DE102017213983A1 - Method for verifying a road map and server device - Google Patents

Abstract

The invention relates to a method for verifying a road map (13), wherein by a server device (10) from at least one motor vehicle (14) respective trajectory data (24) of a motor vehicle (14) traversed, card-illegal Fahrtrajektorie (20) are received along a in the road map (13) missing piece of track (27) leads, and then checked on the basis of a predetermined evidence criterion (38), whether the road map (13) with respect to the path piece (27) is to be checked. If the evidence criterion (38) is fulfilled, a possible faultiness of the road map (13) is signaled. The invention provides that, for a predetermined area (32) comprising the path section (27), trajectory data (24) for at least one reference vehicle (17) is converted to a map-compliant driving trajectory (23), to which no deviation from the road map (13) is detectable, are available. The evidence criterion (38) comprises that a quantity value (34) of the card-adverse driving trajectories (20) and a quantity value (35) of the card-conforming driving trajectories (23) must fulfill a predetermined ratio condition (39).

Description

The invention relates to a method for verifying a road map, i. It is checked whether the road map completely maps or maps a predetermined road network. The invention also includes a server device for operating on a data network, for example the Internet. By means of the server device, the method according to the invention can be carried out.

A method for verifying a road map is, for example, from US 7,024,307 B2 known. According to the known method, a motor vehicle can detect whether it is moving along a driving trajectory that is not recorded in a digital road map of the motor vehicle. In this case, the motor vehicle can send differential data to a server device, which collects these as well as further difference data from other motor vehicles. If a predetermined minimum amount of difference data with a predetermined minimum quality indicates that a road section is actually missing or not mapped in the road map, an update of the road map can be triggered.

A disadvantage of this method is that the road map is also changed when the difference data come from such vehicles whose drivers are not driven on an official public road, but take, for example, a shortcut on an unofficial stretch that is not in the Road map should be mapped. An example of such an unofficial stretch is a parking lot or meadow over which a driver abbreviates a path. To interpret a lack of such an unofficial stretch of road in the road map as a mistake would be a false alarm in the verification of the road map.

A driving trajectory in the sense of the invention is a position sequence, ie a sequence of position information of such positions, which were taken in succession by a motor vehicle. Another term for such a position sequence or sequence of position information is Trace. To check whether such a position sequence deviates from a road map, a so-called map-matching method can be applied, as for example from the DE 10 2010 028 090 A1 is known. According to the known map matching method, to detect a deviation of a position sequence from a road map, a confidence level indicating how relevant a deviation is can be calculated.

If several motor vehicles transmit their driving trajectories to a server device, it must be possible to detect in the server device which driving trajectories belong together, ie lead via the same route or road segment. This is from the DE 10 2013 208 521 A1 A method is known that associates driving trajectories with each other to group all those driving trajectories that travel the same distance.

The invention has for its object to reduce a false alarm rate when verifying a road map, which may arise because individual motor vehicles report a deviation of their driving trajectories from the road map, although this is the driving of an unofficial stretch of road.

The object is solved by the subject matters of the independent claims. Advantageous developments of the invention are given by the dependent claims, the following description and the figure.

The invention provides a method for verifying a road map. The method may be performed by a server device. From at least one motor vehicle, respective trajectory data of a travel trajectory traveled by the motor vehicle is received. This is referred to below as a card-illegal driving trajectory, because it leads along a not cartographed in the road map, so a missing piece of land. The trajectory data can be received from the motor vehicle by the server device, for example via a mobile radio connection and / or an internet connection.

On the basis of a predetermined evidence criterion is checked whether the road map with respect to the path to correct or at least check. If the evidence criterion is met then a possible faultiness of the road map is signaled. On the basis of the evidence criterion, it is thus recognized whether an off-board driving trajectory, such as that traveled or driven by the at least one motor vehicle, constitutes evidence or proof that the road map is faulty or incomplete because the path in the road map is not mapped , This is then signaled by the server device as a possible faultiness of the road map.

It is known from the prior art described at the outset how to count how many card-adverse driving trajectories are present in order to determine the relevance of the deviation. If, however, several motor vehicles are driven along a map-unfavorable driving trajectory, which leads via an unofficial part of the route, which is not even to be mapped in the road map, the simple counting of card-adverse driving trajectories can be used cause a false alarm, ie the road map is changed by completing the path, although this is not intended.

According to the invention, therefore, instead of checking at a predetermined area that includes the part of the path in question, trajectory data for a map-compliant travel trajectory are available from at least one further motor vehicle. A map-compliant driving trajectory is characterized by the fact that no deviation from the road map can be determined for it. In the following, the at least one further motor vehicle is designated for better distinction as the respective reference vehicle. It is thus determined whether and if so how many reference vehicles each have traveled in the area along a map-compliant driving trajectory, which therefore does not include the missing part of the route. The area in which testing is carried out may be determined, for example, by a predetermined corridor around the missing part of the route.

If at least one card-invalid driving trajectory, which leads across the non-cartographed path, is known and if the number of available card-compliant driving trajectories is likewise known, then a quantity value of the card-adverse driving trajectories, ie their number, and a quantity value of the card-conforming driving trajectories, ie For example, their number can be determined. The said evidence criterion, by means of which it is determined whether the road map is to be checked with regard to the path, comprises according to the invention that the quantity value of the card-adverse driving trajectories and the quantity value of the card-conforming driving trajectories must fulfill a predetermined ratio condition. Thus, the quantity value of the card-adverse driving trajectories is set by the ratio condition in relation to or in relation to the quantity value of the card-conforming driving trajectories. In this way, it is recognized how many reference vehicles could travel along the already mapped road network and how many motor vehicles drove in relation to it along card-illegal driving trajectories over the non-mapped path section. The same period can be used to examine these two cases.

The advantage of the invention is that it is not the absolute number of card-invalid driving trajectories that determines whether the road map is classified as faulty, but rather how many driving trajectories are present overall in the area of the route and how many of them are related to the route Total have dissipated over the stretch. As a result, the false alarm rate can be reduced in an advantageous manner. If, for example, 10,000 motor vehicles travel along a cartographic road network per day and, in comparison to this, deviate 100 motor vehicles with their card-adverse driving trajectories, a correct road map can still be assumed. In comparison, for the same number of 100 unlucky driving trajectories in an area that has been traversed by as few as 105 motor vehicles in total, it can be assumed that the road map is faulty.

The invention also includes additional embodiments which provide additional advantages.

In order to obtain from the at least one reference vehicle respective trajectory data for a map-compliant driving trajectory, one can generally provide that each motor vehicle transmits the trajectory data of all its trajectories to the server device. This then results in a correspondingly large amount of data to be transmitted. In order to be able to transmit trajectory data to map-compatible driving trajectories in a targeted manner only, an embodiment provides for checking whether trajectory data for a map-compliant driving trajectory are available to which at least one reference vehicle is sent a transmission command, which transmits the reference vehicle for transmitting the trajectory data from the reference vehicle respectively in the area trajectory instructed. In other words, such trajectory data of map-compliant driving trajectories are determined or requested only when necessary, when a non-cartographed path has been detected. In addition, the request can be limited to the area, for example, by sending the transmission command only to each such reference vehicle, which is located in the area or is approaching. As a result, the data volume of the trajectory data to be transmitted can be kept low.

Said ratio condition, which is provided for the quantity values of the card-adverse and card-conforming driving trajectories, according to one embodiment, comprises the ratio of the quantity value of the card-adverse driving trajectories to the sum of the quantity values of the driving trajectories (card-adverse as well as card-conforming driving trajectories) being greater than a predetermined threshold value. For example, the threshold may range from 30% to 70%. Thus, therefore, the proportional condition is based on the percentage of card-adverse driving trajectories. As a result, statistical deviations of individual driving trajectories can be compensated for in an advantageous manner.

Since a road network can change, for example, due to reconstruction measures, it should not be based on any arbitrary old trajectory data for checking the ratio condition. According to According to one embodiment, a time stamp is determined for each of the driving trajectories, which determines the detection date of the respective driving trajectory, in accordance with the method for determining the quantity value of the card-adverse driving trajectories and / or the quantity value of the card-conforming driving trajectories. It can therefore be specified the day on which the respective driving trajectory was measured or observed. Only travel trajectories from a predetermined time window are then taken into account. The time window preferably starts on the current day or in the current week and goes back to the past by a predetermined period of time. The duration of the time window can be in the range of one day to one week, or one month or one year. As a result, it is advantageously taken into account that road construction measures may be added to a path which has not been previously mapped in the road map. Driving trajectories from the past before the construction of the section, which can then always be only map-compliant driving trajectories, are then outside the time window and are then excluded in the review of the ratio condition.

Previously, the quantity value of the respective driving trajectories (card-compliant and card-adverse) was specified as the respective number. However, it is preferably provided that when the respective quantity value is formed, the respective age of the respective driving trajectory is taken into account as well, the older the driving trajectory is, the smaller the contribution of the respective driving trajectory to the quantity value. Each driving trajectory is thus weighted into the quantity value, whereby the weighting can be formed by a multiplicative factor, which is the smaller, the older the driving trajectory is. As a result, the response time of the server device to an actual change in the road network can be advantageously reduced. The weighting may generally be done with a weighting function selectable by the skilled person, e.g. with a sigmoid function or simply linear. In principle, any possible weighting function can be provided, in which the weighting takes place depending on the age.

If the possible faultiness of the road map is signaled in the manner described, because the criterion of evidence is satisfied with its ratio condition, according to a further development, preferably the map data of the road map is supplemented by a description of the path section on the basis of the trajectory data of at least one card-illegal driving trajectory. The map data thus supplemented are then preferably transmitted to the at least one motor vehicle and / or the at least one reference vehicle. Generally, the supplemented map data is sent out as a map update to at least one arbitrary receiver vehicle or provided for this. This results in the advantage that the road map is automatically updated by recognizing a non-mapped part of the road through the server device. Additionally or alternatively, a respective notification signal for indicating the defective in the area road map and / or preconditioning at least one of the road map and / or a copy of the road map dependent vehicle function can be sent to at least one receiver vehicle. Thus, the at least one vehicle function can be prepared for a deviation of the road map from the actually existing road network. The "copy of the road map" refers to map data that the respective receiver vehicle holds onboard as a copy. The at least one vehicle function may include an infotainment system (information entertainment system) and / or a driver assistance system and / or a navigation system.

Several embodiments of the invention relate to the question of how a card-adverse driving trajectory can be detected.

One embodiment provides that the card-adverse driving trajectory of the respective motor vehicle is detected by assigning a position sequence describing successive positions of the motor vehicle to roads mapped in the road map by means of a predetermined map matching method. In contrast to the prior art, it is now provided that this map matching is not only able to find positions in the road map, which can be assigned to the actual positions of the motor vehicle, but also positions as "offroad" (outside the known Road network) if this assignment is possible only with a probability less than a predetermined minimum probability. Thus, an assignment result is identified which satisfies a predetermined corresponding off-road condition. The map-matching method is thus limited in that it does not map each position of the motor vehicle to a street mapped or registered in the road map, but at a predetermined deviation of the position of each mapped road classifies this position as "off-road". The deviation can be specified by the specialist due to the off-road condition.

In order to be able to distinguish between probable, irrelevant deviations, such as may be caused by a scattering of the vehicle sensors of the position detection, and a relevant deviation due to a driving of a non-mapped route, an embodiment provides that the map matching method is based on a Hidden Markov model is based, which is a scattering of the vehicle sensors through which the consecutive positions of the motor vehicle are detected, modeled. This advantageously reduces a false alarm rate with respect to an off-road condition. Examples of said vehicle sensors are each an odometry and a GNSS location (GNSS - Global Navigation Satellite System), such as a GPS location (GPS - Global Positioning System) or a location based on GLONASS or Galileo or Beidou.

The off-road condition itself, according to one embodiment, comprises an assignment probability of the underlying position sequence to the nearest cartographed road being less than a predetermined threshold value. This assignment probability can for example be formed or calculated on the basis of said Hidden Markov Model (HMM). A hidden Markov model concludes observations (observables, that is, the known positions of sensors) with known statistical methods on hidden states ("suspected positions" on the road network). An HMM is thus a statistical model for the assumed "true" position sequence itself, which is used to find the most likely correct assignment for the said map matching of the observed position sequence to the "true" position sequence. The use of an assignment probability has the advantage that statistical fluctuations or variations can be taken into account.

A refinement provides that any position sequence which fulfills the off-road condition is not classified immediately or unconditionally as a card-invalid driving trajectory, but is first assigned a relevance value by means of a predetermined evaluation function. Only such a position sequence whose relevance value is greater than a predetermined threshold value is then classified as card-adverse driving trajectories. If the described check takes place in the motor vehicle itself, then the trajectory data of this position sequence, which is classified as card-adverse driving trajectory, can then be transmitted to the server device. If, instead, the checking of the position sequence already takes place in the server device, then the trajectory data of this position sequence can be reused for the described checking of the road map.

One embodiment provides said evaluation function. In this case, the evaluation function generates an even smaller relevance value the closer the positional sequence is to a cartographed road and / or the smaller a difference between a shape or a shape described by the positional sequence and a predetermined maneuvering pattern. The latter difference can be calculated using a method of pattern matching known from the prior art. By checking for at least one predetermined maneuvering pattern, there is the advantage that previously known, typical driving maneuvers which can lead to a false alarm or a misjudgment of the positional sequence are excluded. An example of such a maneuver is a parking maneuver.

So far, it has only been described to identify a single, non-mapped stretch in a road map. Preferably, however, trajectory data of several, different card-illegal driving trajectories are received. Thus, in order to advantageously detect a plurality of non-cartographed stretches, associated travel trajectories which lead via the same non-mapped stretch of road are recognized or combined by means of a clustering method.

In order to carry out the method according to the invention, the server device is provided in the manner described for operation on a data network. The data network may be, for example, the Internet or include. The server device may be formed by a computer or by a computer network. The server device thus has a computing device that is configured according to the invention to perform an embodiment of the method according to the invention.

Further features of the invention will become apparent from the claims, the figures and the description of the figures. The features and combinations of features mentioned above in the description as well as the features and feature combinations mentioned below in the description of the figures and / or shown alone in the figures can be used not only in the respectively indicated combination but also in other combinations or in isolation.

The invention will now be described with reference to a preferred embodiment and with reference to the drawings. It shows the sole figure (Fig.) An embodiment of the server device according to the invention in carrying out an embodiment of the method according to the invention.

The figure shows a server device 10 , which may be, for example, a computer or a computer network. The server device 10 can connect to a data network 11 connected, which may be, for example, the Internet. Furthermore, in the figure, a road network comprising several streets 12 shown as it is in a road map 13 can be mapped. The road map 13 can for example, be provided as a database. In the road network 12 can motor vehicles 14 . 15 . 16 . 17 drive. Through a respective position sequence 18 individual positions 19 the respective motor vehicle results in a total of a respective Fahrtrajektorie 20 . 21 . 22 . 23 of the respective motor vehicle 14 . 15 . 16 . 17 , For the sake of clarity, the individual positions 19 a position sequence 18 only for the driving trajectory 20 of the motor vehicle 14 shown. The motor vehicles 14 . 15 . 16 . 17 Thus, each can represent a plurality of motor vehicles, based on a similar Fahrtrajektorie 20 . 21 . 22 . 23 drive along each. Of course, the individual, individual driving trajectories may differ, because not every motor vehicle in the road network travels on exactly the same position sequence. A clustering method can be used to associate related travel trajectories that run along the same road or lane.

The motor vehicles 14 . 15 . 16 . 17 can to the server device 10 respective trajectory data 24 send out. The trajectory data 24 can each, for example, via a mobile phone connection 25 or a wireless LAN connection (WLAN - Wireless Local Area Network) and / or via an Internet connection 26 to the server device 10 be transmitted. The trajectory data 24 can the respective driving trajectory 20 . 21 . 22 . 23 of the respective motor vehicle 14 . 15 . 16 . 17 describe or specify.

The server device 10 can be based on the trajectory data 24 the road map 13 Verify, so check or determine if the road map 13 the road network 12 completely mapped or mapped or described.

To illustrate the operation of the server device 10 are two scenarios in the figure S1 and S2 illustrated.

In the scenario S1 drives the motor vehicle 14 along a stretch of road 27 that in the card 13 not registered or mapped. The driving trajectory 20 of the motor vehicle 14 is thus card-bad. In other words, the course or the shape of the card-adverse driving trajectory gives way 20 of the possible course of the road network 12 off, as he is in the road map 13 is described. The card-illegal driving trajectories 20 So has an off-road section 28 on, at which the positions 19 the position sequence 18 to the road of the cartographic roads of the road network 20 a difference 29 which is greater than a minimum value.

This compares to the scenario S1 the car 17 exclusively on cartographed roads of the road network 12 , so that a map-compliant driving trajectory 23 results.

At the way piece 27 it may be a newly built or newly opened road, for example a turning lane, as illustrated in the figure. This path 27 should therefore be in the road map 13 be entered or mapped.

In the scenario S2 puts the motor vehicle 15 also an unlawful driving trajectories 21 back, but over a meadow or a parking lot 30 away, for example, to avoid a turn at the junction shown, so for example to drive around a traffic light. The parking lot 30 can this two road accesses 31 exhibit. The parking lot 30 So it represents an unofficial stretch that is not part of the road network 12 in the street map 13 should be mapped.

To the scenario S2 also includes the map-compliant driving trajectory 22 of the motor vehicle 16 , which performs the said, prescribed turning operation at the illustrated intersection.

Through the server device 10 can now to the scenario S1 be recognized that the card-illegal driving trajectories 20 an indication of a missing path 27 in the street map 13 is. This can be distinguished by the server device 10 the card-illegal driving trajectory 21 Not as an extra part of the road map 13 should be mapped.

The server device 10 compares to this in the scenario S1 the frequency with which motor vehicles such as the motor vehicle 14 along a card-illegal driving trajectory 20 drive along the stretch of road 27 leads, with the frequency of card-compliant driving trajectories 23 , The car 17 and other motor vehicles with card-conform driving trajectory 23 thus represents a reference vehicle.

The card-compliant driving trajectories are used to determine the frequencies 23 in a predetermined area 32 around the path 27 selected around. The server device 10 may be used to determine the associated trajectory data 24 to reference vehicles such as the reference vehicle 17 a send command 33 to send out their trajectory data 4 output.

For the scenario S1 can therefore be a number or generally a value set 34 the total determined card-illegal driving trajectories 20 and a number or generally a quantity value 35 the one in the area 32 observable map-compliant driving trajectories 23 be determined. Accordingly, to the scenario S2 a number or quantity 36 the card-illegal driving trajectories 21 and a number or generally a quantity value 37 the card-compliant driving trajectories 22 be determined.

An evidence criterion 38 can specify that a respective ratio of the quantity value 34 . 36 to the corresponding quantity value 35 . 37 or to the sum of the quantity values a predetermined ratio condition 39 must meet.

At satisfied ratio condition 39 it then turns out that the recognized stretch of road 27 in the street map 13 missing or the road map 13 at least a missing part of the way 27 should be checked. In contrast, the parking lot 30 be recognized as a rarely used, unofficial stretch, because it can be assumed that in proportion very few drivers will use the parking lot as a shortcut.

Thus it can be prevented, then the parking lot 30 or generally an unofficial stretch as part of the navigable roads in the road map 13 is recorded. For the basis of navigation assistance in modern vehicles are digital road maps, in which the traffic route network or road network is represented for example in the form of semantic, geometric and topological information. A navigation assistance should only include routes that do not have unofficial routes, such as the parking lot 30 , to lead. Therefore, such unofficial stretches (recognizable by the injured evidence criterion) should not be considered a navigable option in the road map 13 be recorded or registered.

A particularly preferred embodiment is described below. It provides the location information or positions obtained with the help of vehicle sensors (eg odometry, GPS) 19 using a special so-called Map-Matcher to map the road map in the vehicle. In this case, this map-matcher is able to find points on a mapped road not only by means of a suitable method (such as a Hidden Markov model), which with a certain probability, the input data (location information, positions 19 ) can be assigned; but also to evaluate positions as "off-road" if, for example, after carrying out predetermined plausibility considerations, no suitable road can be assigned (determined by an off-road condition).

In a further step, the positions marked as "offroad" are collected and evaluated. The score sorts the collected positions into so-called "hotspots" (position sequence that satisfies the off-road condition), where a hotspot is given by a sequence of off-road positions and corresponds to an event possibly involving a map error. An evaluation function assigns each hotspot a "score" or relevance score that expresses the likely likelihood of a card failure. The relevance value includes how similar the sequence of off-road positions 19 geometric to the nearest route or street on the road map 13 is and whether the position sequence 18 off-road positions 19 has a form that is similar to frequent "false positives" (eg "short cut over private property").

1. a) Weiterverarbeitung im Kraftfahrzeug: Weist der Relevanzwert eines Hotspots einen ausreichend hohen Wert auf, kann dem Fahrer angezeigt werden, dass die Straßenkarte an dieser Stelle möglicherweise veraltet bzw. fehlerhaft ist. Ist die Qualität und Quantität der ermittelten Positionen ausreichend hoch und der Relevanz deutet mit sehr hoher Wahrscheinlichkeit auf einen Kartenfehler hin, können die Geometrie und auch die Topologie der betroffenen Straße in der Straßenkarte mit Hilfe der erfassten neuen Positionen 19 aktualisiert werden. Dabei ist allerdings zu beachten, dass nicht alle in der Onboard-Karte des Kraftfahrzeugs enthaltenen Informationen ebenfalls aktualisiert werden können, da einige nur durch spezielle Messfahrzeuge und ggf. manuelle Nachbearbeitung erfasst werden können.
2. b) Weiterverarbeitung im Backend (Servervorrichtung 10): Die ermittelten Hotspots können über Mobilfunk an das durch die Servervorrichtung 10 gebildete Backend als Trajektoriendaten 24 übertragen werden. Dort können Positionsfolgen, die sich auf den gleichen Hotspot beziehen, z.B. mit Hilfe eines geeigneten Clustering-Verfahrens als zusammengehörig erkannt und gruppiert werden. Werden nun bezüglich eines bestimmten Hotspots weitere Trajektoriendaten von anderen Fahrzeugen ausgewertet, auch und insbesondere von solchen, die dort keinen Hotspot erkannt haben (etwa, weil sie im Rahmen der Sammlung von Flottendaten ohnehin empfangen wurden oder indem Referenzfahrzeuge gezielt angewiesen wurden, an diesen Stellen Positionsfolgen zu sammeln und an das Backend zu senden), dann lässt sich auch eine „negative Evidenz“ zu einem Kandidaten für einen Kartenfehler finden. Betrachtet man nun die Hotspot-Meldungen als eine „positive Evidenz“ und setzt diese ins Verhältnis zu der vorliegenden „negativen Evidenz“, dann erhält man ein Maß dafür, wie groß der Anteil der Fahrzeuge war, die einen Kartenfehler „sahen“ im Verhältnis zur Menge aller Fahrzeuge, die in einem bestimmten Zeitraum in der Vergangenheit die betreffende Straße bzw. das betreffende Straßensegment durchquert haben.

Then there are two options:

1. a) Further processing in the motor vehicle: If the relevance value of a hotspot has a sufficiently high value, the driver may be informed that the road map at this point may be out of date or faulty. If the quality and quantity of the determined positions are sufficiently high and the relevance indicates with very high probability a map error, the geometry and also the topology of the affected road in the road map can be determined with the help of the detected new positions 19 to be updated. It should be noted, however, that not all information contained in the onboard map of the motor vehicle can also be updated, since some can only be detected by special measuring vehicles and possibly manual post-processing.
2. b) further processing in the backend (server device 10 ): The detected hotspots can be sent via mobile to the server device 10 formed backend as trajectory data 24 be transmitted. There, position sequences that refer to the same hotspot, for example, using a suitable clustering method can be recognized as belonging together and grouped. If further trajectory data are evaluated by other vehicles with regard to a particular hotspot, and especially by those who have not recognized a hotspot there (for example, because they were received in the context of the collection of fleet data anyway or by reference vehicles were specifically instructed, position sequences at these locations to collect and send to the backend), then a "negative evidence" can be found to a candidate for a map error. If one considers the hotspot messages as a "positive evidence" and sets them in relation to the present "negative evidence", then one gets a measure of how large the proportion of vehicles that "saw" a card error in the Ratio of the number of vehicles that have crossed the road or segment in question in a given period in the past.

If this percentage is large enough, this significantly increases the likelihood of a true card error. If, on the other hand, the proportion is low, then there may have been only a malfunction of the sensors of individual vehicles or a temporary obstacle has been avoided.

The window in the past being considered can be weighted with a function of arbitrary shape; it is possible, for example, a sigmoid function so that recent events are weighted higher than older ones. This results in the said quantity values. The width of this window makes it possible to shift the sensitivity of the card error detection between the two poles "high inertia" (very wide window, ie changes are not recognized until they have existed for a long time) and "high reaction speed" (very narrow window, ie changes are detected very quickly). If the window is very narrow, changes in the course of the road will be detected quickly, but may not be considered very safe, since the statistical base (number of vehicles reporting the change) is still too small and, if necessary, the change is short-lived consists. If the window is very wide, the statistical base becomes large and it can be assumed with more certainty that it is a (semi-) permanent change of the road guidance; At the same time, however, the change will only be recognized as such relatively late. The window must therefore be adjusted meaningfully between those two poles by the expert.

The hotspots considered to be the most likely candidates for a card error after evaluation and classification (i.e., as described: classification into certain error classes and, in particular, rejection of more frequent "false alarms") can now be considered e.g. forwarded to the card manufacturer, who can make targeted a remeasurement with his measuring vehicles. Alternatively (analogously to the procedure in the above scenario a)) hot patches or map updates for the maps with modified geometry can be sent to the vehicles or even the information is transmitted that a map error is present so that this can be displayed to the drivers.

Advantage of Scenario a): Users get the impression that their card "learns", i. E. after a while recognizes changed street directions and displays them correctly. If navigation over a certain route is not consistently possible due to outdated map data, the driver can be informed. Infotainment, driver assistance, and other subsystems may be preconditioned appropriately with the knowledge that the map data is wrong for a particular section of the route.

Advantage of Scenario b): Card manufacturers can fully automatically look for promising candidates for map errors or obsolete locations in map maps of the road map 13 be notified and then (or possibly after a manual sighting) make a remeasurement. In the process, the places where a particularly large number of motor vehicles are confronted with the faulty image of the obsolete map are more likely to be reported, as more of these posts are received. In this way, the most important errors can be remedied most quickly. Updated map sections can then be sent out to the motor vehicles, so that all navigation functions are available again via the changed routes.

Overall, thus trajectory data from motor vehicles can be used to detect and repair map errors.

LIST OF REFERENCE NUMBERS

10

server device

11

Data network

12

road network

13

road map

14

motor vehicle

15

motor vehicle

16

reference vehicle

17

reference vehicle

18

position sequence

19

position

20

Map-contrary driving trajectory

21

Map-contrary driving trajectory

23

Map-compliant driving trajectory

22

Map-compliant driving trajectory

24

trajectory

25

radio link

26

Internet connection

27

span

28

Off-road part

29

difference

30

parking spot

31

driveway

32

area

33

send command

34

volume value

35

volume value

36

volume value

37

volume value

38

evidence criterion

39

ratio condition

S1

scenario

S2

scenario

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The 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

• US 7024307 B2 [0002]
• DE 102010028090 A1 [0004]
• DE 102013208521 A1 [0005]

Claims (13)

A method for verifying a road map (13), wherein by a server device (10) from at least one motor vehicle (14) respective trajectory data (24) of a motor vehicle (14) traversed, card-illegal driving trajectory (20) are received, the card-illegal Fahrtrajektorie (20) along a in the road map (13) missing piece of land (27) leads, - is checked on the basis of a predetermined evidence criterion (38), whether the road map (13) with respect to the stretch (27) is to check, and - if the evidence criterion (38) a possible faultiness of the road map (13) is signaled, characterized in that - to a predetermined, the path (27) comprising area (32) is checked whether from at least one reference vehicle (17) each trajectory data (24) to a map-compliant driving trajectory (23), to which no deviation from the road map (13) can be detected, are available and - the Eviden zkriterium (38) comprises that a quantity value (34) of the card-adverse driving trajectories (20) and a quantity value (35) of the card-conforming driving trajectories (23) satisfy a predetermined ratio condition (39). Method according to Claim 1 in which, for checking whether trajectory data (24) for a map-compliant driving trajectory (23) are available, to the at least one reference vehicle (17) a transmission command (33) which the reference vehicle (17) for transmitting the trajectory data (24) of the Reference vehicle (17) each instructed driving trajectory (23) instructs, is sent out. Method according to one of the preceding claims, wherein the ratio condition (39) comprises that the ratio of the quantity value (34) of the card-adverse driving trajectories (20) to the sum of the two quantity values (34, 35) is greater than a predetermined threshold value. Method according to one of the preceding claims, wherein for determining the quantity value (34) of the card-illegal driving trajectories (20) and / or the quantity value (35) of the card-conforming driving trajectories (23) for each driving trajectory (20, 23) a time stamp, the their date of detection is determined, and only Fahrtrajektorien (20, 23) are taken into account from a predetermined time window. Method according to one of the preceding claims, wherein when forming the respective quantity value (34, 35) the respective age of the respective driving trajectory (20, 23) is taken into account by a contribution of the driving trajectory (20, 23) to the quantity value (34, 35) smaller, the older the driving trajectory (20, 23) is. Method according to one of the preceding claims, wherein in the event that the possible faultiness of the road map (13) is signaled, then a) on the basis of the trajectory data (24) at least one card driving route (20) map data of the road map (13) are supplemented by a description of the path piece (27) and the supplemented map data are sent to at least one receiver vehicle and / or b) a respective indication signal for indicating the in the area (32) faulty road map (13) and / or preconditioning at least one of the road map (13) and / or a copy of the road map (13) dependent vehicle function to at least one receiver vehicle becomes. Method according to one of the preceding claims, wherein the card-adverse driving trajectory (20) of the respective motor vehicle (14) is detected by an assignment of a position sequence (18), the successive positions (19) of the motor vehicle (14) describes in the road map ( 13), cartographic roads are performed by a predetermined map matching method and an association result satisfying a predetermined off-road condition is identified. Method according to Claim 7 wherein the map-matching method is based on a hidden Markov model which models a spread of the vehicle sensor system by which the successive positions (19) are detected. Method according to Claim 7 or 8th wherein the off-road condition comprises that an assignment probability of the underlying position sequence to the nearest cartographed road is less than a predetermined threshold. Method according to one of Claims 7 to 9 in which each position sequence (18) which satisfies the off-road condition is assigned a relevance value by means of a predetermined evaluation function, and each such position sequence whose relevance value is greater than a threshold value is classified as a card-invalid driving trajectory (20). Method according to Claim 10 in that the evaluation function produces an even smaller relevance value the closer the positional sequence (18) is to a cartographic road and / or the smaller a difference between a shape described by the positional sequence (18) and a predetermined maneuvering pattern. Method according to one of the preceding claims, wherein trajectory data (24) of different card-invalid driving trajectories (20) are received and associated driving trajectories (20), which lead across the same path (27), are detected by means of a predetermined clustering method. A server device (10) for operation on a data network (11), the server device (10) having a computing device adapted to perform a method according to any one of the preceding claims.

Images