Method for controlling a fleet-based condition monitoring system for a road section of a road network, as well as associated system, motor vehicle, and associated server device

Description

[0001] The invention relates to a method for controlling fleet-based condition monitoring of a road section of a road or traffic network. The invention provides a system for implementing the method, the components of which, namely a server device and motor vehicles, are also considered part of the invention. It is known from the prior art to monitor a road section of a road network using a so-called vehicle fleet in order to monitor or check the condition of the road section with regard to, for example, road damage and / or existing traffic signs and other traffic infrastructure components. The vehicle fleet used for this purpose generally does not include dedicated measuring vehicles; instead, for example, motor vehicles belonging to private individuals are used, which may also be equipped with suitable sensors to carry out condition monitoring.

[0002] For condition monitoring, each vehicle in the fleet can collect observation data describing at least one object located on the road section. "Object" here refers to a damaged area (road damage) or a traffic infrastructure component. The description data can then be transmitted to a central server, where all monitored objects can be analyzed, for example, for at least one property (e.g., size in the case of road damage). The server can thus generally use the observation data to determine the current condition of at least one object.Examples of suitable observation data include image data from a motor vehicle camera, analysis data from an acceleration sensor, detection data from an object detection, and geoposition data of an object (as determined by a receiver for a position signal from a GNSS - "Global Positioning Satellite System"). The status can indicate whether a specific object, such as road damage such as a pothole, is present and / or the size of the road damage. The status can also indicate the presence and / or type of a traffic infrastructure component (e.g., a road sign or traffic light). A suitable server device can be, for example, an internet server platform.

[0003] Fleet-based condition monitoring of a road section is described as an example in US 2014 / 0160295 A1. How road damage in a road section can be monitored using a conventional motor vehicle is described, for example, in GB 2549088 A and US 2017 / 0243370 A1.

[0004] In fleet-based condition monitoring of a road section, the amount of data sent from the vehicles to the central server can be very large, as some of the data sent to the server is redundant. Incorrect detection (so-called "false positives"), such as those that can occur with curbs or speed bumps, also results in unnecessary data traffic. Continuously operating the sensors in a fleet to acquire new observation data also results in unnecessary energy consumption in the sensors, which is particularly undesirable for battery-operated sensors, such as those found in vehicle wheels.

[0005] US 2018 / 0068495 A1 discloses a method for fleet-based detection of road damage that operates on a threshold basis to prevent even small vibrations, such as those that can occur during normal driving, from leading to an incorrect signaling of a suspected road damage.

[0006] DE 10 2016 003 969 A1 describes a method for acquiring environmental data which describe at least one property of an environment, such as a temperature, road surface smoothness or the position of an infrastructure element, wherein observation data in the environment are acquired by a plurality of motor vehicles by means of a respective sensor device, and wherein a portion of the observation data is received by a stationary server device and the received observation data of different motor vehicles are combined into updated environmental data on the basis of a predetermined aggregation criterion.Based on environmental data already stored in a central storage device, the server device determines a minimum requirement criterion that must be met by updated environmental data. A query message containing the minimum requirement criterion is sent to the motor vehicles. Furthermore, the motor vehicles store the observation data in a respective vehicle memory, and those observation data that meet the minimum requirement criterion are sent to the server device.

[0007] The invention is based on the object of freeing the data traffic caused between the motor vehicles and the server device in a fleet-based condition monitoring of a road section of a traffic network from redundant data and / or false triggers.

[0008] This object is achieved by the subject matter of the independent patent claims. Advantageous embodiments of the invention are described by the dependent patent claims, the following description, and the figures.

[0009] The invention provides a method for controlling fleet-based condition monitoring of a road section of a road network. Fleet-based condition monitoring here means that observation data relating to at least one object located in the road section are determined by motor vehicles of a vehicle fleet and transmitted to a central server device. The server device then determines a current condition of the at least one object based on the observation data. This part of the method can be implemented in a manner known from the prior art and using means known from the prior art, as already described above.

[0010] In order to prevent redundant observation data and / or observation data based on a false positive from being unnecessarily sent from the motor vehicles to the server device, the invention provides that a data reduction device reduces the amount of observation data to be transmitted by operating at least one restriction function in at least one of the motor vehicles and / or in the server device. The respective restriction function thus implements logic to prevent redundant observation data and / or observation data not required in the server device from being generated in the first place or at least to prevent their transmission to the server device. The respective restriction function triggers the determination and / or transmission of new observation data only if a predetermined requirement criterion for new observation data is met.In other words, the data reduction mechanism represents the set of measures provided in the system consisting of a server device and a vehicle fleet to avoid superfluous observation data. Each measure can be implemented or realized through a restriction function. What is superfluous or unnecessary can be determined by the expert based on the requirements criteria for the respective specific or concrete condition monitoring.

[0011] The data reduction device is realized by implementing one or more restriction functions in a respective motor vehicle. A restriction function can be provided, for example, on the basis of a program code or software for a processor circuit of the motor vehicle, as can be realized, for example, by one or more control units. Additionally or alternatively, a restriction function can also be realized in the server device as a component of the data reduction device. One or more such restriction functions for the server device can be provided by a program code or software for a processor circuit of the server device. A server device can be realized by a computer or a computer network, which can be operated, for example, on the Internet as an Internet server.The transmission of the observation data from a respective motor vehicle to the server device can be realized, for example, on the basis of a mobile radio connection and / or a WLAN connection (WLAN - wireless local area network) and / or an Internet connection.

[0012] A respective restriction function checks whether the respective observation data is actually necessary for condition monitoring in the server device before it is sent or even before it is actually generated or determined. To do this, the restriction function can check a specific condition as a requirement criterion and only enable or trigger the determination and / or transmission of new observation data if the corresponding condition of the requirement criterion is met. The requirement criterion can therefore specify all those conditions that determine when new observation data is required for condition monitoring in the server device. By specifying the conditions, a specialist can thus set how redundant or unnecessary observation data can be detected and blocked in the system for a specific monitoring task or condition monitoring.

[0013] The invention provides the advantage that in the system comprising a server device and a vehicle fleet, wherever redundant or unnecessary observation data could be generated or transmitted, these can be prevented by means of a corresponding restriction function.

[0014] The invention also includes embodiments which provide additional advantages.

[0015] According to the invention, condition monitoring relates to road damage. This means that a particular object can be road damage, such as a pothole or a crack in a road surface, or generally an unevenness or a damaged area (damaged area). In general, an object in the present sense is, in particular, a stationary object.

[0016] In one embodiment, additionally or alternatively, it is provided that at least one traffic infrastructure component is monitored by means of condition monitoring, i.e. a monitored object can be a traffic infrastructure component. A traffic infrastructure component is an element provided in the road section, at the roadside, or on the carriageway that serves to guide and / or regulate road traffic. Examples of a traffic infrastructure component are: a traffic light, a traffic sign, a construction site barrier, a guardrail, and a lane marking on a road surface. Objects of this type (road damage and traffic infrastructure components) can advantageously be monitored using motor vehicles that are used privately or are on the road for another commercial purpose, thus eliminating the need for dedicated measuring vehicles.

[0017] In one embodiment, the data reduction device comprises, as a restriction function, a classifier operated in one or more or each of the motor vehicles and / or in the server device, which receives sensor data from at least one sensor circuit of the motor vehicle and, based on the sensor data, classifies an object depicted in the sensor data that is located in the road section, thereby recognizing an object class of the object. The associated requirement criterion of the restriction function, in order to decide on further determination of observation data, includes, in particular, an assignment rule determining for which of the object classes for whose recognition the classifier as a whole is configured, observation data should be transmitted and for which not.In other words, in the case of a vehicle-based classifier, sensor data is not simply transmitted unverified as new observation data from the motor vehicle to the server device. Rather, a local check is first carried out in the motor vehicle to determine whether the sensor data actually describes an object for which observation data is to be transmitted to the server device. If a classifier is operated in the server device, it can be used to decide, based on the assignment rule, whether further, new observation data for an object should be requested from the same or at least from another motor vehicle. The classifier used in a motor vehicle or in the server device can, for example, be based on a machine learning (ML) method, e.g., an artificial neural network, or another artificial intelligence (AI) algorithm.A classifier can be trained with training data to recognize object classes. Such training data represents sensor data, to which so-called label data is additionally provided. This label data can, for example, be created by an operator and indicates which object or, more generally, which object class is actually described (ground truth) in the sensor data. If new sensor data is then passed to the trained classifier, it can signal the object class represented therein. Possible object classes can be: road damage, road damage of a predetermined size, pothole, pothole of a predetermined size, crack, traffic infrastructure component, traffic infrastructure component of a predetermined type, traffic sign, construction site barrier, road markings – although these are only examples.In particular, object classes can also signal different extents or sizes of damage to a damaged area of ​​a road surface.

[0018] The assignment rule can flexibly determine which of the detected or recognized object classes should then be reported to the server device for further observation data. This assignment rule can also be designed to be adaptive or modifiable. For example, the server device can transmit assignment data to the respective vehicle to establish a currently valid assignment rule. The assignment rule can be designed, for example, as a look-up table.

[0019] Classification in the motor vehicle and filtering using an assignment rule, in particular an assignment rule that can be adaptively adjusted by the server device, offers the advantage that the generation of undesired observation data can be suppressed or prevented in the respective motor vehicle. In particular, false detections, such as the detection of a speed bump or a curb as an obstacle on a roadway, can be avoided by means of the classification. The classification can receive, for example, camera data from at least one camera of the motor vehicle and / or ultrasonic measurement data from at least one ultrasonic sensor and / or radar data from at least one radar sensor as sensor data and process them for classification.

[0020] One embodiment provides a special combination or fusion of sensor data that could not previously be combined due to a time lag between the generation of this sensor data. To this end, to provide sensor data from both a camera and an acceleration sensor, image data from the camera is temporarily stored in a ring buffer or similar intermediate buffer. In the event that an unevenness in the road surface causes a predetermined triggering pattern in the acceleration data from the acceleration sensor, the acceleration data containing the triggering pattern and, in addition, the corresponding camera data from the intermediate buffer, in which the unevenness must be depicted, are combined as sensor data for input to the classifier.

[0021] This allows camera images to be combined with, for example, acceleration data, without the image having to be recorded at the same time as the acceleration. This allows forward-looking or rear-facing cameras to be used. In other words, the classifier can use sensor data from two different sensors—namely, sensor data from a camera and an acceleration sensor—to perform the classification, making the classification more robust.

[0022] If a camera is filming a section of road ahead, for example, an unevenness will already be depicted in the image data before the vehicle reaches or drives over it. Only then, when the vehicle drives over it, can this unevenness also be measured using the acceleration sensor. If this unevenness is identified as relevant, i.e. it corresponds to the trigger pattern, for example a predetermined temporal progression of a time signal of the vertical acceleration, it must then be possible to access the image data from the camera that was previously recorded when the unevenness was still in front of the vehicle in order to obtain image data about the unevenness. For this purpose, the ring buffer or similar temporary buffer is used, from which the corresponding image data from the past or into the future can be read.The size of the buffer can be adjusted by a specialist to suit the frame rate of the camera and the driving speeds achievable by the vehicle.

[0023] In order to find the said correct or corresponding image data, one embodiment provides that the image data corresponding to the acceleration data are selected from the intermediate buffer: a) based on a current driving speed and a time offset between the image data and a time at which the unevenness is passed over (as can be determined from a time at which the triggering pattern is detected) and / or b) based on a wheel speed of at least one wheel and a radius value of the respective wheel.

[0024] In the intermediate buffer, the refresh rate or so-called frame rate can be used to determine how far back in time from a current memory position or a current memory pointer in the buffer to past image data must be traveled in order to reach image data from a predetermined point in the past. Starting from the time of travel, the driving speed and the known geometric orientation of the camera can be used to determine at what point in time the unevenness must have been within the camera's detection range. The time offset can be determined accordingly, and the corresponding image data can then be read from the buffer. Based on the geometric orientation of the camera, the distance between the camera's detection range on the road surface and the position of a particular wheel is known.Using the wheel speed (rotational speed) or number of wheel revolutions and a radius of the respective wheel, the number of wheel revolutions and, based on the wheel speed, the corresponding time offset can also be calculated. The radius value is preferably dynamic, meaning it is adjusted to the current deformation of the wheel's tire, which can result, for example, from wear and / or rotational speed and / or cornering position.

[0025] With regard to the case according to the invention in which road damage, i.e. a damaged area in the road surface, is observed or monitored as at least one object, the invention provides that the data reduction device comprises, as a restriction function, that in the server device, based on initial observation data (which describe the damaged area), a model for a temporal progression of a damage expansion or damage change is operated, which model estimates a growth of the damaged area in the road surface based on weather data and / or traffic data and determines an estimated condition therefrom. Then, initially, no further observation data is necessary at all. However, the requirement criterion for new observation data includes that the initial observation data are older than a maximum age and / or the estimated condition corresponds to a predetermined critical condition that the expert can specify.In other words, unnecessary generation or transmission of observation data can also be prevented on the server side by using the model to determine an estimated condition of a road damage in the server, so that no observation data is necessary to detect the condition. The observed and monitored damage site may have been measured or described once using initial observation data, so that its (real) initial state is known. Based on this, the model can be used to calculate a temporal progression for an enlargement or expansion of the damage site. Such a model can be based on digital data and, for example, operated or implemented by software or program code.

[0026] Using the model, the condition can be estimated, i.e., a virtual condition can be used in the server facility, until the demand criterion is met again to request new observation data. The demand criterion can, for example, stipulate that it is used based on the estimated condition up to a maximum age of the initial observation data, and then (real) observation data can be requested again. Additionally or alternatively, if the estimated condition indicates that the road damage now has a predetermined critical condition, for example, is becoming an obstruction to traffic, a verification of the estimated condition using (real) observation data can be initiated or triggered. The model can influence the influence of weather, which can be described by the weather data (e.g., temperature and / or precipitation and / or frost).The number of motor vehicles driving over the damaged area and causing it to become brittle, for example, can also be taken into account by including traffic data.

[0027] To provide a suitable model, one embodiment provides that the model determines a damage expansion da per number dN of vehicles (e.g., passenger cars and / or trucks) driving over the damaged area as a function of a respective vehicle mass M (model da / dN as a function of M). For example, a model can be used to estimate or simulate crack growth of a crack in the road surface. The more heavy vehicles drive over the damaged area, the faster the damaged area grows, i.e., the greater the rate of damage expansion.

[0028] One embodiment allows for checking in the motor vehicle whether a server device already has correct or sufficiently accurate observation data, eliminating the need to retransmit this data, which would otherwise result in redundant data traffic. For this purpose, the inventory of status data available in the server device is also stored in the motor vehicle and compared with new observation data from the motor vehicle. This status data stored in the motor vehicle is referred to herein as "inventory data."The data reduction device includes, as a restriction function, that in one or more or each of the motor vehicles, inventory data describing the objects already recorded in the server device and / or status data of the respective already recognized status of the respective object is stored, and a processor circuit in the motor vehicle checks whether new observation data determined in the motor vehicle and the inventory data meet a predetermined difference criterion, wherein the requirement criterion of the associated restriction function then includes that the new observation data are only transmitted if the difference criterion is met. Using the difference criterion, the person skilled in the art can therefore specify the extent to which current new observation data may deviate from the inventory data without the new observation data being transmitted from the motor vehicle to the server device.If the difference criterion is met, i.e. the new observation data deviates significantly from the existing data, this new observation data is transmitted to the server device so that it can update the corresponding states of the objects.

[0029] In one embodiment, it is provided that the requirement criterion checks at least one of the following conditions: the new observation data contains image data and an image data transmission is only necessary if the object has not yet been catalogued (i.e. image data transfer only in the case of new damage); a significant change is detected according to a change criterion.

[0030] The existing data can also describe the geometry and / or dimensions of the object, and the new observation data can be used to determine the current geometry and / or dimensions of the object. If the change criterion results in a rate of change (change within a predetermined period of time) that is greater than a threshold and / or if the object has changed beyond a predetermined minimum, this can serve as a trigger for transmitting image data.

[0031] However, even if there are no significant changes on the road section, it has proven advantageous to trigger the request for new observation data based on temporal criteria, for example, in order to rule out a systematic error, which may, for example, consist in changes to the at least one object being signaled by the motor vehicles due to transmission errors. Therefore, in one embodiment, the requirement criterion includes that the observation data present in the server device are recognized as outdated or incomplete according to a predetermined evaluation criterion, and the data reduction device includes, as a restriction function, that in the server facility, measurement orders for obtaining new observation data are distributed among several of the motor vehicles, so that only some of the motor vehicles are expressly commissioned to obtain observation data of a particular type, and / or inspection intervals for obtaining new observation data are timed according to road type and / or traffic density and / or planned monitoring measures for the road section, and / or a comparison of infrastructure data relating to traffic infrastructure components is only carried out in time-limited and time-spaced measurement campaigns and the motor vehicles refrain from transmitting the observation data of the traffic infrastructure components between the measurement campaigns.

[0032] In other words, new observation data can be requested within a time frame. This can be determined by the person skilled in the art using the evaluation criterion, which indicates when observation data available in the server device is considered outdated (because it is older than a predetermined maximum age, for example, older than one day, older than two days, or older than one week) or because it is incomplete, for example because the road section is only described or covered with a certain measurement density or sample density. In order to avoid generating unnecessary data traffic, the data reduction device also ensures that a restriction function is activated or prevents unnecessary new observation data from being omitted. Thus, depending on the type of observation data, only a portion of the motor vehicles can be commissioned to obtain new observation data.The inspection intervals for obtaining new observation data can be determined depending on the load or traffic density of the road section. For traffic infrastructure components that are not subject to wear and tear due to traffic, measurement campaigns can be conducted to monitor their condition or inspect them. However, these can also be separated by measurement breaks, with the measurement campaigns spaced apart. During these measurement breaks, the vehicles refrain from transmitting observation data relating to such traffic infrastructure components.

[0033] To perform condition monitoring of a road section, the invention also provides a system comprising the described server device and a vehicle fleet with multiple motor vehicles. The server device can be configured as a computer or computer network as described and operated, for example, as a server or server platform for the Internet. The vehicle fleet can also be composed of motor vehicles that are additionally operated for another purpose, for example, as a private vehicle or transport vehicle, or as a commercial vehicle used beyond the method.

[0034] The invention also encompasses the system comprising a server device of the type described and a vehicle fleet with several motor vehicles of the type described. A data reduction device is provided in the system and the system is configured to carry out an embodiment of the method according to the invention.

[0035] Also part of the invention is the respective motor vehicle for the system according to the invention, wherein a processor circuit of the motor vehicle is configured to determine observation data relating to at least one object located in a road section and to transmit it to a server device of the system. At least one restriction function of a data reduction device of the system is provided in the motor vehicle, and the respective restriction function only enables the determination and / or transmission of new observation data if a predetermined requirement criterion for new observation data is met. The motor vehicle can be provided as a passenger car or a truck, to name just a few examples.

[0036] The server device for the system according to the invention is a component of the invention, wherein a processor circuit of the server device is configured to determine a current state of at least one object in a predetermined road section based on observation data received from motor vehicles. At least one restriction function of a data reduction device of the system is provided in the server device, and the respective restriction function only enables the determination of new observation data if a predetermined requirement criterion for new observation data is met.

[0037] The respective processor circuit in a motor vehicle or in the server device can each be implemented, for example, on the basis of at least one microprocessor and / or at least one microcontroller. For the method steps to be performed by the motor vehicle or the server device, software or program code with program instructions can be provided which, when executed by the respective processor circuit, cause it to execute the method steps. The program code can be stored in a data memory of the respective processor circuit.

[0038] An embodiment of the invention is described below. It shows: Fig. 1 a schematic representation of an embodiment of the system according to the invention; and Fig. 2 a diagram illustrating a model for determining an estimated state of an object in a road section, the model being stored in a server device of the system of Fig. 1 can be operated.

[0039] The embodiment explained below is a preferred embodiment of the invention.

[0040] Furthermore, the described embodiment can also be supplemented by further features of the invention already described.

[0041] In the figures, functionally identical elements are provided with the same reference numerals.

[0042] Fig. 1 shows a system 10 with a vehicle fleet 11 consisting of several motor vehicles 12 that can drive on the roads 14 of a road network 13. The system 10 can include a server device 15, which can be configured, for example, as a server or an internet server platform 16. The server device 15 can be provided to monitor or check at least one object 17 in the road network 13 for its current state. The state can include checking the presence of the respective object 17 and / or the geoposition of the object 17 and / or its property (size, dirtiness, type, for example, a traffic sign or traffic light, or its appearance). For example, a traffic light 18 and / or a traffic sign 19 and / or road damage 20 can be monitored as an object. A traffic light 18 and a traffic sign 19 each represent an infrastructure component 21.The server device 15 can monitor the objects 17 for those roads 14 of a predetermined, delimited area 22, wherein the roads 14 of the road network 13 in the area 22 represent a road section 23. A road section can thus be a single road or a subnetwork of a road network or a segment of a road, for example, with a length in a range of 100 meters to 10 kilometers.

[0043] The monitored objects 17 can, for example, be mapped in an electronic road map 24, wherein the respective geoposition and the object type can then represent the respective state of the object, and correspondingly mapped objects 17' are available in the road map based on corresponding state data describing the state. Map data 25 of the road map 24 can then, for example, be provided to at least one vehicle so that information about the respective mapped object 17' (ie, the state data) is made available in the respective vehicle (not shown), for example for driver assistance and / or for warning a driver of the vehicle.Within the scope of driver assistance, for example, a spring hardness in a chassis of the vehicle can be adjusted depending on the map data 25 relating to the respective object 17 and / or an autonomous driving function and / or a cruise control can adapt a driving speed of the vehicle depending on the map data 25.

[0044] In order for the server device to be able to provide the map data 25, the respective current state of the respective object 17 must be known, e.g. its geoposition and / or its object type.

[0045] For this purpose, the server device 15 in the system 10 can use the motor vehicles 12. In each motor vehicle 12, a sensor circuit 26 can be provided, wherein Fig. 1 For the sake of clarity, the components described below are only shown for one motor vehicle 12. The at least one sensor circuit 26 can generate a total of sensor data 27, which can be received by a processor circuit 28 of the motor vehicle 12. Based on the sensor data 27, the processor circuit 28 can generate observation data 29, which can be transmitted to the server device 15 via a radio module 30 of the motor vehicle 12, for example a mobile radio module and / or a WLAN radio module, via a communication connection 31. For example, the observation data 29 can be transmitted to the server device 15 via a mobile radio network 32 and an internet connection 33. The server device 15 can generate the map data 25 with the objects 17' mapped therein on the basis of the observation data 29 in a manner known per se.The observation data 29 can, for example, be image data from at least one camera and / or recognition data of a recognition result of a recognition of the objects 17 (for example, geoposition and / or appearance and / or object type), as already described above. The observation data 29 can be generated in a manner known from the prior art.

[0046] The system 10 ensures that no superfluous or undesired observation data 29 are calculated and / or transmitted.

[0047] In order to keep the amount of observation data 29 actually transmitted to a minimum, a data reduction device 35 can be provided in the server device 15 and / or in the respective motor vehicle 12, by means of which one or at least one restriction function 36 is provided in the server device 15 and / or the motor vehicle 12, which can be operated or monitored, for example, by a respective processor device 37 of the server device 15 and / or the processor circuit 28 of the respective motor vehicle 12. The respective restriction function 36 can be used to monitor or check a requirement criterion 38 that must be met in order for new observation data 29 to be generated in the respective motor vehicle 12 and / or for previously generated observation data 29 to be transmitted from the respective motor vehicle 12 to the server device 15.A restriction function 36 can also consist in the server device 15 omitting a request to generate new observation data, so that this also ensures that the motor vehicles 12 do not generate any new observation data.

[0048] In Fig. 2 shows how, starting from initial observation data 29, state data of a simulated or estimated state 40 can be generated by not generating any new additional observation data 29 by the motor vehicles 11 and / or transmitting said data to the server device 15, but rather by determining traffic data 41 for the road section 23, i.e., determining the number of motor vehicles that have driven over the respective object 17 on the roads 14 and, if the object 17 is a road damage 20, using this to determine the current state as an estimated state 40 for the road damage 20. The diagram shows how, depending on a respective mass M of a vehicle or carriage that has rolled over the road damage 20 (represented in logarithmic notation logM), a damage expansion rate 42 results, which results in the damage expansion 43 as da per number dN of vehicles or carriages driving over.This is also plotted logarithmically as log da / dN. Assuming that the cars or vehicles have a mass M in the range between M0 and M1, a so-called Paris line can be used in a model 44, for example, to distinguish three zones 45 of the damage expansion rate 42. The model 44 shown is only an example; a specialist can determine a model through tests and / or simulations depending on the type of road damage 20 and / or the condition of the road surface.

[0049] The condition of a road and its infrastructure can thus be monitored, for example, to plan maintenance measures. The idea here is to utilize the swarm intelligence of all motor vehicles, provided they are equipped with certain components (communication with the server device 15 as the so-called backend, electronic tire pressure monitoring, acceleration sensor, front camera).

[0050] This allows a route network or road network to be analyzed within a specified timeframe and with minimal effort. In a second step, these observation data can be used to supplement an existing road map.

[0051] Preferably, at least one of the following functions is implemented in System 10: Functional group F1: 1) Detection of damage to the road surface 2) Categorization or classification of the damage (crack, pothole, ...) 3) Detection of "non-damage" e.g. speed bumps, curbs, Functional group F2: 4) Detection of roadside traffic infrastructure components (e.g. signs, traffic lights) 5) Categorization or classification of roadside traffic infrastructure components 6) Detection of changes in the roadside infrastructure

[0052] General functional group: 7) Reporting of the respective position and type of damage / sign type 8) Planning of resources / distribution of tasks among the available vehicle fleet using the data reduction device 9) Implementation with low energy / computing and data transmission costs

[0053] Example implementation for F1 road surface monitoring: 1) Detection of road damage / irregularities in the surface structure is achieved via a sensor mounted in the tire (e.g., the eTIS ® system). A vertical acceleration signal Az or a signal from a spring travel sensor is conceivable in addition or alternatively. 2) To classify the damage, the specific acceleration signal nature is used, as well as image data from the front camera of the sensor or sensor circuit 26. For this purpose, the image data is preferably stored periodically in a ring buffer. This is advantageous because the detection range of the camera image does not coincide with the sensor event of an acceleration sensor. If a corresponding event occurs, the camera image corresponding to the position is transmitted to the backend. 3) Classification of the damage via image recognition / AI in the backend (server device 15)

[0054] The following can be provided for image recognition: a. Training the algorithm with a sufficient database of manually classified damage and / or relevant infrastructure elements b. This avoids false alarms, e.g. for speed bumps, curbs 4) Assignment of the severity of the damage via the acceleration signature + damage depth and via the camera image (damage size) 5) Creation of a damage map (e.g. as a road map 24) / recommendations for action for road maintenance departments a. Location (geo-tag or geoposition) of the damage b. Significant change in the damage over time, e.g. taking into account an increase due to frost and / or additional vehicle traffic c. Derivation of a factor for crack propagation speed based on the change in the damage (e.g. increase in area, depth and number of vehicles driving over the damage). (Analogous to the state-of-the-art modeling of crack propagation in metals da / dN vs. deltaK).An example model 44 is shown in . Fig. 2 shown. Analogies to the crack propagation model can be: dK corresponds to M, the vehicle mass: Two classes of cars / trucks and / or the application for the right lane of a motorway / remaining lanes or a mixed calculation and / or a dependence on the damage geometry are conceivable here. da - corresponds to the crack growth per vehicle dN - number of vehicles passing over

[0055] The extent of damage can be estimated using a material-specific parameter and the frequency of overrun of the damaged area (route utilization).

[0056] Example implementation for F2 traffic signal monitoring: 1) Camera recording of the road signs via front camera 2) Interpretation of the signs via image recognition algorithm in the vehicle 3) Sending the sign type with geo-tag (e.g. geoposition) to the backend to generate a database

[0057] Example implementation for reducing energy / computation and data transmission costs: 1) Route sections are only inspected periodically by one (or a smaller group of) vehicles. The query is triggered via an algorithm in the backend. a. Only minimal impact on the battery life of the wheel sensors (a higher sampling rate is necessary for the scanning process) b. Avoiding multiple inspections of the same damage c. Image data is only transmitted if necessary + not already cataloged 2) Image data is only transmitted to the backend in the case of new damage / significant deterioration of the damage: a. for the initial classification of the damage, b. for documenting the damage development. 3) Damage tracking is possible through flexibly defined inspection intervals, e.g. frequently used roads are inspected more often, b. The focus of the inspection can be set after repairs and / or changes to the route and / or a predefined criterion.4) At defined intervals, comparison of the infrastructure data (signs and / or other traffic infrastructure components) with current data from vehicles: a. Sign, position and type match the database (existing data), then no action, only update of the timestamp for the last update; b. New sign on the route, then transmission of the type and position to the backend; c. Expected sign not found, then image data from the front camera to the backend and, for example, check by the road maintenance department whether the sign was removed improperly.

[0058] Damage classification via an ML / AI model (e.g., an artificial neural network) can consist of image data and acceleration signature as input, and the output is the damage type as a class (e.g., crack in the road surface, pothole) and damage severity (size of damage, depth of damage). A statistical output can be modeled using Model 44 as a damage propagation rate, analogous to crack propagation speed in metals.

[0059] This advantageously enables long-term investigation of road conditions for maintenance reasons, estimation of maintenance intervals using failure models, and so-called "predictive maintenance." To select the appropriate image from the ring buffer, the exact delta t (the moment the wheel passes over the damage (FL - front left / FR - front right / RL - rear left / RR - rear right) and the time of capture by the front camera are calculated from the wheel speed and dynamic wheel radius and / or the vehicle speed.

[0060] The use of a "swarm" (vehicle fleet) enables data reduction for individual vehicles, namely, distributing the payload or data generation across many individual vehicles, thus avoiding redundant data recording. Not every damage is necessarily detected by every vehicle activated for the respective route (different track gauges, trajectories, to name just a few examples), but a sufficient number of vehicles in a network can reliably detect damage.

[0061] The server device 15 enables cloud-based or network-based functions for condition monitoring and the use of machine learning (ML) and artificial intelligence (AI).

[0062] Overall, the example shows how fleet-based condition monitoring of a road surface can be provided.

Claims

1. Method for controlling fleet-based condition monitoring of a road section (23) of a road network (13), wherein the fleet-based condition monitoring comprises determining observation data (29) concerning at least one object (17) located in the road section (23) by means of motor vehicles (12) in a vehicle fleet (11) and transmitting said data to a central server device (15) and the server device (15) determining a current condition of the at least one object (17) on the basis of the observation data (29), wherein a data reduction device (35) reduces an amount of observation data (29) to be transmitted by operating at least one restriction function (36) in one or some or all of the motor vehicles (12) and / or in the server device (15) and the respective restriction function (36) triggering a determination and / or transmission of new observation data (29) only if a predetermined requirement criterion (38) for new observation data (29) is met, characterized in that at least one object (17) is road damage (20) and the data reduction device (35) comprises, as a restriction function (36), operating a model (44) for a temporal progression of a spread of damage (da, 43) of the road damage (20) in the server device (15) starting from initial observation data, which model, on the basis of weather data and / or traffic data (41), estimates a growth of a damage site constituting the road damage (20) in a road surface of the road section (23) and from this determines an estimated condition (40) of the road surface, and the requirement criterion (36) for new observation data involves the initial observation data being older than a maximum age and / or the estimated condition matching a predetermined critical condition.

2. Method according to Claim 1, in which, in order to provide sensor data (27) as observation data (29), data are taken both from a camera and from an acceleration sensor, image data from the camera are buffered in a buffer and, if a predetermined trigger pattern is caused by an unevenness in the road surface in acceleration data from the acceleration sensor, the acceleration data containing the trigger pattern and the corresponding image data from the buffer, in which the unevenness is represented, are combined as sensor data (27) for a classifier.

3. Method according to Claim 2, wherein sensor data (27) both from a camera and from an acceleration sensor are combined, and wherein a time offset between the formation of these sensor data is present and is compensated for.

4. Method according to Claim 2 or 3, wherein the image data corresponding to the acceleration data are selected from the intermediate buffer: a) based on a current driving speed and a time offset between the image data and the time at which the unevenness is driven over and / or b) based on a wheel speed of at least one wheel and a radius value of a radius of the respective wheel.

5. Method according to Claim 1, wherein the model (44) determines a spread of damage (da, 43) per number (dN) of the vehicles driving over the damage site on the basis of a respective mass (M) of the respective vehicle.

6. Method according to one of the preceding claims, wherein the data reduction device (35) comprises, as a restriction function (36), respectively keeping inventory data, which describe the objects (17') already recorded in the server device (15) and / or condition data relating to the respective already detected condition of the respective recorded object (17'), stored in one or more or each of the motor vehicles (12) and a processor circuit (28) in the motor vehicle (12) checking whether new observation data (29) determined in the motor vehicle (12) and the inventory data meet a predetermined difference criterion, and the requirement criterion (38) for new observation data involves the new observation data (29) being sent only in this case.

7. Method according to one of the preceding claims, wherein the requirement criterion (38) for new observation data comprises recognizing, in the server device (15), the observation data (29) already available there as obsolete or incomplete according to a predetermined evaluation criterion and the data reduction device (35) comprising, as a restriction function (36), • dividing measurement orders for acquiring new observation data (29) between a plurality of the motor vehicles (12) in the server device (15), such that only some of the motor vehicles (12) are explicitly instructed to acquire observation data (29) of a respective type, and / or • clocking inspection intervals for acquiring new observation data (29) according to road type and / or traffic density and / or planned monitoring of measures for the road section (23), and / or • comparing infrastructure data relating to traffic infrastructure components (21) only in measurement campaigns that are of a limited time and are separated from one another in terms of time, and the motor vehicles (12) refraining from sending the observation data (29) relating to the traffic infrastructure components (21) between the measurement campaigns.

8. Method according to one of the preceding claims, wherein the respective object (17) is road damage (20) or a traffic infrastructure component (21).

9. Method according to one of the preceding claims, wherein the data reduction device (35) comprises, as a restriction function (36), respectively operating a classifier in one or more or each of the motor vehicles (12) and / or in the server device (15), which classifier receives sensor data (27) from at least one sensor circuit (26) of the motor vehicle (12) and, on the basis of the sensor data (27), classifies an object (17) which is represented in the sensor data (27) and is located in the road section (23), and thereby detects an object class of the object (17), wherein the requirement criterion (38) for new observation data comprises, in particular, using an assignment rule to specify for which of the object classes, for whose detection the classifier is configured, observation data (29) are transmitted and / or for which observation data are not transmitted.

10. System (10) comprising a server device (15) and a vehicle fleet (11) having a plurality of motor vehicles (12), wherein the server device (15) is configured to monitor a current condition of at least one object (17) in a road network (13) on the basis of observation data (29), and wherein each of the motor vehicles (12) is respectively configured to determine the observation data (29) concerning at least the object (17) located in the road section (23) and to transmit said data to the server device (15), wherein a data reduction device (35) is provided in the system (10), characterized in that the system (10) is configured and intended to perform a method according to one of the preceding claims.

11. System (10) according to Claim 10, wherein a processor circuit (28) of the motor vehicle (12) is configured to determine the observation data (29) concerning at least the object (17) located in the road section (23) and to send said data to the server device (15) of the system (10).

12. System (10) according to Claim 10, wherein a processor circuit (37) of the server device (15) is configured to determine a current condition of the at least one object (17) in a predetermined road section (23) based on the observation data (29) received from the motor vehicles (12).