WO2016075245A1

WO2016075245A1 - Improvement of a driving horizon system by self-learning the most probable driving path

Info

Publication number: WO2016075245A1
Application number: PCT/EP2015/076451
Authority: WO
Inventors: Karsten Mattmüller; Rolf Haimerl; Manuel Rauch; Raimund VARNHAGEN
Original assignee: Continental Automotive Gmbh
Priority date: 2014-11-14
Filing date: 2015-11-12
Publication date: 2016-05-19
Also published as: DE102014223331B4; DE102014223331A1

Abstract

The invention relates to a method and a device for creating the most probable driving path in a driving horizon system. According to the invention, a driving horizon that is based on the most probable driving path is generated only after creation of the most probable driving path. The frequency of the creation of the most probable driving path and driving horizon is thus reduced by self-learning.

Description

description

Improvement of a driving horizon system by self-learning of the most probable travel path

The invention relates to a method and a device for creating the most probable driving path in a driving horizon system. The most probable path, also called Most Probable Path or MPP, indicates which path a vehicle is likely to drive in the future. The creation of the most probable driving path usually takes place within a driving horizon system, also called horizon provider or eHorizon. So far, the most probable driving path is mainly determined by the road classes ahead of the vehicle. It essentially consists of data stored statically in a map database.

The driving horizon system can detect by positioning or navigation software (GPS, Dead Reckoning, Map Matching or the like) when the most probable travel path is left. In this case, the most probable driving path is invalidated and a new most probable driving path and a driving horizon based thereon are calculated. Therefore, it is in the creation of the most probable

Driving path advantageous if it reflects the future driving behavior as accurately as possible, since a deviation from the highest ^¬ probable driving path additional, re-calculations of both the most likely driving path as well as the driving horizon entails. It occurs by a tarry ^¬ approximately trouble because the driving horizon is available again after its creation.

A generic method for improving the most probable driving path is known from DE 102011078 946 AI. In this case, the most probable travel path and preferably a driving horizon based thereon is first determined by the horizon Provider created and then passed to a driver behavior component, which due to the driver behavior of the past changes to the original most likely

Travel path within the limits of the existing Fahrhorizonts makes. The disadvantage here is that under certain circumstances only very short pieces of the initially created most probable driving path and driving horizon are used, and still a frequent re-creation of the most probable driving path and driving horizon are necessary.

The object of the present invention is therefore to improve the efficiency in the creation of the most probable travel path. This object is achieved with a method according to claim 1 and an apparatus according to claim 13 so that the driving behavior shown in the Ver ^¬ gear integral is taken into account during the creation of the most likely driving path. A driving horizon based on the most probable driving path will only become available after the most probable

Fahrpfades generated. Thus, the frequency of creating the most probable travel path and driving horizon is reduced by self-learning.

In a system, for example a driving horizon system, for creating a most probable travel path, a first database is used in which information about the road network, for example in the form of a map database, is stored. This first database contains as the first data nodes and road segments, the road segments are assigned to different road classes, due to which the creation of the most probable driving path is independent of the driving ^¬ behavior.

In the method according to the invention, at least one further, dynamic second database is used in addition, in which second data about the driving behavior shown in the past are stored. These second data are inventively with Stored with respect to the nodes of the first database, so that at least a subset of the nodes of the map database is also found in the additional, second database. In the second database, the second data can be marked as valid, so that it is known in the preparation of highly true ^¬ translucent current driving path, which second data to which node the to be at least be a second database ^¬ taken into account, and at which node the first data of first map database are to be considered.

During the creation of the most probable travel path, it is then checked for each node, which is relevant for the current calculation of the most probable travel path, whether there are valid second data in the second database with respect to this node. If this is the case, then this node are the valid second data of the second database when creating the most likely travel path be taken into account with respect to ^¬. Are in the second database no valid second data relating to a node exist, only the first data of the first database when creating the most likely driving path into account are ^¬ Untitled for this node.

According to the invention, it is provided that the creation of the

Fahrhorizonts happens only after completion of the creation of the ^¬ highest probable travel path. This also means that a renewed creation of a most probable travel path only takes place when the most probable travel path is left by the vehicle, or when the remaining remaining length of the most probable travel path falls below a specified value. Because the driving horizon is created only after From ^¬ circuit creating the most likely travel path that a computationally intensive ^¬ he imputed driving horizon is rejected at least in part, is avoided. Due to the fact that when falling below a defined residual length is planned again before the end of the most probable driving path, a predictable delay is avoided. Leaving the most probable route is when a vehicle enters a road segment (for example, an intersection) represented by a node of the first database, in another road segment than in that road segment which is most likely

Driving path to the current node would have to follow.

Falling below the residual length can be detected in various ways. For example, this can be set via parameters a number of meters, and the remaining length while driving on the basis of lengths of the individual road segments are determined. Or the remaining length may refer to the number of remaining, still ahead of the vehicle road segments of the most probable driving path. In addition, a relative definition of the restoring to under ^{¬ bordering} residual length is conceivable. Here, the total length, measured for example in meters, or in number of streets ^¬ ßensegmente, the created most likely driving path is taken as a reference and is carried out a rebuild of the most likely driving path when a Festge ^¬ inserted portion of the travel path has been traversed.

When the vehicle leaves the most probable travel path during a journey, second data about the occurred deviation is written to the second database. In addition, at least the validity of the second data stored with respect to the node where the deviation occurred is stored in the second database. According to the invention, the newly added second data or already existing second data can be marked as valid. It is also possible that existing valid second data will be invalid, and that second data will be invalid for future scheduling. According to the invention, the latter happens in particular when the vehicle deviates from the most probable travel path in a manner which corresponds to the most probable travel path which, at the same initial situation (same position and direction of travel) at this node. only based on the first data of the first database.

In addition, according to the invention, no route planning system has to be used in the creation of the most probable travel path, the creation thus takes place without a destination being known.

However, as an extension of the method and apparatus according to the invention, it is possible to combine the claimed method with an external route planning system, but this is not necessary for the functionality of the claimed method. Advantageous developments of the method according to the invention are described in the dependent claims.

Thus, it is advantageous if the node-related second data of the second database at least contain information about the road segment from which the vehicle came, the road segment into which the vehicle has entered, and a time stamp. Here, each record can be stored individually, or the records can be grouped by road segments.

In addition, these second data can be stored in the form of node-related lists, which has the advantage that these lists can be efficiently sorted and searched, and at least a subset of the second data changed at different times remain individually represented by the time stamp. It is also possible to remove the unused second data after a specified time in order to keep the size of the lists small. A further variant of the method according to the invention provides that the second data of the second database are stored in the form of node-related turn-off probabilities. the. One advantage of this variant is that access to the second data can take place in a constant time.

The disadvantage here is that the access to the individual, changed at different times second data is lost.

A further development of the method according to the invention provides that the validity of the second data of the second database is primarily based on when and how often deviations were detected at this node. Here, the detected differences relate to the highest true ^¬ translucent current driving path that would be created with the same initial situation, that is, at least at the same position and driving direction of the vehicle, relying solely on the first data from the first database. According to the invention, it is possible that a one-time deviation from the most probable travel path does not always have to lead to a change in the future calculation of the most probable travel path. Rather, it is possible to control the influence of the creation of the most probable driving path by the second data of the second database by parameters. By a Be ^¬ consideration of the time stamp, it is also made possible that the creation of the most likely driving path occurs differently to different days of the week, or different times. This recurring situations ^¬ Fluk can be considered in traffic.

According to the invention it could also be of advantage if the second data with respect to a node of the second data base are marked as invalid once the vehicle at that node once the most likely driving path not taking into ^¬ account the second database (with the same output ^¬ situation same Position and direction) would be created. Here, too, further parameterizations are possible in order to take into account, for example, the time stamp, or to mark the second data with respect to a node of the second database as invalid only if the vehicle has repeatedly passed the time stamp is followed most likely driving path, which would be based solely on the first data of the first database with the same output ^¬ situation. Also, according to the invention, further data, such as the road segment from which the vehicle came, could also be taken into account.

Furthermore, it may be advantageous according to the invention, if at least one of the databases used ^¬ is stored locally in the vehicle. It may also be advantageous if at least one of the databases used is stored on a server. It can also be advantageous if at least one of the databases used is a SQLite database.

Depending on the field of application, it may be advantageous for the second data of the second database to be personalized or stored in person. This can be the

Personalization of the second data to the vehicle, the driver or both.

The calculation of the most probable travel path is carried out according to the invention by an arrangement which comprises at least one processor. The most probable driving path can be created locally on at least one processor located in the vehicle. This has the advantage that the creation of the most probable driving path can take place even if the connection to a central server is interrupted.

Alternatively, the creation of the most probable travel path and / or the driving horizon can take place on at least one centrally located server. In this case, the need existing system only communicate locally in the vehicle, the relevant for the creation of the most likely driving path the second data to the server and receive the highest ^¬ likely driving path and associated travel horizon from the server after preparation and the vehicle and / or driver in a suitable form put. According to the invention, the use of a route planning system is not necessary. By using a route planning system ^¬ creating the most likely driving ^¬ path is greatly simplified because for a known target deviations from the calculated route can be expected only in exceptional cases.

According to the invention, a parameterization of the method which is as free as possible is preferred. Parameters can resolve possible conflicts between personal and impersonal second data, as well as conflicts between different valid second data. Conflicts between first and second data are solved according to the invention on the validity of the second data.

Possible parameters are for example the weighting of into account of personal and impersonal second data of the second database when creating a highly true ^¬ translucent current driving path. This helps to dissolve the con ^¬ conflicts when another highly likely driving path would be created, for example, based on the provided vehicle ^¬-specific information, such as the second based on the driver ^¬ specific second data.

Furthermore, parameters can control the number of deviations necessary to change the validity. Also, the removal of old or unused second data may be controlled by parameters indicating when second data will be removed or invalidated. For this purpose it is necessary that at least a subset of the second data be provided with time stamps regarding creation and / or access. The parameters can either be set externally, or they can be automatically adjusted by the driving behavior of the past. Various learning algorithms can be used here. It can also be specified by parameters in how far the timestamps stored for the second data are taken into account when creating a most probable travel path. By the parameters temporal fluctuations in Verkehrsauf ^¬ come in the creation of a highly probable driving ^¬ path taken into account, without ignoring the other second data.

It should be noted that both the first data of the first database and the second data of the second database can be distributed, so that, for example, there are different second databases for each vehicle, driver and / or region. In this case, the program for creating the most probable travel path searches for the respectively relevant data from the existing data records of first and second data.

Further features of the method according to the invention and of the device according to the invention are described by the following figures and embodiments.

Fig. 1 assembly and flow chart of the invention

Method,

2 embodiment of the second database,

Fig. 3 change of the most probable travel path based on the driving behavior.

1 shows the schematic structure of the method for improving a driving horizon system by self-learning of the most probable driving path. The first database 101 stores the map data used by the method according to the invention. The map data in this case consist at least of nodes and road segments to which different road ^classes can be ordered to ^¬ . The second database 102 includes the driving behavior data stored due to the recent Fahrver ^¬ holding. In one embodiment, these second data are stored in the second vehicle-specific database so that Be ^¬ special features of the vehicle such as size, weight or possible acceleration can be taken into account. In another embodiment, the second data of the second database is stored impersonally in order to better respond to general impairments, such as accidents and construction sites. In a further exemplary embodiment, the second data of the second database are stored in a driver-specific manner, so that it is easier to respond to personal preferences in the route selection. So it is possible that one driver avoids a certain road segment, while another driver does not.

A mixture of the above-mentioned embodiments is also possible, being able to be set by parameters in the form and the extent to which vehicle-specific, impersonal and driver-specific data included in the second future He ^¬ position of the most likely driving path.

In Fig. 1, both first data of the first database and second data of the second database are used to create the most probable travel path 103. If there are valid second data in the second database, these second data 109, if relevant, for the Creation of the most probable driving path used. If there is no valid second data in the second database with respect to a node relevant to the creation of the most probable travel path, or if the present valid second data is irrelevant to the creation of the most probable travel path, then the first data relating to this node becomes 110 of the first database used. After creating the most probable travel path 103, the created most probable travel path is used to create the driving horizon 104. Here is also possible that the creation of highly ^¬ probable travel path from one module and the subsequent creation of the driving horizon is carried out by another module, or that the creation of the most likely driving path and driving the horizon are executed sequentially by a module.

The Fahr ^¬ horizon based on the most probable route is then used in the vehicle 105 while driving, among other things, based on the driving horizon of the

Shifting behavior of an automatic transmission or speed can be controlled when using Cruise Control. This increases the safety for the driver as well as for other road users. In addition, energy savings are possible. Now, when the vehicle leaves the calculated most likely driving path 106 to recalculate the most probable ^¬ travel path is initiated. In addition, second data about the occurred deviation is written to the second database 102 where it is maintained for future builds of the most probable travel path.

Also when the vehicle is most likely the end of the travel path is approaching 107, so there will be another He ^¬ position of the most probable path and the driving based on driving horizon, but there is no change to the second data.

The route planning system 111 is in this embodiment in no connection to the creation of the most probable driving path.

It should also be emphasized that the first data of the first database remain unchanged due to the driving behavior. However, it is In principle, it is also possible for the first data of the first database to change due to external events, such as a software update. 2, an embodiment of the second database 102 is shown. In this case, the second data in the form of node-related lists 201, 202 and 203 are stored in the second database, wherein at least a subset of the nodes represented in the second database is represented by lists such that each node of the subset is represented by a respective list. These lists contain at least from which road segment the vehicle came ("from"), into which road segment the vehicle has entered ("to"), as well as a time stamp and the validity of the respective data record ("0/1").

In another, not shown, embodiment, the validity could also be stored node-related, so that each node is only valid or invalid, and

Mixed forms, as shown in FIG. 2 possible, do not occur.

In Fig. 3 a - Fig. 3e, the effect of the method according OF INVENTION ^¬ dung is exemplified. The vehicle first comes in Fig. 3a from the left. The most probable travel path calculated first based on the first map data of the first database predicts that the vehicle will go straight on remaining on the main road. The actual travel path is represented by the dashed line in FIGS. 3a-3e. In this example, however, there is a problem 305 with the road which is not represented in the first database and which makes it impossible for the driver to follow the created most probable travel path. This problem could be road damage, low-hanging branches or the like, leading to a de facto blocking of the road for this vehicle. The vehicle turns left based on the lock 305 at the node 301. Thus, the most likely driving path is invalidated, see Fig. 1, 106 and a new highly true ^¬ scheinlicher driving path, and based thereon driving horizon is created. In addition, second data about the made deviation 108 is written into the second database and these second data are validly marked.

When the vehicle now turns right at node 302 in FIG. 3 b and thus leaves the most probable travel path again, a new most probable travel path, as shown in FIG. 3 c, is created and second data about the deviation carried out are written to the second database and these second data are marked valid. The vehicle now follows the most probable travel path to nodes 303, 304.

If the same route is traveled through again, as shown in FIG. 3d, the most probable travel path is likewise created on the basis of the second data of the second database in such a way that the blockage 305 is bypassed.

If the blocking 305 now disappears in FIG. 3e, then the most probable travel path created is still the same as when the barrier was present. Now, if the vehicle is once again running straight at the node 301, which corresponds to You Create solely on the basis of the first data of the first database highly true ^¬ scheinlichem travel path at that node, the information available at the node 301 in the second database second data is marked as invalid, so that in the future the most probable travel path again, as in Fig. 3a, goes straight ahead.

In another embodiment, the second data of the second database is checked for consistency at intervals. In this case, for example, second data of adjacent nodes can be matched and the validity changed. This can be advantageous since detours usually result in changes to several nodes. If the reason for a bypass, however is omitted, it is advantageous to update the second data of all affected ^¬ nodes of the second database.

Claims

Method for determining a most probable travel path (MPP) where a. in addition to a first database 101, which uses a road ^¬ network on the basis of first data, which at least Kno ^¬ th, road segments and road classes, at least one second database 102 is used in at least a subset of nodes of the first database 101 second data about that at the respective nodes 201, 202, 203 ge ^¬ showed driving behavior are kept;

b. at any time at least a portion of the second data in the second database 102 are marked as valid ^¬ ge;

c. the valid second data from the second database 102 for He ^¬ position of the most likely driving path are attracted forth ^¬ with respect to a relevant for the creation of the most likely driving path node in the second data ^¬ bank 102 with respect to this node in the presence of a valid second data 109;

d. the first data 110 is in the presence of no valid second data bezüg ^¬ Lich one for the preparation of the relevant highly probable ^¬ current driving path node in the second database 102 with respect to this node from the first database 101 for Erstel ^¬ development of the most likely driving path Hérange ^¬ subjected;

e. a driving horizon creates 104 only after the presentation of the ^¬ He most likely travel path 103;

f. if a deviation from the highest ^¬ probable driving path takes place during the journey, with respect to the node on which the deviation has taken place, data on the deviation occurred 108 in the second database 102 is written and the validity of the second data related to this node is adjusted;

G. a new highly likely driving path and based thereon driving horizon will only be created when the most likely driving path will leave 106 or the remaining length of the most probable path running below a festgeleg ^¬ th value 107;

H. when creating a route planning system 111 is used.

The method of claim 1, wherein the data of the second database 102 for at least a subset of the nodes 201, 202, 203 at least information on road segments from which a vehicle came to the road segment into which the vehicle is retracted, and a timestamp bein ^¬ haltet.

The method of claim 2, wherein the data is stored in the form of lists.

The method of claim 1, wherein the data of the second database 102 are stored in particular in the form of node-related Abbiegewichlichkeiten.

The method of claim 1, wherein the data of the second database 102 are stored in particular in the form of node and time-related Abbiegewichlichkeiten.

The method of claim 1, wherein the validity of data of the second database 102 is based on when and how often occurred at this node deviations from the alone by the first data of the first database 101 at the same initial situation at this node defined most ^¬ probable travel path ,

7. The method of claim 1, wherein the second data of the second database 102 with respect to a node are invalid if the last time at this node shown driving behavior to the sole by the first data of the first

Database 101 at the same initial situation at this node defined most probable travel path ent ^¬ spoke. 8. The method of claim 1, dadurchgekenn ^¬ records that at least one of the databases 101, 102 is stored locally in the vehicle 105.

9. The method of claim 1, wherein: at least one of the databases 101,

102 is stored remotely on a server.

10. The method of claim 1, wherein at least one of

Databases 101, 102 is a SQLite database.

11. The method of claim 1, dadurchgekenn ^¬ records that at least a subset of the second data of the second database 102 are stored and used personalized.

12. The method of claim 1, dadurchgekenn ^¬ records that at least a subset of the second data of the second database 102 are stored and used impersonally.

13. An apparatus for carrying out the method according to claim 1, wherein a calculation is carried out for generating the travel path 103, at least one processor is used.

14. The device according to claim 13, dadurchge ^¬ indicates that both a local processor in the vehicle as well as a remote processor on a server.