METHOD, DEVICE AND COMPUTER-READABLE STORAGE MEDIUM WITH INSTRUCTIONS FOR DETERMINING THE LATERAL POSITION OF A VEHICLE RELATIVE TO THE LANES OF A ROAD

Description

[0001] The present invention relates to a method, a device, and a computer-readable storage medium containing instructions for determining the lateral position of a vehicle relative to the lanes of a roadway, and in particular for determining a relative lateral position with sub-lane accuracy. The present invention further relates to a vehicle with such a device and a lane geometry map with lane center geometries and lane edge geometries for use with such a method or device.

[0002] Modern vehicles are becoming increasingly autonomous, meaning they provide the driver with more and more functions and systems that assist in controlling the vehicle by providing guidance or taking over parts of the vehicle control. Such functions and systems require a wide range of information about the vehicle and its surroundings.

[0003] For the "lane-accurate navigation" function, for example, knowledge of the lane in which the vehicle being navigated, the "ego vehicle," is located is required. This lane is also referred to as the "ego lane." Furthermore, for automated driving and car-to-car applications, in addition to knowledge of the ego lane, more precise information regarding the ego vehicle's lateral position relative to the ego lane is needed. The ego vehicle's lateral position relative to the roadway must be known at all times with sub-lane accuracy.

[0004] Document US 2014 / 0358321 A1 discloses a method for detecting and tracking the boundaries of a traffic lane. The method uses maps containing information on road geometry, GPS data, historical data, and the positions of other vehicles to determine the current position.

[0005] The publication EP 2 899 669 A1 describes a method for determining the lateral position of a vehicle relative to the lane of a road. A camera is used to detect geometric aspects of the lanes, such as road markings. These detected aspects are classified and used for position determination. The classification process requires training of the classification unit.

[0006] German patent application DE 10 2012 104 786 A1 describes a system for accurately estimating the track a vehicle is traveling in. A track determination system provides estimated tracks determined in various ways. Examples include lane markings captured by a camera, a guide vehicle, or GPS / maps accurate to the track level. The estimated tracks are accompanied by confidence information. The estimated tracks and their corresponding confidence information are fused to produce a determined track.

[0007] In summary, three main approaches are currently being pursued for determining the lateral position of a vehicle relative to a roadway.

[0008] One approach involves using a highly accurate digital lane geometry map with centimeter-level accuracy in conjunction with a high-precision dual-frequency GPS system. Here, the position on the map is determined using the GPS sensor without additional imaging sensors. However, due to GPS and map inaccuracies regarding absolute position, it is often impossible to assign the ego vehicle to the correct lane. Furthermore, a solution using a highly accurate map and GPS is very expensive.

[0009] Another approach involves using imaging sensors, such as a camera system. This allows the ego-vehicle to be assigned to lanes based on lanes detected by the sensors. However, using imaging sensors without a digital map often results in only one or two lanes being recognized by the sensors.

[0010] The positioning of the ego vehicle can then only be relative to the detected lanes, but not relative to all lanes.

[0011] A third approach combines imaging sensors with map information regarding the number and marking type of lanes. By using imaging sensors and information from a digital map about the number of lanes and their marking type (dashed, solid, etc.), the ego-vehicle can be assigned to all lanes. However, due to distance errors in the lane markings detected by the sensors relative to the ego-vehicle, the accuracy of the lateral positioning relative to the corresponding lane is insufficient for the applications mentioned above.

[0012] The article by T. Wu et al., "Vehicle Localization using Road Markings," presented at the IEEE Intelligent Vehicles Symposium on June 23, 2013 (pages 1185-1190), describes a method for the visual localization of a vehicle using road markings such as arrows, pedestrian crossings, and speed limits. This method automatically detects and evaluates such markings. Features of the markings, particularly their corner points, are then used to calculate the vehicle's location. This calculation relies on a database containing previously measured features of the markings.

[0013] It is an object of the invention to provide a method and a device for determining a lateral position of a vehicle relative to the lanes of a roadway, which enables a determination of the relative lateral position with sub-lane accuracy.

[0014] This problem is solved by a method having the features of claim 1, by a device having the features of claim 4, and by a computer-readable storage medium with instructions according to claim 7. Preferred embodiments of the invention are the subject of the dependent claims. Fig. 1 schematically shows a method for determining the lateral position of a vehicle relative to the lanes of a roadway; Fig. 2 shows a first embodiment of a device for determining the lateral position of a vehicle relative to the lanes of a roadway; Fig. 3 shows a second embodiment of a device for determining the lateral position of a vehicle relative to the lanes of a roadway; and Fig. 4 shows a preferred embodiment of the device described in Fig. 1 described procedure.

[0015] To better understand the principles of the present invention, embodiments of the invention are explained in more detail below with reference to the figures. It is understood that the invention is not limited to these embodiments and that the described features can also be combined or modified without leaving the scope of protection of the invention as defined in the appended claims.

[0016] Fig. 1 Figure 1 schematically shows a procedure for determining the lateral position of a vehicle relative to the lanes of a roadway. In a first step, geometric and property information of the road markings is determined.10 Next, an approximate position of the vehicle is determined.11 Finally, by comparing the determined geometric and property information of the road markings with road marking geometries at the approximate position determined for the vehicle from a lane geometry map, the lateral position of the vehicle is determined.12 For this purpose, the lane geometry map contains lane center geometries and lane edge geometries with high relative accuracy.

[0017] A set of possible lateral positions of the vehicle is determined, and the best possible lateral position is then determined by applying at least one evaluation function. A history of past lateral positions determined for the vehicle can be taken into account. If a lateral position of the vehicle cannot be determined by comparing the determined geometry information and property information of the road markings with the road marking geometries, an approximate position can be generated.

[0018] Fig. 2 Figure 1 shows a simplified schematic representation of a first embodiment of a device 20 for determining a lateral position of a vehicle relative to the lanes of a roadway.

[0019] The device 20 has an image processing unit 22 for determining 10 geometric and property information of lane markings. For this purpose, the image processing unit 22 uses image information from a camera unit 25, which is received via an input 21 of the device 20. The device 20 also has a position determination unit 23 for determining 11 an approximate position of the vehicle. The approximate position of the vehicle is determined, for example, based on received data from a GPS receiver 26, which can also be received via the input 21. An evaluation unit 24 determines 12 the lateral position of the vehicle by comparing the determined geometric and property information of the lane markings with lane marking geometries at the approximate position determined for the vehicle from a lane geometry map.For this purpose, the lane geometry map contains lane center geometries and lane edge geometries with high accuracy relative to each other.

[0020] Evaluation unit 24 determines a set of possible lateral positions of the vehicle. The best possible lateral position is then determined by applying at least one evaluation function. A history of past lateral positions determined for the vehicle can be taken into account when determining the current lateral position. If a lateral position cannot be determined by comparing the determined geometry information and property information of the road markings with the road marking geometries, evaluation unit 24 can generate an approximate position.

[0021] The lateral position of the vehicle, determined by the evaluation unit 24, is preferably made available for further processing via an output 27 of the device 20, for example, for processing in a lane guidance system. It can also be stored in a memory 28 of the device 20, for example, for later evaluation. The input 21 and the output 27 can be implemented as separate interfaces or as a combined bidirectional interface. The image processing unit 22, the position determination unit 23, and the evaluation unit 24 can be implemented as dedicated hardware, for example, as integrated circuits. Of course, they can also be partially or fully combined or implemented as software running on a suitable processor.

[0022] Fig. 3 Figure 1 shows a simplified schematic representation of a second embodiment of a device 30 for determining the lateral position of a vehicle relative to the lanes of a roadway. The device 30 comprises a processor 32 and a memory 31. For example, the device 30 is a computer or a workstation. Instructions are stored in the memory 31 which, when executed by the processor 32, cause the device 30 to perform the steps according to one of the described methods. The instructions stored in the memory 31 thus embody a program executable by the processor 32, which implements the method according to the invention. The device has an input 33 for receiving information. Data generated by the processor 32 is provided via an output 34. Furthermore, it can be stored in the memory 31.Input 33 and output 34 can be combined into a bidirectional interface.

[0023] The processor 32 can comprise one or more processor units, such as microprocessors, digital signal processors, or combinations thereof.

[0024] The memory elements 28, 31 of the described embodiments can have volatile and / or non-volatile memory areas and can include a wide variety of storage devices and storage media, for example hard disks, optical storage media or semiconductor memory.

[0025] A preferred embodiment of the invention will now be described in detail. The method is based on a series of input data. First, the geometry and property information of the visible road markings, determined by the imaging sensor system, is required. These are referred to below as BV lines (BV for image processing). Also required is absolute position information, including direction and speed. This can be provided, for example, by the vehicle's GPS.In the present embodiment, a distinction is made between absolute position information that represents the direct result of a positioning by a GNSS system (GNSS: Global Navigation Satellite System) (GNSS position data) and that which has been interpolated based on a previous GNSS positioning by dead reckoning (dead reckoning navigation) (absolute position data). Optionally, relative, non-jump position information can also be used, which is determined, for example, by motion estimation. Additionally, map data with high relative accuracy regarding lanes and lane markings is used. This data is provided, for example, by a map data server 110. The map information is referred to below as DLM lanes (DLM: Detailed Lane Model) and DLM lane markings.Preferably, results from the previous iteration or iterations of the procedure are included as history, except of course in the first iteration.

[0026] The basic procedure is divided into several parts, which are in Fig. 4 some are grouped together in blocks shown with dashed lines.

[0027] As part of the processing of the input data, an approximate vehicle position is determined. This serves as the starting point for comparing road markings with map data. In addition, the geometries of the BV lines are converted and aggregated.

[0028] The geometries of the road markings detected by the camera system are typically described by clothoids in a vehicle-relative Cartesian coordinate system. The clothoid descriptions are converted into polylines that closely approximate the clothoid geometries. This conversion to polylines is performed because it significantly simplifies the implementation of the subsequent algorithms operating on the road marking geometries. In the present embodiment, the coordinates of the polylines are converted from the vehicle-relative Cartesian coordinate system to the WGS84 coordinate system. The subsequent algorithms operate in the WGS84 coordinate system because the map data and the vehicle position and movement information are also typically in this coordinate system.

[0029] The geometries of the lane markings detected by the camera system always begin shortly before the vehicle and extend for a few meters approximately in the direction of travel / camera view. Occasionally, the camera system initially detects and transmits lane markings correctly, but then fails to detect them shortly thereafter. Therefore, the BV lines present during an algorithm iteration are always temporarily stored and compared with the geometries newly transmitted by the camera system in the next iteration.

[0030] Subsequently, essential geometric aspects are extracted 50. Here, points from BV lines and DLM lane markings are calculated for the longitudinal vehicle position. For this purpose, an orthogonal line of a configurable length is first constructed at a configurable distance from the approximate vehicle position 51. Then, the intersection points of the DLM lane markings and the DLM lanes with the orthogonal line are determined 52. For each intersection point, its lateral position relative to the vehicle is preferably recorded, e.g., as the distance from the midpoint of the orthogonal line, and, in the case of DLM lane markings, information on the type of associated lane marking (dashed / solid line, guardrail, lane edge, etc.). In a further step, the intersection points of the orthogonal line with the BV lines are calculated 53.Again, for each intersection point, its lateral position relative to the vehicle and the information on the type of detected road marking (dashed / solid line, guardrail, road edge,...) are preferably recorded.

[0031] Based on the results of the previous step, a set of possible vehicle positions is determined 60. These are subsequently evaluated by a sequence of corresponding evaluation functions 70, whereby the possible vehicle positions are supplemented or modified as necessary. Penalty points are assigned using the evaluation functions. A first evaluation function 71 considers the assignment of the line types detected by the camera to the line types stored in the map. For this evaluation, a configurable matrix is ​​preferably provided, which assigns a specific value to each combination of BV line and map lane marking type. In this way, frequently occurring misassignments by the camera, e.g., the identification of a solid line as a dashed line, can be assigned only a slightly poor rating, while unlikely misassignments by the camera, e.g.,The detection of the road edge as a guardrail is associated with a significantly poor rating. A second rating function 72 considers the history of vehicle positions. Possible vehicle positions that deviate significantly from the history are characterized, for example, by high penalty points. In the present embodiment, a third rating function 73 evaluates the lane type. The vehicle is presumed to be in a regular lane of the roadway. Possible vehicle positions in lanes not intended for use (shoulders, "unknown lanes" of the DLM, and emergency lanes, etc.) are therefore rated poorly, while possible vehicle positions in drivable lanes are rated neutrally. As a further example, significantly higher penalty points are assigned for possible vehicle positions in the opposite lane than for positions in the direction of travel.The penalty point system depends on the sensor system used and the digital map employed.

[0032] This allows for very easy adjustments to the systems used. As a result of the evaluation process, the best-rated possible vehicle position is ultimately selected (80).

[0033] During the process, it may occur at various points that, due to missing or insufficient input data, a position determination is not possible and the process cannot continue. In such cases, it is possible to exit the modular process at the relevant point and initiate error handling (90), for example, generating a result without position information or with position information determined approximately by another method. Fig. 4The paths of the regular process are indicated by solid arrows, while paths where the process deviates from the regular process are indicated by dashed arrows.

[0034] For example, if map data for the current vehicle position is unavailable according to absolute position information, no determination of possible vehicle positions is performed. If absolute position information is unavailable (e.g., no GNSS reception due to buildings), and relative position information is available, the absolute position information can be replaced by a previous absolute position and relative position information. If no relative position information is available, no determination of possible vehicle positions is performed if absolute position information is also unavailable. In such cases, a corresponding error message is displayed.If, however, too few BV lines were detected, a sufficiently accurate assignment of DLM lane markings to BV lines cannot be found, or no possible vehicle positions were determined, an approximate position information can be obtained using a map-based method. One possibility, for example, is to assume that the vehicle movement continues along the lane centerline according to the map. In a final step, the results are processed, made available for further processing, and incorporated into the history.

Claims

1. Method for determining a lateral position of a vehicle relative to the lanes of a roadway, comprising the steps of: - ascertaining (10) geometry information and property information of roadway markings, by means of an image processing unit (22) which processes image information from a camera unit (25), the geometry information describing a course of the roadway markings in the direction of driving for a limited region in front of the vehicle and the property information describing a type of the relevant roadway markings; - determining (11) an approximate position of the vehicle by means of a position determining unit (23); and - determining (12) the lateral position of the vehicle by means of an evaluation unit (24) by comparing the ascertained geometry information and property information of the roadway markings with roadway marking geometries at the approximate position determined for the vehicle from a lane geometry map; characterized in that the lane geometry map contains lane center geometries and lane edge geometries having high accuracy relative to each other, and, when determining (12) the lateral position of the vehicle by comparing the ascertained geometry information and property information of the roadway markings with the roadway marking geometries from the lane geometry map, a set of possible lateral positions of the vehicle is determined, and a best possible lateral position of the vehicle is ascertained by applying a sequence (70) of corresponding analysis functions, a first analysis function (71) analyzing an assignment, which underlies the relevant possible lateral position, of the line types of the roadway markings detected by the camera unit (25) to the line types of the roadway markings stored in the lane geometry map, a second analysis function (72) analyzing a deviation of the relevant possible lateral position from a history of the positions, and a third analysis function (73) analyzing a location of the relevant possible lateral position on a provided drivable or non-drivable lane.

2. Method according to claim 1, wherein the ascertained property information of the roadway markings indicates that the roadway marking is a dashed line, a solid line, a crash barrier, or a roadway edge.

3. Method according to claim 1 or 2, wherein, in the event that a lateral position of the vehicle cannot be determined by comparing the ascertained geometry information and property information of the roadway markings with roadway marking geometries, an approximated lateral position is generated.

4. Device (20) for determining a lateral position of a vehicle relative to the lanes of a roadway, wherein the device (20) comprises the following: - an image processing unit (22), which processes image information from a camera unit (25), for ascertaining (10) geometry information and property information of roadway markings, wherein the geometry information describes a course of the roadway markings in the direction of driving for a limited region in front of the vehicle and the property information describes a type of the relevant roadway markings; - a position determining unit (23) for determining (11) an approximate position of the vehicle; and - an evaluation unit (24) for determining (12) the lateral position of the vehicle by comparing the ascertained geometry information and property information of the roadway markings with roadway marking geometries at the approximate position determined for the vehicle from a lane geometry map; characterized in that the lane geometry map contains lane center geometries and lane edge geometries having high accuracy relative to each other, and the evaluation unit (24) is configured, when determining (12) the lateral position of the vehicle by comparing the ascertained geometry information and property information of the roadway markings with the roadway marking geometries from the lane geometry map, to determine a set of possible lateral positions of the vehicle, and to ascertain a best possible lateral position of the vehicle by applying a sequence (70) of corresponding analysis functions, a first analysis function (71) analyzing an assignment, which underlies the relevant possible lateral position, of the line types of the roadway markings detected by the camera unit (25) to the line types of the roadway markings stored in the lane geometry map, a second analysis function (72) analyzing a deviation of the relevant possible lateral position from a history of the positions of the vehicle, and a third analysis function (73) analyzing the location of the relevant possible lateral position on a provided drivable or non-drivable lane.

5. Device (20, 30) according to claim 4, wherein the ascertained property information of the roadway markings indicates that the roadway marking is a dashed line, a solid line, a crash barrier, or a roadway edge.

6. Device (20, 30) according to claim 4 or 5, wherein, in the event that a lateral position of the vehicle cannot be determined by comparing the ascertained geometry information and property information of the roadway markings with roadway marking geometries, the evaluation unit (24) is configured to generate an approximated lateral position.

7. Computer-readable storage medium comprising instructions that, when executed by a computer, cause the computer to execute the steps of a method according to any of claims 1 to 3 in order to determine a lateral position of a vehicle relative to the lanes of a roadway.

8. Autonomous or manually controlled vehicle, characterized in that it comprises a device according to any of claims 4 to 6.