Methods for detecting road damage and methods for operating a vehicle

Abstract

The invention relates to a method for detecting road damage (FS) on a roadway (FB). According to the invention, lateral vehicle positions (POS) of a plurality of vehicles (3.1 to 3.n) within a lane driven by each of these vehicles are collected relative to lane boundaries (B1, B2) and / or to a respective lane centerline by means of a central processing unit (2). Based on the collected lateral vehicle positions (POS), a statistical position distribution (SV) on a given roadway segment (S) is determined, whereby a deviation between the determined statistical position distribution (SV) and a given reference position distribution (RV) is calculated, and a conclusion is drawn that at least one road damage defect (FS) is present if the calculated deviation exceeds a predetermined limit value. The invention further relates to a method for operating a number of vehicles (3.1 to 3.n).

Description

[0001] The invention relates to a method for detecting road surface damage on a roadway.

[0002] The invention further relates to a method for operating a vehicle.

[0003] Road surface damage, such as potholes, can impair driving comfort and damage vehicle components, resulting in increased costs. Such road surface damage can be detected after driving over or through the surface or through image recognition using machine learning methods.

[0004] DE 10 2012 018 122 A1 discloses a method for autonomously guiding a motor vehicle along a roadway based on a planned target trajectory. A sensor device detects unevenness in the roadway, and a target trajectory is planned based on the detected unevenness in such a way that it avoids the unevenness. The sensor device comprises a stereo camera, a radar sensor, a lidar sensor, or a PMD sensor (PMD = Photo Mixed Detector).

[0005] The invention is based on the object of providing a novel method for detecting road surface damage on a roadway and a novel method for operating a vehicle.

[0006] The object is achieved according to the invention by a method for detecting road surface damage, which has the features specified in claim 1, and by a method for operating a vehicle, which has the features specified in claim 8.

[0007] Advantageous embodiments of the invention are the subject of the subclaims.

[0008] In a method for detecting road damage on a roadway, the invention uses a central processing unit, for example, a backend server, to collect lateral vehicle positions of a plurality of vehicles within a lane they are each traveling in relative to lane boundaries and / or to a respective lane center. Based on the collected lateral vehicle positions, the processing unit determines a statistical position distribution on a predetermined section of road and a deviation between the determined statistical position distribution and a predetermined reference position distribution. At least one existing road damage is inferred if the determined deviation exceeds a predetermined limit.

[0009] In contrast to image-based detection of road damage using machine learning methods, which require additional detection of potholes that do not affect traffic, and in contrast to detection of road damage based on driving over or through these, which does not allow detection of road damage if the vehicle does not drive over or through them, the present method can be used to detect road damage in a simple and reliable manner. This method is based on the detection of steering behavior that deviates from normal driving in a lane and reports road damage if a limit value is exceeded.

[0010] The procedure is based on the fact that vehicle behavior and driving patterns when avoiding road defects deviate from normal vehicle behavior and driving patterns, since the road defects are either avoided or driven over between the wheels.

[0011] This means that a vehicle traveling along a section of road can be provided with early information about road damage, preventing it from being driven over or through. This increases driving comfort and reduces vehicle wear and tear. This simultaneously leads to increased customer satisfaction. Furthermore, the method can be used to particularly advantageously increase driving safety, particularly through fleet-supported detection of road damage. Information on road damage obtained using the method can also be used to maintain a traffic network, resulting in increased efficiency in improving road quality. The aforementioned advantages are achieved using digital means (English: Digital Luxury) through an intelligent, data protection-compliant methodology for road damage detection.

[0012] In one possible embodiment of the method, a statistical position distribution determined at an earlier point in time is specified as the reference position distribution. Thus, it is not necessary to calculate the reference position distribution in the form of trajectories; instead, a comparison is simply made between statistical position distributions determined using the vehicle fleet over time, since changing behavior of the vehicle fleet also indicates road changes. For example, a change in a trajectory distribution over time, such as a shift or change in shape, can be detected, or trajectory distributions from various earlier points in time can be compared with a current trajectory distribution. If the difference exceeds a threshold, road damage can be assumed.

[0013] In another possible embodiment of the method, the reference position distribution is determined based on digital map data using a trained artificial neural network. This enables a simple, automated, and reliable determination of the reference position distribution.

[0014] In another possible embodiment of the method, a normal distribution is specified as the reference position distribution. When driving on an undamaged road, no or only minimally differing steering maneuvers or trajectories are to be expected within a vehicle fleet. This expectation can be easily and reliably approximated by modeling a normal distribution.

[0015] In another possible embodiment of the method, the lateral vehicle positions are determined based on data acquired by on-board sensors. This makes it particularly advantageous to use vehicle sensors that are already installed in the vehicles for other functions, such as driver assistance functions. Thus, no additional hardware is required in the vehicles to implement the method.

[0016] In another possible embodiment of the method, a temporal progression of a steering angle determined from data acquired by the sensors is used to determine the lateral vehicle positions. By monitoring the steering angle in this way, the steering behavior of vehicle users can be easily monitored. Furthermore, by monitoring the steering angle, the lateral vehicle positions within the lane relative to lane boundaries and / or the respective lane center can be determined particularly easily and reliably.

[0017] In another possible embodiment of the method, the lateral vehicle positions for the given route section are aggregated, and an average value of a trajectory traveled by the multiple vehicles in this route section is calculated from the aggregated lateral vehicle positions. By calculating such an average value, the statistical position distribution of the vehicles can be determined particularly easily and reliably.

[0018] In the method according to the invention for operating a number of vehicles, road surface damage on a roadway is determined according to the aforementioned method, and if road surface damage is detected in this way, road surface damage information relating to this damage is transmitted to the number of vehicles. This enables vehicles traveling on a section of road to be provided with early information about road surface damage on this section, so that driving over or through the road surface damage can be avoided. This increases driving comfort and reduces vehicle wear. This simultaneously leads to increased customer satisfaction. Furthermore, the method can be used to particularly advantageously increase driving safety, in particular through fleet-supported detection of road surface damage.The aforementioned advantages are achieved by digital means (English: Digital Luxury) through an intelligent, data protection-compliant methodology for road damage detection.

[0019] In one possible embodiment of the method, after receiving road damage information from the group of vehicles, information concerning the corresponding road damage is output within the group of vehicles, and / or an automatic longitudinal and / or lateral control of the group of vehicles is adapted to the corresponding road damage. This allows drivers of the vehicles to be reliably warned of road damage, and vehicle damage and a loss of comfort for vehicle occupants can be avoided or at least reduced.

[0020] Embodiments of the invention are explained in more detail below with reference to drawings.

[0021] Showing: Fig. 1 schematically shows a vehicle on a roadway, a central processing unit and a reference position distribution and Fig. 2 schematically shows a vehicle on a road with road damage, a central processing unit, a vehicle fleet, a reference position distribution and a statistical position distribution.

[0022] Corresponding parts are provided with the same reference numerals in all figures.

[0023] In Fig. 1 shows a vehicle 3.1 on a roadway FB in a section S, a central processing unit 2 and a reference position distribution RV. Fig. 2 shows the vehicle 3.1 on a roadway FB with a roadway damage FS in a road section S, a central processing unit 2, a vehicle fleet 3 with a plurality of vehicles 3.1 to 3.n, a reference position distribution RV and a statistical position distribution SV.

[0024] Road surface damage (FS), such as potholes, can impair driving comfort and damage vehicle components (3.1 to 3.n), resulting in increased costs. Therefore, it is essential to detect such road surface damage (FS) in a timely manner in order to avoid it.

[0025] For such detection, it is assumed that vehicle behavior and driving patterns when avoiding road surface defects FS deviate from normal vehicle behavior and driving patterns, since the road surface defects FS are either avoided or driven over between the wheels of vehicles 3.1 to 3.n.

[0026] To detect road surface damage FS, the steering behavior of vehicle users of vehicles 3.1 to 3.n, for example, by monitoring a steering angle, and / or trajectories T, T1, T2 driven by vehicles 3.1 to 3.n are determined using on-board sensors. This means that lateral vehicle positions POS within a lane traveled by vehicles 3.1 to 3.n relative to lane boundaries B1, B2 and / or to a respective lane center are collected. This is described in the present Fig. 1 and Fig. 2 using the example of vehicle 3.1.

[0027] The lateral vehicle positions POS determined by vehicles 3.1 to 3.n are transmitted to the central processing unit 2, which collects them. The central processing unit 2 is, for example, a central data center, a cloud, and / or a server, such as a backend server.

[0028] When driving on an undamaged roadway FB, no or only minimally differing steering maneuvers or trajectories T, T1, T2 are to be expected between the vehicles 3.1 to 3.n. This expectation can be approximated by modeling a reference position distribution RV, such as a normal distribution. This reference position distribution RV can be calculated by calculating trajectories T, T1, T2 for a given map. Each road per roadway FB is divided into sections S of a certain length, for example, 5 m. For each of these sections S, the trajectory T is calculated that a vehicle 3.1 to 3.n would take at this location.This means that the expected steering behavior and / or the expected trajectories T, T1, T2 are / are derived, for example, from digital map data for the respective route section S, for example, through a heuristic description of the expected trajectory T, T1, T2 as a function of a curve radius, a road width, etc. and / or with a machine-trained model. The reference point on vehicle 3.1 to 3.n can be any point in the width, as long as it matches an actually measured point of other fleet individuals, i.e., the other vehicles 3.1 to 3.n of fleet 3. This means that for each vehicle 3.1 to 3.n, its lateral vehicle position POS relative to lane boundaries B1, B2 and / or to the respective lane center is determined.After this determination, the lateral vehicle positions POS and the resulting trajectory T for this route section S are used as the mean value of the reference position distribution RV, such as a normal distribution, and defined as the expected driving behavior using a given standard deviation.

[0029] In order to detect road surface damage FS, a statistical position distribution SV on the specified route section S is determined by means of the processing unit 2 from the lateral vehicle positions POS determined using the vehicles 3.1 to 3.n and the lateral vehicle positions POS collected by the central processing unit 2. This means that in this route section S, all steering behaviors of vehicle users of the vehicles 3.1 to 3.n of the vehicle fleet 3 and / or trajectories T, T1, T2 driven by all vehicles 3.1 to 3.n of the vehicle fleet 3 are transmitted to the processing unit 2 in the form of the respective lateral vehicle positions POS of the vehicles 3.1 to 3.n and aggregated by means of this in the statistical position distribution SV.

[0030] Subsequently, the aggregated behavior of the vehicles 3.1 to 3.n is compared with an expected steering behavior and / or an expected trajectory T, T1, T2 in this route section S, i.e. with the reference position distribution RV.

[0031] If a road surface damage FS were to occur on lane FB, it would increasingly be bypassed by vehicles 3.1 to 3.n or—depending on the extent of the road surface damage FS—driven over between their wheels. Both of these actions require a steering action, since road surface damage FS is primarily caused by loads on the road surface FB and usually occurs in the area of ​​a lane overrun by the wheels of vehicles 3.1 to 3.n. Thus, the road surface damage FS also primarily occurs in the lanes and not in the center of the road surface FB.

[0032] If these evasive or over-driving steering maneuvers occur increasingly in a section of road S, this can be detected in the computing unit 2, as a strong deviation from the expected reference position distribution RV occurs. If a limit value is exceeded and / or the deviation is statistically significant, this deviation can be classified as road surface damage FS and noted as such in the computing unit 2 and transmitted to the vehicles 3.1 to 3.n of the vehicle fleet 3. This means that if a deviation of the fleet behavior in the form of the statistical position distribution SV from the reference position distribution RV is statistically significantly greater or at least an evasive value is greater than a previously defined threshold value and thus deviates significantly from the previously determined, in particular dynamically adaptable, reference position distribution RV, the computing unit 2 concludes that road surface damage FS is present.

[0033] Such road damage FS is stored, in particular, with associated position data on the computing unit 2, and corresponding road damage information FS-I is transmitted to the number of vehicles 3.1 to 3.n via a communication connection. This enables information concerning the corresponding road damage FS to be output within the number of vehicles 3.1 to 3.n, so that vehicle users in the relevant road section S or ahead of it can adapt their driving style to the road damage FS in good time and / or avoid it in a timely manner. Depending on the number of vehicles 3.1 to 3.n, an automatic longitudinal and / or lateral control system can also be adapted to the corresponding road damage FS.

[0034] In one possible embodiment, no normal behavior in the form of trajectories T, T1, T2, i.e., no reference position distribution RV needs to be calculated. Instead, a comparison is simply made between the data transmitted from vehicles 3.1 to 3.n to computing unit 2 over time, since behavior that changes over time can also indicate road changes. For example, a change, such as a shift or change in shape, in a trajectory distribution over time can be detected, or a trajectory distribution at various earlier points in time can be compared with a current trajectory distribution. If the difference exceeds a threshold value, road damage FS is inferred. This means that a statistical position distribution SV determined at an earlier point in time can be specified as the reference position distribution RV. QUOTES CONTAINED IN THE DESCRIPTION

[0000] This list of documents submitted by the applicant was generated automatically and is included solely for the convenience of the reader. This list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions. Cited patent literature

[0000] DE 10 2012 018 122 A1

[0004]

Claims

[1] Method for detecting road surface damage (FS) on a road surface (FB), characterized by that by means of a central processing unit (2) - lateral vehicle positions (POS) of a plurality of vehicles (3.1 to 3.n) within a lane travelled by them relative to lane boundaries (B1, B2) and / or to a respective lane centre are collected, - a statistical position distribution (SV) on a given route section (S) is determined based on the collected lateral vehicle positions (POS), - a deviation between the determined statistical position distribution (SV) and a given reference position distribution (RV) is determined and - at least one existing road surface damage (FS) is concluded if the determined deviation exceeds a specified limit. [2] Method according to claim 1, characterized bythat a statistical position distribution (SV) determined at an earlier point in time is specified as the reference position distribution (RV). [3] Method according to claim 1 or 2, characterized by that the reference position distribution (RV) is determined from digital map data using a trained artificial neural network. [4] Method according to one of the preceding claims, characterized by that a normal distribution is specified as the reference position distribution (RV). [5] Method according to one of the preceding claims, characterized by that the lateral vehicle positions (POS) are determined on the basis of data recorded in the vehicles (3.1 to 3.n) by means of on-board sensors. [6] Method according to claim 5, characterized bythat to determine the lateral vehicle positions (POS), a temporal progression of a steering angle determined on the basis of data recorded by the sensors is used. [7] Method according to one of the preceding claims, characterized by that the lateral vehicle positions (POS) for the given route section (S) are aggregated and an average value of a trajectory (T, T1, T2) driven by the several vehicles (3.1 to 3.n) in this route section (S) is formed from the aggregated lateral vehicle positions (POS). [8] Method for operating a number of vehicles (3.1 to 3.n), wherein - road surface damage (FS) located on a road surface (FB) is determined according to a method according to one of the preceding claims and - if road surface damage (FS) is detected in this way, road surface damage information (FS-I) relating to this damage is transmitted to the number of vehicles (3.1 to 3.n). [9] Method according to claim 8, wherein after receiving road damage information (FS-I) from the number of vehicles (3.1 to 3.n) - information concerning the corresponding road surface damage (FS) is output within the number of vehicles (3.1 to 3.n) and / or - an automatic longitudinal and / or lateral control of the number of vehicles (3.1 to 3.n) is adapted to the corresponding road surface damage (FS).

Images