerline. This geographical information can be incorporated into the three-dimensional road map as a geometric model. With this information, the three-dimensional road map can include, in particular, the following elements:

[0041] • Polygon line of the left lane edge

[0042] • Polygon line of the right lane edge

[0043] • Polygon line of the road center

[0044] • Lane edge points of the left road edge

[0045] • Lane edge points of the right road edge

[0046] • Multipoint lane center

[0047] • Polygons of the lane segments

[0048] • Lane boundary polygons

[0049] Lane segments also allow for more precise control and / or validation of autonomous vehicle behavior. For example, deviations from the expected driving behavior of other road users can be more easily detected if the vehicle accesses a three-dimensional road map containing polygon segments. In particular, unusual braking and acceleration behavior can be detected more easily and accurately. Thus, there are three lines per lane: the left and right lane edge lines, and the lane center line. For these three lines, the following parameters can be calculated for each segment:

[0050] • Lane length

[0051] • Lane width

[0052] • Lane line type

[0053] • Lane line color

[0054] • Lane direction

[0055] • Lane curvature

[0056] • Lane gradient

[0057] • Lane cross slope

[0058] • Lane line quality

[0059] The lane line type, lane line color, and lane line quality can be derived from the images or the edge detection algorithm. The remaining parameters are calculated as shown in the table below:

[0060] Further advantages are achieved if the method further comprises the step:

[0061] Calculating a lane center zone as an area located between the first lane edge line and the second lane edge line, each separated by a certain distance value from the first lane edge line and the second lane edge line. Lane center zones also allow for more precise control of autonomous vehicles. For example, deviations from the expected driving behavior of other road users can be more easily detected if the vehicle accesses a three-dimensional road map that includes lane center zones. In particular, unusual steering behavior can be detected more easily and accurately. Lane keeping assistants can also react more quickly and reliably in unusual driving situations.

[0062] Further advantages are achieved if the latter method further comprises the step:

[0063] Calculating a lane edge area as the difference between the area spanned by the first lane edge line and the second lane edge line and the lane center area.

[0064] This creates a three-dimensional map that defines an edge area and a center area. This further enhances the benefits provided by the lane center area.

[0065] Further advantages are achieved if the method further comprises the step of linking static or dynamic lane parameters to the lane.

[0066] Static lane parameters that can be provided include, in particular: a lane direction, a lane width, a lane gradient, a lane cross slope, a lane length, a lane curvature of the left lane edge line, a lane curvature of the right lane edge line and / or a lane curvature of the lane center line.

[0067] In particular, relative distances from the vehicle to the lane edge can be specified as dynamic lane parameters. These include the shortest distance to the left lane edge line, the right lane edge line, and the lane center line, the corresponding current distances, the corresponding average distances, and the corresponding orthogonal distances. An angle between the direction of travel and the lane center line or between the direction of travel and one of the lane edge lines can also be specified as a dynamic driving parameter. Here, too, a current angle, an average angle, or a maximum angle can be specified as a dynamic lane parameter.

[0068] Further advantages are achieved if the method further comprises the steps:

[0069] Providing a three-dimensional road map obtained according to one of the preceding claims,

[0070] Crossing the edge of the lane several times with the vehicle and

[0071] Recording the vehicle orientation with the inertial measurement system, the vehicle position with the GNSS antenna and images with the camera device such that at least some of the images show an outer contact point of a tire on the lane edge;

[0072] Determining lane edge points with optimized accuracy from vehicle alignment values ​​recorded at the same time and the images showing the outer contact point of the tire on the lane edge; and

[0073] Validate and / or correct the three-dimensional road map with the accuracy-optimized lane edge points.

[0074] HD map validation provides a measure of the accuracy of the measured route and thus of the previously created three-dimensional road map. The vehicle is used as the validation vehicle for this purpose. Validation is carried out with highly precise recording of the vehicle position, for example with an accuracy of less than ±5 cm, preferably less than ±3 cm and particularly preferably less than ±1.5 cm. An inherent measurement error of the camera during distance recordings is reduced by the fact that the outer contact point of the tire, which is known relative to the GNSS antenna, is also the measurement point. The temporal measurement grid should be as short as possible. In particular, the images should be recorded at intervals of 50 ms or less, preferably 20 ms or less and particularly preferably at intervals of 10 ms or less. For each of the methods, recording the images can also mean recording a video.

[0075] Lane edges are deliberately crossed. The point of crossing is determined from the images or, preferably, the video and compared with the position determination results. From this, a relative distance can be calculated between the lane edge line of the previously output three-dimensional road map and the newly calculated lane edge line values. If a lane center line is present, validation can also be performed for the lane center line.

[0076] According to a second aspect, the present invention provides a three-dimensional road map obtained by a method according to any one of the preceding claims.

[0077] The three-dimensional road map comprises at least the first and second lane edge lines in three-dimensional coordinates and with a high accuracy of preferably ±3 cm or less, particularly preferably ±1.5 cm or less, and may further contain the lane center line and / or the lane segments and / or the lane center area and / or the lane edge area.

[0078] According to a third aspect, the invention provides a method for using a three-dimensional road map to control an autonomous vehicle.

[0079] For this purpose, the three-dimensional road map can be integrated into the vehicle's control system, or the vehicle can access the three-dimensional road map for control via a communications network. The lane edge lines and / or the lane center line and / or the lane segments and / or the lane center area and / or the lane edge area can be used to execute driving maneuvers and to assess the driving behavior of other road users and adapt the control of the vehicle accordingly. According to the third aspect of the invention, an autonomous vehicle can, in addition to accessing sensor data relating to the current driving situation, also use the data integrated in the three-dimensional road map to assess road conditions and driving situations and thereby improve its own driving behavior.

[0080] According to a fourth aspect, the invention provides a method for using a three-dimensional road map according to claim 13 for detecting crossing of a lane edge line, in particular in a racing sports competition, comprising the steps of: providing a three-dimensional road map, in particular according to claim 13;

[0081] Moving a detection vehicle with a GNSS device on a road area covered by the three-dimensional road map;

[0082] Recording a vehicle position using the GNSS device;

[0083] Comparing the vehicle position with coordinates of the first lane edge line and / or the second lane edge line; and

[0084] Detecting crossing of a lane edge line by comparison.

[0085] For this purpose, the detection vehicle preferably also has a GNSS measuring device, and its dimensions, i.e., its maximum extent, are more preferably known by the GNSS measuring device. A vehicle trajectory is created by recording the vehicle position over a period of time. The maximum vehicle dimensions can be added using offsets. This also makes it possible to detect when the detection vehicle, which is currently participating in a racing competition, for example, drives over a lane edge with only one tire or part of a tire.

[0086] Further advantages, features, and details of the invention will become apparent from the following description, in which exemplary embodiments of the invention are described in detail with reference to the drawings. They show schematically:

[0087] Fig. 1 shows a vehicle with a measuring device for carrying out a method according to the invention,

[0088] Fig. 2 a calibration device with recorded measured values ​​for calibrating the measuring device,

[0089] Fig. 3 is an image from a road measurement for measuring a distance between the outer contact point of the tire and a lane edge point of the lane edge,

[0090] Fig. 4 a three-dimensional road map with two lane edge lines and one lane center line, Fig. 5 a three-dimensional road map with two lane edge lines, one

[0091] Lane centerline and lane segments,

[0092] Fig. 6 a three-dimensional road map with two lane edge lines, a lane center area and a lane edge area,

[0093] Fig. 7 is a flow chart of a method according to the invention, and

[0094] Fig. 8 shows an alternative flow chart of a method according to the invention.

[0095] Figure 1 schematically shows a vehicle 10 with a measuring device 12, which is suitable for carrying out a method according to the invention for creating a three-dimensional road map. The measuring device 12 comprises a camera device 13 with three time-synchronized cameras 14. A first camera 14 is arranged on the left side of the vehicle 10 and a second camera 14 on the right side of the vehicle 10. The first and second cameras 14 can in particular be arranged in an area between the B-pillar of the vehicle 10 and a front wheel plane 19. The front wheel plane 19 runs perpendicular to the vehicle center plane 15 and parallel to the wheel axis. Particularly preferably, the cameras 14 can be arranged vertically above the tires 16, 18 on the vehicle 10.In such an arrangement, the fields of view of the cameras 14 can run parallel to the axial direction, minimizing perspective distortion of the images in only one direction and consequently increasing the accuracy of the method. The first camera 14 is aligned such that its field of view encompasses an outer contact point of a left tire 16, and the second camera is aligned such that its field of view encompasses an outer contact point of a right tire 18. The cameras 14 are configured as video cameras to enable a sufficiently high image capture rate.

[0096] The measuring device 12 further comprises a GNSS antenna 20, which operates synchronously with the camera device 13. The GNSS antenna comprises a transmitting and receiving device as well as an evaluation device (neither explicitly shown) for evaluating the position data. Time-synchronous operation means that the position data recorded by the GNSS antenna 20 have a time stamp, and the images recorded by the camera device 13 also have a time stamp, and that the time stamps match with each other if the recording times are the same. This allows an assignment between the position data and the images, and a synchronization between the images and the data can be achieved. An inertial measuring system that is also time-synchronous is not explicitly shown.A third camera 14 is located centrally on the vehicle, with its field of view directed forward, so that during operation, its field of view shows the front of the vehicle and the lane. The third camera can thus simplify the analysis.

[0097] The position of the GNSS antenna relative to the outer tire support points is known.

[0098] The vehicle is arranged on a calibration device 22. The calibration device 22 is designed as a calibration belt 24 and has a scale that is arranged transversely to the direction of travel, in the axial direction below the front tires. In this position, calibration of the camera device 13 can be carried out. For a particularly precise vertical arrangement of the calibration belt 24, a laser measuring device 21 is used. With the laser measuring device 21 arranged at one end of the calibration belt 24, the laser beam 17 emanating from it and with it the calibration belt 24 can be arranged particularly precisely parallel to the front wheel plane 19 and perpendicular to the vehicle center plane 15. Since the precise arrangement of the calibration belt 24 affects the calibration of the measuring device 12, the laser measuring device 21 increases the accuracy of the method.

[0099] Figure 2 schematically shows a calibration tape 24. The calibration tape 24 has calibration points 26 marked by circles on a line. For the recording of a camera 14, the calibration points are linked to the pixels of the camera image to calibrate the camera 14. This can be done manually or automatically. For this purpose, the calibration tape 24 is preferably arranged so that its distance measurement runs along a pixel line of the camera 14. The calibration can be performed in the process using software that interprets mouse clicks as the setting of a calibration point. The calibration points can then be used to determine the ratio of pixel distances to the actual distance measurement.

[0100] In an example measurement, there are a total of 69 pixels in the outer edge area of ​​the image within a distance segment of 1 cm. In this example, the measurement can be accurate to 0.015 cm.

[0101] Figure 3 shows an image of a road measurement taken while driving along a lane. The image shows an outer contact point of the right front tire 14.

[0102] For measuring a distance between the outer contact point 32 of the tire 14 and a lane edge point 34 of the lane edge 36. In this example, the lane edge is defined by the lower edge of a curb 38 facing the lane 30. In other cases, the lane edge 36 can be defined by a road marking or in some other way. The lane edge 36 can be determined and / or defined using an edge detection algorithm. The errors from machine edge detection can optionally be compensated for in order to achieve greater accuracy and, in particular, eliminated using human logic. This can compensate for effects that arise, for example, because sunlight makes detection difficult, shadows, leaves, sand, or other objects block the lane markings, or because no lane markings are present.

[0103] To determine a lane edge point 34, a distance 42 between the outer contact point 32 of the tire 14 and a lane edge point 34 of the lane edge 36 is determined from vehicle orientation values ​​and the image recorded at the same time in step b). The known offset between the GNSS antenna 20 and the outer contact point 32 of the tire 14 is taken into account.

[0104] The distance 42 is determined from a connecting line 40 between the outer contact point 32 of the tire 14 and a lane edge point 34, which runs according to the pixel line of the camera used during calibration. An absolute position in GNSS coordinates of the lane edge point is then calculated from the vehicle position and the distance 42 for the time point.

[0105] These steps are repeated for both lane edges. For this purpose, the vehicle 10 can be equipped with only one camera 14, which, for example, exclusively records the right tire 14. In this case, the lane 30 is traveled twice, each time in a different direction, so that absolute positions of both lane edges 36 can be recorded.

[0106] If, however, the camera device has two cameras 14 arranged on different sides of the vehicle 10, it may be sufficient for the lane 30 to be driven on only once and for the images of both lane edges 36 to be recorded simultaneously.

[0107] By connecting or interpolating a plurality of temporally successive absolute positions of the first and second lane edges 36 separately, lane edge lines of the first and second lane edges are determined. In this way,

[0108] Figure 4 shows a three-dimensional road map 44 comprising a first lane edge line 46, a second lane edge line 48, and a lane center line 50. The first and second lane edge lines 46, 48 were determined by interpolating respective absolute positions 52, 54 and therefore represent them in the three-dimensional GNSS coordinate system.

[0109] The lane center line 50 is calculated from the two lane edge lines 46, 48 and / or their absolute positions 52, 54. The lane center line 50 serves as a reference and is interpolated at a constant distance. The resolution of the lane center line 50 can, in principle, be varied as desired, particularly depending on the lane characteristics. In the example shown, the lane center line 50 can be determined by first calculating a center position 56 as the geographical center point of an absolute position 52 of the first lane edge line 44 and an absolute position 54 of the second lane edge line 48 opposite the absolute position 52 of the first lane edge. The lane center line 50 can then be determined by connecting or interpolating the temporally successive center positions 56.

[0110] Fig. 5 shows a three-dimensional road map 44 with lane segments 58. According to the invention, the lane segments 58 are defined as a polygon consisting of two consecutive center positions 56 and the two absolute positions 52 of the first lane edge line 46 and the second lane edge line 48 with which the two center positions 56 were calculated. A lane is defined as the sum of the lane segments 58. The sum of the lane segments 58 thus forms a surface in three-dimensional space. This surface, and thus the lane 60, can be further smoothed by means of a three-dimensional interpolation between the lane segments 58 and / or the absolute positions 52, 54 and center positions 56. A lane segment 58 thus consists of a total of six support points. These support points each contain two consecutive GNSS points of the left and right lane edges, as well as the center of the lane 60.

[0111] Figure 6 shows a three-dimensional road map 44 with two lane edge lines 46, 48, a lane center area 62 and a lane edge area 64, wherein the lane edge area 64 consists of two spatially separated areas.

[0112] The lane center area 62 is an area located between the first lane edge line 46 and the second lane edge line 48, each spaced apart by a distance value from the first lane edge line 46 and the second lane edge line 48. It can be calculated using simple geometric considerations. The lane edge area 64 can be calculated as the difference between the area spanned by the first lane edge line 46 and the second lane edge line 48 and the lane center area 62.

[0113] Fig. 7 shows a flow chart of a method according to the invention.

[0114] In step a), a vehicle is provided which comprises a measuring device comprising an inertial measuring system, a GNSS antenna and a camera device time-synchronized with the inertial measuring system and the GNSS antenna.

[0115] In step b), while driving along the lane, the vehicle orientation is recorded with the inertial measurement system, the vehicle position with the GNSS antenna, and images are captured with the camera system. At least some of the images show the outer contact point of a tire and the edge of the lane.

[0116] In the subsequent step c), the distance between the outer tire contact point and a lane edge point is determined. This distance is determined from vehicle alignment values ​​recorded at the same time in step b) and one of the images.

[0117] In step d), an absolute position of the lane edge point defined in step c) is calculated. The calculation is based on the vehicle position and the distance determined in step c) for the time point.

[0118] Step e) is a repetition step. Here, steps c) and d) are repeated for a plurality of consecutive absolute positions. The number of repetitions depends on the length of the lane to be traversed or the total measurement duration of the measurement unit. In principle, the number of repetitions n can be freely selected.

[0119] In step f), a first lane edge line of the lane is determined. The determination is performed by connecting or interpolating the majority of the temporally consecutive absolute positions.

[0120] Parallel to step f), after step f) or before step f), step g) is performed, in which steps b) to e) are repeated for a second lane edge opposite the first lane edge. This is a one-time repetition, which is typically performed separately if the camera system only records images on one side of the vehicle.

[0121] Subsequently, in step h), a second lane edge line of the lane opposite the first lane edge is determined. The determination is performed by connecting or interpolating the majority of the temporally consecutive absolute positions of the second lane edge.

[0122] Finally, in step i), the three-dimensional road map, which includes at least the first lane edge line and the second lane edge line, is output.

[0123] Figure 8 shows an alternative method according to the invention, which includes all the steps described in Figure 7. In contrast to the method described in Figure 7, steps b), c), d), and e) are performed in parallel for the first and second lane edges to determine the data for determining the first and second lane edge lines. The parallel execution of these steps is possible if the vehicle's camera system is configured to capture images on both sides of the vehicle simultaneously, for example, using two or more cameras.

[0124] The above explanations of the embodiments describe the present invention exclusively by way of examples.

[0125] List of reference symbols

[0126] 10 vehicles

[0127] 12 Measuring device

[0128] 13 Camera setup

[0129] 14 Camera

[0130] 15 Vehicle center plane

[0131] 16 left tire

[0132] 17 Laser beam

[0133] 18 right tire

[0134] 19 front wheel plane

[0135] 20 GNSS antenna

[0136] 21 Laser measuring device

[0137] 22 Calibration device

[0138] 24 calibration tape

[0139] 26 Calibration point

[0140] 30 lanes

[0141] 32 outer support point

[0142] 34 Lane edge point

[0143] 36 Lane edge

[0144] 38 curb

[0145] 40 connecting line

[0146] 42 distance

[0147] 44 three-dimensional road maps

[0148] 46 first lane edge line

[0149] 48 second lane edge line

[0150] 50 lane center line

[0151] 52 Absolute position of the first lane edge line

[0152] 54 Absolute position of the second lane edge line

[0153] 56 Middle position

[0154] 58 lane segment

[0155] 60 lanes

[0156] 62 Lane center area

[0157] 64 Lane edge area

Claims

Claims 1. A method for creating a three-dimensional road map (44), comprising the steps of: a) providing a vehicle (10) which comprises a measuring device (12) comprising an inertial measuring system, a GNSS antenna (20), and a camera device (13) time-synchronized with the inertial measuring system and the GNSS antenna (20); b) recording the vehicle orientation with the inertial measuring system, the vehicle position with the GNSS antenna (20), and images with the camera device (13) while driving along a lane (60), such that at least some of the images show an outer contact point (32) of a tire (16, 18) and a lane edge (36); c) determining a distance (42) between the outer contact point (32) of the tire (16, 18) and a lane edge point (34) of the lane edge (36) from values of the vehicle alignment recorded at the same time in step b) and one of the images;d) calculating an absolute position (52) of the lane edge point (34) defined in step c) from the vehicle position and the distance (42) for the time determined in step c); e) repeating steps c) and d) for a plurality of temporally successive absolute positions (52); f) determining a first lane edge line (46) of the lane by connecting or interpolating the plurality of temporally successive absolute positions (52); and g) repeating steps b) to e) for a second lane edge (36) opposite the first lane edge (36), and subsequently h) determining a second lane edge line (48) of the lane (60) opposite the first lane edge (36) by connecting or interpolating the plurality of temporally successive absolute positions (54) of the second lane edge (36); and; i) outputting the three-dimensional road map (44) comprising the first lane edge line (46) and the second lane edge line (48).

2. Method according to claim 1, wherein the recording in step b) takes place at time-synchronous intervals.

3. Method according to one of the preceding claims, wherein the camera device (13) has a first camera (14) on the left side of the vehicle (10) and a second camera (14) on the right side of the vehicle (10), wherein the first camera (14) is aligned such that its field of view includes an outer support point (32) of a left tire (16) and the second camera (14) is aligned such that its field of view includes an outer support point (32) of a right tire (18).

4. Method according to one of the preceding claims, further comprising step a1) after step a): a1) calibrating the camera device (13) with a calibration device (22) arranged on the floor of the lane.

5. The method according to claim 4, wherein the calibration device (22) comprises a measuring device with calibration markings arranged on the measuring device.

6. Method according to one of the preceding claims, wherein the lane edge (36) is defined in step e) using an edge detection algorithm.

7. Method according to one of the preceding claims, further comprising the steps: Calculating a center position (56) as the geographical center of an absolute position (52) of the first lane edge line (46) and an absolute position (54) of the second lane edge line (48) opposite the absolute position (52) of the first lane edge (36); and Determining a lane center line (50) by connecting or interpolating the temporally successive center positions (56).

8. The method of claim 7, further comprising the step: Calculating lane segments (58) as a polygon from two consecutive center positions (56) and those two absolute positions (52, 54) of the first lane edge line (46) and the second lane edge line (48) with which the two center positions (56) were calculated, wherein the lane (60) is defined as the sum of the lane segments (58).

9. The method according to any one of the preceding claims, further comprising the step: Calculating a lane center region (62) as a region lying between the first lane edge line (46) and the second lane edge line (48), each spaced by a distance value from the first lane edge line (46) and the second lane edge line (48).

10. The method of claim 9, further comprising the step: Calculating a lane edge area (64) as the difference between the area spanned by the first lane edge line (46) and the second lane edge line (48) and the lane center area (62). 1 1. Method according to one of the preceding claims, further comprising the step: Linking static or dynamic lane parameters to the lane (60).

12. Method according to one of the preceding claims, further comprising the steps: Providing a three-dimensional road map (44), in particular obtained according to one of the preceding claims, Driving over the edge of the lane (36) several times with the vehicle (10) and thereby Recording the vehicle orientation with the inertial measurement system, the vehicle position with the GNSS antenna (20) and images with the camera device (13), so that at least some of the images show an outer contact point (32) of a tire (16, 18) on the lane edge (36); Determining accuracy-optimized lane edge points (34) from vehicle alignment values recorded at the same time and the images showing the outer contact point (32) of the tire (16, 18) on the lane edge (36); and Validating and / or correcting the three-dimensional road map (44) with the accuracy-optimized lane edge points (34).

13. A three-dimensional road map (44) obtained by a method according to any one of the preceding claims.

14. Use of a three-dimensional road map (44) according to claim 13 for controlling an autonomous vehicle.

15. Use of a three-dimensional road map (44) according to claim 13 for detecting crossing of a lane edge line (46, 48), in particular in a racing competition, comprising the steps: Providing a three-dimensional road map (44) according to claim 13; Moving a detection vehicle with a GNSS device on a road area covered by the three-dimensional road map (44); Recording a vehicle position using the GNSS device; Comparing the vehicle position with coordinates of the first lane edge line (46) and / or the second lane edge line (48); and detecting crossing of a lane edge line (46, 48) by the comparison.

Images