DE102013217061B4 - Accurate positioning of a vehicle - Google Patents

Abstract

Method for estimating the position of a vehicle in a digital map, wherein the vehicle further comprises means for determining the absolute geographical position of the vehicle, in particular a satellite-based positioning system, wherein the map defines lanes, wherein the method comprises:Providing a group of candidate positions, wherein each candidate position is assigned a rating;wherein the rating assigned to each candidate position further takes into account a respective absolute distance which essentially corresponds to the distance between the respective candidate position and the determined absolute geographical position of the vehicle, in particular in a coordinate system;For each candidate position: determining a cluster center candidate position, wherein the cluster center candidate position is one of the candidate positions, wherein the lane distances between candidate positions are taken into account during the determination;Where a lane distance is the distance from one candidate position to another candidate position along the lane or, in particular at junctions, along the lanes;Determining a rating for each of the determined cluster center candidate positions, wherein the rating for the respective cluster center candidate position is based on the ratings of the candidate positions for which the respective cluster center candidate position was determined;selecting a cluster center candidate position as an estimate for the position of the vehicle taking into account the ratings of the determined cluster center candidate positions.

Description

The invention relates to a method for estimating the position of a vehicle in a digital map as well as a computer program and a vehicle for the same purpose.

Nowadays, navigation systems for vehicles, such as cars, trucks, motorcycles or bicycles, are widespread. These have a digital map and means of determining position, in particular satellite navigation receivers such as GPS, GLONASS and/or Galileo receivers. In navigation systems that are factory-installed in vehicles, steering angle sensors and speed sensors on the wheels are also often used as means of determining position. In a digital map, roads are typically divided into individual sections and in more detailed digital maps the number of lanes for each direction of travel is specified for at least some sections of the road. A road has at least one carriageway, where a carriageway is a continuous area that can be driven by a vehicle. At least one new carriageway begins at a junction. Digital maps are often structured according to the GDF (geography data file) standard. During operation, the navigation systems determine the position of the vehicle in the digital map. To do this, the navigation system uses the means for determining the position to first record the geographical position (in geo-coordinates) of the vehicle in order to then find this again in the digital map. This process is also called map matching. When determining the position of the vehicle in the map, the position of the vehicle is advantageously determined on a road and, if necessary, also on a lane on the digital map. The determination of the geographical position is often subject to inaccuracies relevant to positioning on a map due to measurement errors of the means for determining the position; this is particularly the case with the satellite navigation receivers currently used in widely used commercial navigation systems. Furthermore, the position information of the streets in the digital map is also subject to inaccuracies. For these reasons, it often happens that a geographical position determined for the vehicle does not coincide with the position on the map that should actually be determined for the vehicle. This is particularly true if positioning is to be carried out precisely down to the lane, which is sometimes also referred to as lane-accurate positioning. Lane-accurate positioning is particularly advantageous at intersections or junctions, as it allows the driver of a vehicle to be given particularly precise driving instructions and enables advanced driver assistance functions.

It is known from the literature that so-called particle filters are used for map matching. This is described, for example, in: A. Selloum, D. Betaille, E. Le Carpentier and F. Peyret, “Lane level positioning using particle filtering”, Proc. 12th IEEE International Conference on intelligent Transportation Systems 200, pages 1 to 6 However, this work uses a digital map that is accurate to within decimetres and therefore does not correspond to the digital maps in widely used commercial navigation systems.

Furthermore, the so-called median shift for clustering is known from the field of image processing, with the help of which modes can be recognized in a point cloud. This is described in: L. Shapira, S. Avidan and A. Shamir, “Mode-detection via median-shift”, in Proc. IEEE International Conference on Computer Vision, 2009, pages 1909 to 1916 However, this work does not mention or present any possibilities for applying the median shift presented there for vehicle positioning. At the same time, the median shift presented there is not adapted to the properties and problems of vehicle positioning.

The document DE 10 2010 042 314 A1 discloses a method for navigation with a navigation system with which a current location can be determined based on satellites and found on a digital map. It is provided that an identifiable location is determined on the digital map and furthermore a picture of the identifiable location is taken and compared with the identifiable location on the digital map, wherein a spatial difference between the identifiable location determined on the map and the identifiable location of the picture is determined and used for location correction.

The document EP 1 537 541 B1 discloses a method for determining whether at least one vehicle is driving on at least one toll road section by means of a position determination system which is designed to detect the current position of the at least one vehicle, wherein positions of the at least one vehicle are compared with the position of at least one reference point which is characteristic of an entrance to a toll road section.

The object underlying the invention is to improve the positioning of vehicles in digital maps and to make the best possible selection from several possible vehicle positions in the digital map.

The object is achieved by a method, a computer program and a vehicle according to the independent claims. Advantageous further developments are defined in the dependent claims.

In one aspect, a method for estimating the position of a vehicle in a digital map, the map defining lanes, the method comprising: providing a group of candidate positions, each candidate position being assigned a rating; the candidate positions being located in particular on the lanes of the digital map; for each candidate position: determining a cluster center candidate position, the cluster center candidate position being one of the candidate positions, the lane distances between candidate positions being taken into account during the determination; a lane distance being the distance from one candidate position to another candidate position along the lane or, in particular at junctions, along the lanes; determining a rating for each of the determined cluster center candidate positions, the rating for the respective cluster center candidate position being determined based on the ratings of the candidate positions for which the respective cluster center candidate position was determined; selecting a cluster center candidate position as an estimate for the position of the vehicle, taking into account the ratings of the determined cluster center candidate positions. The set of candidate positions is typically obtained by applying a particle filter, where the candidate positions represent particles.

The vehicle further comprises means for determining the absolute geographical position of the vehicle, in particular a satellite-based positioning system (for example based on GPS and/or Galileo and/or GLONASS), wherein the evaluation assigned to each candidate position also takes into account a respective absolute distance, which essentially corresponds to the distance between the respective candidate position and the determined absolute geographical position of the vehicle, in particular in a coordinate system. Here, the distance that would result between two coordinates in a geographical coordinate system is taken into account for the evaluation of the candidate positions, and therefore not the roadway distance. Physical measured values are taken into account when evaluating the candidate positions, which also affect the evaluations of the cluster center candidate positions and thus the estimated vehicle position.

In this way, the advantages of the median shift known from other areas of technology are utilized for vehicle positioning. The method proposes the application of the median shift for the group of candidate positions (i.e. possible vehicle positions). The median shift known from the literature is adapted for vehicle positioning specifications. Using a median shift, clusters are first formed (each of which is represented by the specific cluster center candidate position and all candidate positions for which the respective cluster center candidate position was determined). One of the cluster center candidate positions is then selected as an estimate of the position. As a key difference to the median shift presented in the literature, the lane distances are taken into account during clustering (comprising the determination of the cluster center candidate positions). Furthermore, the method presented here also differs from the method presented in the literature in the selection of the cluster center candidate position for estimating the position of the vehicle. According to the method disclosed here, evaluations of the candidate positions themselves are used for the selection of the estimate. These evaluations are determined in particular based on position measurements using satellite navigation receivers or the measurement of the orientation of the vehicle and thus have a reference to the physical measurement results.

The road distance can easily be determined by specifying the distance of the vehicle's position from the start of the section when roads are divided into sections, as is common in digital maps. However, in order to correctly determine the road distance at junctions, this often requires that junctions always start new sections, which is the case with typical digital map formats (e.g. the GDF format).

In a further development, determining the cluster center candidate position for the respective candidate position comprises an iterative procedure in which candidate positions are selected one after the other that satisfy an optimization criterion, in particular a minimization criterion, based on the lane distance to other candidate positions, whereby the iterative procedure terminates and the cluster center candidate position is found when the same candidate position is chosen directly one after the other, whereby this candidate position is determined as the cluster center candidate position.

The candidate position is therefore determined based on the lane distances between a candidate position and other candidate positions and an optimization criterion based on this. The optimization criterion typically consists of minimizing a measure, whereby the measure is determined based on the lane distances. In this way, the median shift known from the literature is further adapted to the specifics of the vehicle positioning and individual, suitable criteria are applied for the selection of the medians and clustering.

In a first alternative, the method is designed such that determining the cluster center candidate position for the respective candidate position comprises an iterative method in which the respective candidate position is the first starting candidate position, the iterative method comprising: determining the surrounding candidate positions, namely those candidate positions that are located within a predetermined roadway distance from the starting candidate position, the starting candidate position also being an surrounding candidate position; determining the minimum candidate position, namely that surrounding candidate position that satisfies an optimization criterion, in particular minimizing the sum of the roadway distances to the other surrounding candidate positions or the sum of measures based on the roadway distances, in particular the roadway distances weighted with the respective evaluations of the other surrounding candidate positions; repeating the two preceding steps, the minimum candidate position being the starting candidate position; Abort the repetition if the starting candidate position and the designated minimum candidate position are identical, where the starting candidate position is then the cluster center candidate position.

In a second alternative, which is more economical in terms of computational effort, the minimum candidate position is first determined and saved for all candidate positions. The resulting list is then used to find the cluster center candidate position: The method further comprises: For each candidate position: Determining the surrounding candidate positions, namely those candidate positions that are within a predetermined roadway distance from the respective candidate position, wherein the respective candidate position is also an surrounding candidate position; Determining the minimum candidate position, namely that surrounding candidate position that satisfies an optimization criterion, in particular the sum of the roadway distances to the other surrounding candidate positions or the sum of measures based on the roadway distances, in particular the roadway distances weighted with the respective evaluations of the other surrounding candidate positions, is minimized; Assigning the minimum candidate position to the respective candidate position; Determining the cluster center candidate position for each candidate position based on the assignment, wherein this step in particular comprises an iterative process in which the respectively assigned determined minimum candidate position is the first starting candidate position, and the iterative process comprises: a) determining the minimum candidate position assigned to the starting candidate position; b) repeating the previous step, wherein the determined minimum candidate position becomes the starting candidate position, until the starting candidate position corresponds to the assigned minimum candidate position, wherein this starting candidate position is then the cluster center candidate position for the respective candidate position.

In another development, the vehicle further comprises: means for determining the orientation of the vehicle, in particular in relation to cardinal directions; the rating assigned to each candidate position is based on the difference between the candidate orientation and the determined orientation of the vehicle, the candidate orientation being the orientation of the roadway or a section of the roadway in the map on which the respective candidate position is located. The orientation of the vehicle is in particular the direction of travel, which is sometimes also called heading. A further criterion is thus introduced into the evaluation of the candidate positions, namely the orientation of the vehicle and the candidate position. The orientation of the candidate position is determined using the digital map. Here, too, physical measurements are taken into account when evaluating the candidate positions, which also affect the evaluations of the cluster center candidate positions and thus the estimated vehicle position.

Those candidate positions for which the same cluster center candidate position has been determined can be referred to as a cluster or cluster group.

In a further development, lanes are also assigned to the candidate positions; the lane distance is determined in particular independently of the lanes; the lanes are divided into sections; the method further comprising: for each lane of the section on which the selected cluster center candidate position is located; summing or forming an average of the evaluations of the candidate positions that are located on the section in the respective lane; determining the lane for the estimation of the vehicle position based on the sum or the average for the respective lane, in particular the highest sum or the highest average. By selecting the cluster center candidate position, the lateral position of the estimate is determined, i.e. the position in the longitudinal direction or the longitudinal position in relation to the direction of travel. This can be done in particular by specifying a distance from a reference point, for example the start of a lane section or a road section. The lane is then determined based on the distribution of the other candidate positions across the lanes within the section on which the selected cluster center candidate position is located.

Another aspect relates to a computer program which causes a computer to carry out one of the above methods when the computer program is executed. The computer can be a microcontroller, a general-purpose computer or can comprise dedicated circuits, provided that these can be set up by programming. The computer program can be comprised of a computer program product.

In a further aspect, a vehicle comprises: electronic computing means configured to carry out one of the above methods, wherein the vehicle comprises the means defined in the respective method. The electronic computing means may be a computer.

1. 1. Zusatzverfahren zum Schätzen der Position eines Fahrzeugs in einer digitalen Landkarte, wobei die Landkarte Fahrbahnen und Fahrspuren definiert, wobei die Position die Entfernung des Fahrzeugs von einem Referenzpunkt entlang der Fahrbahn und die Fahrspur angibt, wobei das Fahrzeug umfasst:
   • Mittel zur Schätzung der Fahrtdistanz, nämlich der Distanz entlang der Fahrbahn, die das Fahrzeug von einem ersten zu einem zweiten Zeitpunkt zurückgelegt hat;
   • eine Kamera, die auf einen Teil der das Fahrzeug umgebenden Fahrbahn, insbesondere die dem Fahrzeug vorausliegende Fahrbahn, gerichtet ist; und
   • Rechenmittel zur Bildverarbeitung, wobei die Rechenmittel dazu eingerichtet sind, in den Aufnahmen der Kamera eine Fahrspurmarkierung zu erkennen;
   • Wobei eine Fahrspurmarkierung eine Markierung auf der Fahrbahn umfasst, die insbesondere die Grenze einer Fahrspur anzeigt;
   wobei eine Kandidatenposition die Entfernung des Fahrzeugs von einem jeweiligen (nicht notwendigerweise verschiedenen) Referenzpunkt entlang der Fahrbahn und die Fahrspur angibt;

The method can also be carried out in combination with the following additional methods, additionally carried out by the computer program below or additionally implemented in the vehicle below as described therein, wherein the first group of candidate positions and their evaluations according to the additional methods are the group of candidate positions and their evaluations for which the method presented so far is carried out. Additional features in the additional methods supplement the features of the methods presented above.

1. 1. An additional method for estimating the position of a vehicle in a digital map, the map defining roadways and lanes, the position indicating the distance of the vehicle from a reference point along the roadway and the lane, the vehicle comprising:
   • Means for estimating the travel distance, namely the distance along the roadway travelled by the vehicle from a first to a second point in time;
   • a camera directed at a part of the road surrounding the vehicle, in particular the road ahead of the vehicle; and
   • Computing means for image processing, wherein the computing means are configured to recognize a lane marking in the camera images;
   • Wherein a lane marking comprises a marking on the roadway which in particular indicates the boundary of a lane;
   where a candidate position indicates the distance of the vehicle from a respective (not necessarily different) reference point along the roadway and the lane;

• Bereitstellen einer ersten Gruppe von Kandidatenpositionen für den ersten Zeitpunkt;
• Bestimmen einer Bewertung für jede Kandidatenposition der ersten Gruppe von Kandidatenpositionen unter Berücksichtigung der erkannten Fahrspurmarkierung;
• Bestimmen einer zweiten Gruppe von Kandidatenpositionen für den zweiten Zeitpunkt basierend auf Kandidatenpositionen der ersten Gruppe; wobei die von einer Kandidatenposition der zweiten Gruppe angegebene Entfernung basierend auf der von einer Kandidatenposition der ersten Gruppe angegebenen Entfernung und der Fahrtdistanz bestimmt wird;
• Bestimmen einer Kandidatenposition aus der ersten Gruppe von Kandidatenpositionen als Schätzung für die Position des Fahrzeugs unter Berücksichtigung der Bewertungen der jeweiligen Kandidatenpositionen der ersten Gruppe.

The method comprises repeatedly performing the following steps, using the second group of candidate positions as the first group of candidate positions when repeating the steps:

• Providing an initial group of candidate positions for the first time point;
• Determining a score for each candidate position of the first group of candidate positions taking into account the detected lane marking;
• Determining a second group of candidate positions for the second time based on candidate positions of the first group; wherein the distance indicated from a candidate position of the second group is determined based on the distance indicated from a candidate position of the first group and the travel distance;
• Determining a candidate position from the first group of candidate positions as an estimate for the position of the vehicle, taking into account the evaluations of the respective candidate positions of the first group.

In the typical application when implementing a particle filter, detected lane markings are taken into account when evaluating the particles, i.e. the candidate positions. In this way, the evaluation criteria usually used for particle filters are expanded to include evaluations of the lane markings. Due to the additional source of information from the lane markings, more robust, lane-accurate positioning is possible.

The means for estimating the travel distance can be included in a navigation system, or only comprise a satellite navigation receiver and corresponding signal processing, or can be implemented via motion and/or wheel sensors of the vehicle. In addition, a (loose) coupling of the various sensor systems, for example in a Kalman filter, is conceivable.

1. 2. Zusatzverfahren nach Zusatzverfahren 1, wobei die Rechenmittel ferner dazu eingerichtet sind, zusätzlich den Typ der erkannten Fahrspurmarkierung zu erkennen, insbesondere, ob es sich um eine durchgezogene Markierung oder gestrichelte Markierung handelt oder in welche Richtung ein erkannter Richtungspfeil zeigt; wobei bei dem Bestimmen der Bewertung für jede Kandidatenposition der ersten Gruppe zusätzlich der Typ der Fahrspurmarkierung berücksichtigt wird.
2. 3. Zusatzverfahren nach Zusatzverfahren 2, wobei die Rechenmittel ferner dazu eingerichtet sind, zu erfassen, ob sich die Fahrspurmarkierung rechts oder links des Fahrzeugs befindet;
   • wobei bei dem Bestimmen einer Bewertung für jede Kandidatenposition berücksichtigt wird, ob sich die erkannte Fahrspurmarkierung links oder rechts des Fahrzeugs befindet.
3. 4. Zusatzverfahren nach einem der Zusatzverfahren 2 oder 3, wobei das Bestimmen der Bewertung für jede Kandidatenposition umfasst:
   • Bestimmen einer Einflussgröße für die Bewertung der jeweiligen Kandidatenposition anhand einer vorgespeicherten Zuordnung, insbesondere einer Tabelle, wobei die Zuordnung einem erkannten Typ der Fahrspurmarkierung und einer von einer Kandidatenposition angegebenen Fahrspur eine Einflussgröße zuordnet.
4. 5. Zusatzverfahren nach einem der vorhergehenden Zusatzverfahren 1 bis 4:
   • wobei die Rechenmittel ferner dazu eingerichtet sind, in den Aufnahmen der Kamera eine weitere Fahrspurmarkierung zu erkennen; und
     • insbesondere dazu eingerichtet sind, den Typ der erkannten Fahrspurmarkierung zu erkennen;
     • und weiter insbesondere dazu eingerichtet sind, zu erkennen, ob sich die weitere Fahrspurmarkierung links oder rechts des Fahrzeugs befindet;

A lane marking can delimit a lane or provide information about traffic management. The delimitation can be such that it can be crossed by the vehicle according to the road traffic regulations (dashed line or marking) or not crossed (solid line or marking). In addition, further classifications or typifications of delimitations are possible. A marking for traffic management can, for example, be a directional arrow that indicates whether the lane is a turning lane. A road can comprise several lanes.

1. 2. Additional method according to additional method 1, wherein the computing means are further configured to additionally recognize the type of lane marking recognized, in particular whether it is a solid marking or a dashed marking or in which direction a recognized directional arrow points; wherein the type of lane marking is additionally taken into account when determining the evaluation for each candidate position of the first group.
2. 3. Additional method according to additional method 2, wherein the computing means are further configured to detect whether the lane marking is located to the right or left of the vehicle;
   • wherein when determining a score for each candidate position, it is taken into account whether the detected lane marking is located to the left or right of the vehicle.
3. 4. Additional procedure according to one of the additional procedures 2 or 3, wherein determining the rating for each candidate position comprises:
   • Determining an influencing factor for the evaluation of the respective candidate position based on a pre-stored assignment, in particular a table, wherein the assignment assigns an influencing factor to a recognized type of lane marking and a lane specified by a candidate position.
4. 5. Additional procedure according to one of the preceding additional procedures 1 to 4:
   • wherein the computing means are further configured to detect a further lane marking in the camera recordings; and
     • in particular, they are designed to recognise the type of lane marking detected;
     • and are further designed in particular to detect whether the further lane marking is to the left or right of the vehicle;

• insbesondere den Typ der erkannten weiteren Fahrspurmarkierung berücksichtigt, und weiter insbesondere für jede erkannte Fahrspurmarkierung und deren Typ berücksichtigt, ob sich die Fahrspurmarkierung links oder rechts des Fahrzeugs befindet.

Wherein determining a score for each candidate position of the first group of candidate positions also takes into account the further lane marking; and

• in particular, the type of additional lane marking detected is taken into account, and further, for each detected lane marking and its type, in particular, whether the lane marking is to the left or right of the vehicle.

By taking into account the other lane markings, their type and in particular whether they are on the left or right of the vehicle, a more precise assignment of the vehicle to a relative lane position is possible. By taking into account a lane marking on another side of the vehicle, it is also possible to detect whether the vehicle is in the lane on the far left or the far right of a multi-lane road. If only one lane is detected, it may only be possible to make a statement about one edge position (i.e. lane on the far left or another lane or lane). lane on the far right or other lane). By determining both sides, an improved evaluation of the candidate positions can be made.

6. Additional method according to additional method 5 with reference to additional method 4, wherein the table assigns an influencing variable to a recognized type of the first and a recognized type of the second lane marking and to a lane indicated by a candidate position, taking into account in particular whether the respective lane marking and its type is recognized to the left or right of the vehicle.

• wobei die Mittel zur Schätzung der Fahrtdistanz umfassen: Mittel zur Bestimmung der absoluten geographischen Position des Fahrzeugs;

7. Additional procedure according to one of the preceding additional procedures 1 to 6,

• wherein the means for estimating the travel distance comprise: means for determining the absolute geographical position of the vehicle;

• Schätzen der absoluten geographischen Position des Fahrzeugs zu dem zweiten Zeitpunkt;
• Für jede Kandidatenposition der ersten Gruppe: Bestimmen der Absolutdistanz, nämlich der Distanz zwischen der jeweiligen Kandidatenposition und der absoluten geographischen Position des Fahrzeugs;
• Wobei das Bestimmen einer Bewertung für jede Kandidatenposition ferner die jeweilige bestimmte Absolutdistanz berücksichtigt.

The procedure includes:

• estimating the absolute geographical position of the vehicle at the second time;
• For each candidate position of the first group: determining the absolute distance, namely the distance between the respective candidate position and the absolute geographical position of the vehicle;
• Whereby determining a rating for each candidate position also takes into account the respective determined absolute distance.

• wobei das Fahrzeug umfasst: Mittel zur Bestimmung der Ausrichtung des Fahrzeugs, wobei diese Mittel zur Bestimmung der Ausrichtung insbesondere von dem Mittel zur Bestimmung der Fahrtdistanz umfasst sind;
• wobei das Verfahren umfasst:
  • Bestimmen der Ausrichtung des Fahrzeugs zum zweiten Zeitpunkt;

8. Additional procedure according to one of the preceding additional procedures 1 to 7,

• the vehicle comprising: means for determining the orientation of the vehicle, wherein said means for determining the orientation are in particular comprised by the means for determining the travel distance;
• the method comprising:
  • Determining the orientation of the vehicle at the second time;

For each candidate position of the first group: determining the candidate orientation, namely the orientation of the roadway or a section of the roadway in the map on which the respective candidate position is located;

Wherein determining a score for each candidate position further takes into account the difference between the orientation of the vehicle and the respective candidate orientation.

• Bestimmen einer Untergruppe der ersten Gruppe von Kandidatenpositionen unter Berücksichtigung der Bewertung der jeweiligen Kandidatenposition, insbesondere unter Beachtung eines Schwellwertes; wobei die Untergruppe eine zweite Anzahl von Kandidatenpositionen umfasst, die kleiner ist als die erste Anzahl;
• Bestimmen einer Zwischengruppe von Kandidatenpositionen, wobei die Zwischengruppe die Untergruppe umfasst; wobei die Zwischengruppe die erste Anzahl an Kandidatenpositionen aufweist;
• Bestimmen der zweiten Gruppe von Kandidatenpositionen basierend auf der Zwischengruppe;
• wobei die von jeder Kandidatenposition der zweiten Gruppe angegebene Entfernung jeweils basierend auf der von einer Kandidatenposition der Zwischengruppe angegebenen Entfernung zuzüglich der Fahrtdistanz bestimmt wird, so dass die von einer Kandidatenposition der zweiten Gruppe angegebene Entfernung basierend auf der von einer Kandidatenposition der ersten Gruppe angegebenen Entfernung und der Fahrtdistanz bestimmt wird.

9. Additional procedure according to one of the preceding additional procedures 1 to 8,
wherein the first group has a first number of candidate positions;
wherein the second group has the first number of candidate positions;
wherein determining the second group comprises:

• Determining a subgroup of the first group of candidate positions taking into account the evaluation of the respective candidate position, in particular taking into account a threshold value; wherein the subgroup comprises a second number of candidate positions which is smaller than the first number;
• Determining an intermediate group of candidate positions, the intermediate group comprising the subgroup; the intermediate group having the first number of candidate positions;
• Determining the second group of candidate positions based on the intermediate group;
• wherein the distance indicated from each candidate position of the second group is determined based on the distance indicated from a candidate position of the intermediate group plus the travel distance, such that the distance indicated from a candidate position of the second group is determined based on the distance indicated from a candidate position of the first group and the travel distance.

This additional procedure enables so-called sampling and resampling. From the candidate positions in the first group, those are selected that appear to be the most promising for a good estimate of the vehicle's position (subgroup). These are then added to the first Number multiplied (intermediate group) and shifted according to the travel distance to reach the second group.

• Vervielfältigen von Kandidatenpositionen der Untergruppe, so dass die Zwischengruppe mit der ersten Anzahl an Kandidatenpositionen entsteht;
• wobei sich die von jeder Kandidatenposition der zweiten Gruppe angegebene Entfernung jeweils basierend auf der von einer Kandidatenposition der Zwischengruppe angegebenen Entfernung berechnet, zuzüglich der Fahrtdistanz und zuzüglich eines für die jeweilige Kandidatenposition der zweiten Gruppe bestimmten Betrages, der zufällig bestimmt wird, insbesondere gemäß einer Normalverteilung oder der Summe von Normalverteilungen.

10. Additional procedure according to additional procedure 9,
wherein determining the intermediate group comprises:

• Duplicating candidate positions of the subgroup so that the intermediate group with the first number of candidate positions is created;
• wherein the distance indicated by each candidate position of the second group is calculated based on the distance indicated by a candidate position of the intermediate group, plus the travel distance and plus an amount determined for the respective candidate position of the second group, which is randomly determined, in particular according to a normal distribution or the sum of normal distributions.

The distances of the candidate positions are thus "noisy". For this purpose, a different noise can be added to each distance indicated by a candidate position in the intermediate group. This prevents the duplicated candidate positions in the intermediate group from all indicating the same position after the shift by the travel distance and also improves the estimate overall.

11. Additional method according to one of the preceding additional methods 1 to 10, wherein the digital map also specifies, for at least some of the sections for each lane, the lane or lanes following the respective lane in the direction of travel;

Wherein determining a candidate position of the second group takes into account which lane or lanes follow the lane indicated by the candidate position of the first group on which the candidate position of the second group is based;
wherein the subsequent lane is taken into account in particular when a candidate position of the second group is located on a different section than the section on which the candidate position of the first group is located, based on which the candidate position of the second group was determined.

When updating the candidate positions (determining the candidate positions of the second group based on candidate positions of the first group), the lane in which the original candidate position is located and the subsequent lanes for the lane of the original candidate position are taken into account. This makes it possible to shift the candidate positions in a meaningful way during the execution of the procedure (or a particle filter if necessary). The shift corresponds to the lane behavior that a vehicle could perform. In this way, candidate positions are generated in a more intelligent and improved way. Consequently, the estimate of the vehicle's position based on the candidate positions is also improved.

12. Additional method according to one of the preceding additional methods 1 to 11, wherein the computing means are further configured to detect the crossing of a lane marking in the camera images and are further configured in particular to detect a lane change of the vehicle,
wherein the determination of the lanes of the candidate positions of the second group takes into account the detected lane change.

Thus, another source of information, namely the camera and a lane change, is used to update the candidate positions according to the observed behavior of the vehicle. This enables the determination of candidate positions whose position information corresponds better to the actual position of the vehicle and thus enables an improved estimation.

13. Additional procedure according to additional procedure 12,
wherein the lanes indicated by the candidate positions of the second group are based on the lanes indicated by the candidate positions
the first group and the detected lane change.

Computer program which, when executing the computer program, causes a computer to carry out a method according to one of the preceding additional methods 1 to 13.

• Mittel zur Schätzung der Fahrtdistanz, nämlich der Distanz die das Fahrzeug von einem ersten zu einem zweiten Zeitpunkt zurückgelegt hat;
• eine Kamera, die auf einen Teil der das Fahrzeug umgebenden Fahrbahn, insbesondere die dem Fahrzeug vorausliegende Fahrbahn, gerichtet ist; und
• Elektronische Rechenmittel zur Bildverarbeitung, wobei die Rechenmittel dazu eingerichtet sind, in den Aufnahmen der Kamera eine Fahrspurmarkierung zu erkennen;
• Elektronische Rechenmittel, die programmtechnisch einrichtbar sind;
• Wobei das Fahrzeug dazu eingerichtet ist, ein Verfahren nach einem der Zusatzverfahren 1 bis 13 auszuführen und die in dem jeweiligen Zusatzverfahren definierten Mittel umfasst.

Vehicle, comprising:

• Means for estimating the travel distance, namely the distance travelled by the vehicle from a first to a second point in time;
• a camera directed at a part of the road surrounding the vehicle, in particular the road ahead of the vehicle; and
• Electronic computing means for image processing, wherein the computing means are configured to recognise a lane marking in the camera images;
• Electronic computing devices that can be set up by programming;
• The vehicle is designed to carry out a method according to one of the additional methods 1 to 13 and comprises the means defined in the respective additional method.

SHORT DESCRIPTION OF THE DRAWINGS

• 1 shows schematically a flow chart of a first, optional part of the method according to an embodiment.
• 2 shows a second group of candidate positions and a section of a roadway according to the first part of the method according to an embodiment.
• 3 shows schematically a flow chart of the second part of the method according to an embodiment.
• 4 shows schematically the clustering of candidate positions according to an embodiment.

DETAILED DESCRIPTION OF THE EMBODIMENTS

1 represents steps of a first part of the method according to an embodiment. These steps are optional in the method according to the invention. In the first optional part of the method according to 1 the roads of the digital map each comprise at least one lane, which is divided into individual sections I, sometimes also called links. Furthermore, the digital map also indicates the maximum number of lanes that can be used in one direction of travel for each section. In addition, the digital map specifies which section follows a given section in the direction of travel and also specifies which lane of a section is connected to which lane of the following section. In digital maps, sections end in particular at junctions of streets or lanes. Therefore, several subsequent sections and subsequent lanes can also be specified for a section or lane. In general, the invention can also be carried out with a digital map that does not provide any specification of the respective subsequent sections and lanes.

The estimated position x or candidate position c of a vehicle in a digital map is specified in the digital map by three pieces of information: the identification I of the section of the road, which is a natural number herein; the identification of the lane, which is also a natural number herein; and the distance s from the beginning of the section (which serves as a reference point) along the roadway, which is a real number greater than 0. Thus, the estimated position x at time t*T (where t is a natural number and T indicates a time period, for example 0.1s; 0.5s; 1s; 5s) can be represented as: x _t = (l _t , k _t , s _t ) and the candidate position c at time t can be represented as: c _t = (l _t , k _t , s _t ).

The method for estimating the position of the vehicle in the digital map is carried out by a vehicle that includes means for estimating the travel distance of the vehicle from a first point in time to a second point in time. These means include a satellite navigation receiver, a speedometer on at least one wheel of the vehicle and a yaw rate determination using the steering angle. The outputs of these sensors are linked together in an extended Kalman filter known from the prior art in order to estimate the travel distance d _t that the vehicle has covered from a first point in time (t-1)*T to a second point in time t*T. The vehicle also includes a camera that is aimed at the road ahead of the vehicle. The camera can take pictures in the visible range of light and use a CMOS sensor. Other examples of usable cameras are known in the prior art. The vehicle also includes electronic computing means for image processing, wherein the computing means are set up to recognize the lane markings present there and their type in the camera images. Methods for this are known in the prior art. The lane markings closest to the vehicle are identified and classified as left or right lane markings (seen from the direction of travel). The vehicle further comprises means for determining the orientation h of the vehicle, which are included in the means for determining the travel distance. The orientation h is determined based on the yaw rate determination and/or an electronic compass, wherein the orientation ht of the vehicle is described with respect to a fixed reference system, for example the cardinal direction, at time t*T.

To initiate the method according to the embodiment with reference to 1 the absolute position of the vehicle is determined using the position determination means, S1. The position x _ini in the digital map that is closest to this position is then determined. This position in the digital map is duplicated so that a first number of candidate positions (also called particles) is reached. The first number can be, for example, 10, 20, 30, 50 or more. These candidate positions are then "noisy", i.e. each given a different deviation from the position x _ini , with the deviations being chosen according to a normal distribution. The deviation is made two-dimensionally so that both the distance s _t and the section l _t as well as the lane k _t of a candidate position c _t can change due to the noise. When offsetting a candidate position transversely to the direction of travel, i.e. to another lane, the typical width of a lane is taken into account, unless specified by the digital map. Furthermore, the information on the connection between sections and the connection of lanes across the sections is taken into account in the noise generation, so that the absolute position of a candidate position is represented by the candidate position in the digital map. The (noisy) candidate positions represent the first group of candidate positions, S2.

For each candidate position c of the first group, a rating w _{c, t} is created for the time t*T, S3. The rating consists of three components: a) a rating component w _{c, t, dist} based on the absolute distance; b) a rating component w _{c, t, heading} based on the orientation of the candidate position; and c) a rating component w _{c, t, lane} based on the detected lane markings. The rating w _{c, t} is then calculated by multiplying the individual components: $${\text{w}}_{\text{c}\text{, t}}={\text{w}}_{\text{c}\text{, t}\text{, dist}}*{\text{w}}_{\text{c}\text{, t}\text{, heading}}*{\text{w}}_{\text{c}\text{, t}\text{, lane}}$$

The evaluation component based on the absolute distance w _{c, t, dist} refers to the distance d _abs between an absolute geographical position of the vehicle determined using the means of determining the position and an absolute position of the vehicle determined for the respective candidate position in the map. An absolute position is therefore inferred from the candidate position in the map. The evaluation component w _{c, t, dist} is calculated, for example, as follows: $${\text{w}}_{\text{c}\text{, t}\text{, dist}}=\text{exp}\left(-0.08{\text{*d}}_{\text{abs}}\right).$$

The evaluation component w _{c, t, heading} based on the orientation of the candidate position refers to the absolute orientation of the vehicle and the absolute orientation of the candidate position in the map. The orientation of the vehicle is determined using the means for determining the orientation. The absolute orientation of the candidate position is determined based on the absolute orientation of the section of the road or carriageway in the digital map. It may be intended to use the orientation of only part of the section if it changes its orientation significantly. It is also possible to determine the orientation based on the lane or road on which the vehicle is located, provided that the orientations of the individual lanes differ from one another and are known (for example as additional information in the digital map). The angular difference d _heading (in radians) between the two orientations is determined. The evaluation component w _{c, t, heading} is determined, for example, as follows: $${\text{w}}_{\text{c}\text{, t}\text{, heading}}={\left(1+\text{exp}\left(10{\text{*d}}_{\text{heading}}-7.5\right)\right)}^{-1}$$

The advantage of this calculation of the evaluation component based on the orientation is that small deviations in the orientation of the section in the map do not have too much of an impact. This is particularly useful at junctions or branches of the roads, which only inaccurately reflect the actual orientations, especially in digital and possibly abstracted maps. On the other hand, the evaluation component changes sufficiently when the angle difference increases.

The evaluation component w _{c, t, lane} refers to the detected lane markings. The vehicle's camera and the electronic computing devices detect the lane markings closest to the left and right of the vehicle and their type. Types are differentiated between "solid" and "dashed" as well as "unknown" if the marking cannot be determined as "solid" or "dashed". In the vehicle, in particular in the electronic computing devices, an assignment in the form of a table is also stored, with the help of which the evaluation component w _{c, t, lane} is determined. A value for the evaluation component w _{c, t, lane} is assigned to output variables via the table. The output variables are the type of the detected left lane marking, the type of the detected right lane marking, the number of lanes in one direction of travel on the section in which the respective candidate position is located and the lane that indicates the candidate position. It is also possible to determine the evaluation component w _{c, t, lane} using only one detected lane and its type. In this case, however, the evaluation component may less frequently result in a change in the evaluation of the candidate position.

The evaluation components w _{c, t, lane} given in the table show the extent to which the lane markings that would be expected for a candidate position match the lane markings detected in the camera. The expected lane marking can be inferred from the number of lane markings in a direction of travel and the lane of the candidate position by determining the relative position of the lane in relation to all lanes. For example, it can be concluded that the candidate position is in the leftmost lane if the candidate position indicates that it is in the third lane from the right (for example kt=3) and the section has a total of three lanes in a direction of travel. In this case, the left lane marking would be expected to be a solid line and the right lane marking a dashed line. If the expected markings match the detected markings, the evaluation of the candidate position is improved. This is done by choosing the value for w _{c, t, lane} in the table or assignment accordingly. For example, in this case, w _{c, t, lane} = 1. If, on the other hand, the expected markings do not match the detected markings, the rating of the candidate position can be reduced. This is done by selecting the appropriate value for w _{c, t, lane} in the table or assignment. For example, in this case, w _{c, t, lane} = 0.5. It should be noted that the detection of lanes and lane types using camera recordings is subject to errors. For this reason, candidate positions whose expected lane markings do not match the detected lane markings should not be completely downgraded due to this difference. In general, complete downgrading is of course still possible, especially if the lane markings are detected particularly reliably.

In another example, it can be concluded that the candidate position is in a middle lane on the left if the candidate position indicates that it is in the second lane from the right (for example kt=2) and the section has three lanes in total in one direction of travel. In this case, the left lane marking would be expected to be a dashed line and the right lane marking would be expected to be a dashed line. If the expected markings match the detected markings, the evaluation of the candidate position is improved. This is done by choosing the value for w _{c, t, lane} in the table or mapping accordingly. For example, in the present case, w _{c, t, lane} = 1. If, on the other hand, the expected markings do not match the detected markings, the evaluation of the candidate position can be reduced. This is done by choosing the value for w _{c, t, lane} in the table or mapping accordingly. For example, in the present case, w _{c, t, lane} = 0.5.

If only the right lane marking is detected, it is only possible to differentiate, for example, whether the vehicle is in the rightmost lane or in another lane. In the case of a three-lane road, no separate statement can be made for the leftmost lane indicated by a candidate position. The influence on the evaluation of the candidate positions is therefore limited compared to the detection of lane markings on two sides of the vehicle.

Intermediate values for w _{c, t, lane} are possible, for example w _{c, t, lane} = 0.8 if the type of lane markings is unknown. In general, values of 0.8 and 1 for w _{c, t, lane} can result in an improved rating, while 0.5, for example, results in a reduced rating.

Table 1 gives an example of possible values for w _{c, t, lane} in case the road section has two or more lanes. The headings of the rows and columns indicate the recognized types of lane markings for left and right, respectively. The values for w _{c, t, lane} are given in the result cells. ben. The value to be used is determined by the relative position of the lane, which is determined based on the lane of the candidate position and the maximum number of lanes. If the lane is on the far left, the first value for w _{c, t, lane} should be used. If the lane is on the far right, the last value should be used. If there are more than three lanes on the section and the lane is not an edge lane, the middle value should be used. Table 1 Type Right solid dashed line Unknown Type Links line Solid line 0.5 | 0.5 | 0.5 1 | 0.5 | 0.5 0.8 | 0.5 | 0.5 dashed line 0.5 | 0.5 | 1 0.5 | 1 | 0.5 0.5 | 0.5 | 0.8 Unknown 0.5 | 0.5 | 0.8 0.8 | 0.8 | 0.5 0.5 | 0.5 | 0.5

In the case that a section has only two lanes, for the subcase that dashed lines are detected on the right and left, instead of the division shown in Table 1, the value 1 is to be used for both the right and left lane indication of the candidate position.

The determination of the value for the evaluation component w _{c, t, lane} can also include other criteria. For example, the evaluation can take into account the correspondence between the detected lane markings and the specification of the digital map regarding the connection of a lane with lanes of subsequent sections. For example, if the lane indicated by a candidate position is only connected to a single lane of the subsequent section, but at the same time dashed lane markings are detected on the right and left, this can be used to reduce the evaluation of the candidate position.

In the next step of this exemplary method, a subgroup having a second number of candidate positions, S4, is selected from the first group of candidate positions, which has a first number of candidates. This selection can be made based on a threshold value for w _{c, t} or based on the first number, which can be specified. For example, it can be specified that the subgroup includes those candidate positions of the first group that have the highest ratings, where the second number is half of the first number (or, for example, 30% or 70% of the first number).

Based on the subgroup, an intermediate group of candidate positions z _t-1 is then determined, S5. The intermediate group of candidate positions again includes the first number of candidate positions. To achieve this, candidate positions of the subgroup are duplicated. The candidate positions of the subgroup that are duplicated can be selected based on the evaluation of the candidate position of the subgroup. Likewise, the number of duplications of a candidate position can be based on the evaluation.

Based on the candidate positions z _t-1 of the intermediate group, new candidate positions c _t are determined, which form the second group, S6. The candidate positions of the second group are the continuation of the candidate positions of the intermediate group according to the travel distance dt covered by the vehicle between the first time and the second time. In the continuation, the distance s _{z, t-1} of the respective candidate position of the subgroup is adjusted according to the determined travel distance, which is sometimes also called spreading: $${\text{s}}_{\text{c}\text{, t}}={\text{s}}_{\text{z}\text{, t}-1}+{\text{d}}_{\text{t}}{\mathrm{\alpha }}_{\text{c}}+{\mathrm{\beta }}_{\text{c}}$$

The distance s _{c, t} of a candidate position in the second group is thus calculated as the distance of the candidate position in the intermediate group (where the subgroup provides candidate positions for the time (t-1)*T) plus the travel distance dt, which is distorted by the noise factor α _c and the noise parameter β _c . The noise factor α _c is normally distributed with a mean of 1 and a standard deviation of, for example, 0.5, which represents inaccuracies in the determination of the travel distance. To this is added a normally distributed and mean-free noise parameter β _c , which is primarily intended to take into account errors in the digital map with regard to the positioning of streets, carriageways and lanes. The standard deviation for β _c is, for example, 20 m.

If a distance s _{c, t} is determined for a candidate position c _t of the second group that exceeds the end of the current section of the digital map, the section I _t following section I _t-1 is determined for this candidate position c _t . The distance s _{c, t} of the candidate position of the second group is adjusted accordingly to the new reference position (the start of the new section). Nevertheless, the distance between the candidate position of the intermediate group and the candidate position of the second group corresponds to the distance specified by the calculation, which is now simply distributed over several sections.

If there is more than one subsequent section for a section, the subsequent section is chosen with a probability of 90% according to the digital map specification of the section on which the lane is located that follows lane I _t-1 of the candidate position of the intermediate group. In this case, the lane for the candidate position of the intermediate group is also chosen according to the digital map's successor specification. In the remaining 10% of cases, the subsequent section and the lane for the candidate position of the second group are chosen randomly (in particular, uniformly distributed) among the possible sections and their respective lanes.

If there is only one subsequent section, the successor specification for the lane of the respective candidate position of the intermediate group is taken into account with a probability of 90%. In the other cases, the lane for the candidate position of the second group is chosen randomly (in particular, uniformly distributed). This strategy makes it possible to take into account errors in the digital map, in particular errors regarding the successor specification, and unusual driving maneuvers.

If the digital map does not specify subsequent sections or lanes, the subsequent sections and lanes are chosen randomly and, in particular, evenly distributed.

Optionally, the lanes determined for the candidate positions of the second group can be determined taking into account a further criterion, namely according to a lane change detected using the camera. If, for example, a lane change to the left was detected, in addition to the lane determination described above, the lane is re-determined for each candidate position in the second group, if possible. In the example of the lane change to the left, this is done by assigning a lane one further to the left than previously calculated. A lane change is detected by observing the distances of the left and right lane markings to the vehicle. Since lane change detection can be error-prone, a random (and evenly distributed) lane change to the left or right is assumed for a small part of the candidate positions in the second group, for example 10%, 20%, or 30%, and is reproduced accordingly in the determination of the lane for the corresponding candidate position. For the random lane change, it can be assumed that the vehicle changes lane with a probability of 20%. For this part of the candidate positions, the detected lane change plays no role.

This second group of candidate positions serves as the first group when the steps of the procedure are executed again and the steps of the procedure are executed starting from this group.

Finally, the position of the vehicle is estimated, S7, using the 3 described method is used.

2 shows an example of the second group of candidate positions. 2 shows a schematic representation of sections of roadways in a digital map. The absolute position 1 of the vehicle is marked with an arrow. The candidate positions are marked with points 2. In this example, the candidate positions are centered on the middle of the lanes. A lane branches off from the four-lane roadway. Vehicle 1 is on this lane, and its camera consequently recognizes a solid line as the right lane marking and a dashed line as the left lane marking. Using this knowledge, the candidate positions of the first group (not shown) are evaluated. According to the method outlined above, those candidate positions that are on the branching lane are rated better than the other candidate positions. Accordingly, candidate positions on the branching lane are frequently represented in the second group.

3 shows a schematic flow chart for the second part of the method according to an embodiment. In this second part, the position of the vehicle is estimated.

In the first step S8, the first set of candidate positions is provided together with their respective scores. This set of candidate positions and their scores can be determined using the 1 and 2 The procedure described can be determined using a different origin for the group of candidate positions and their respective evaluation.

In the next step S9, environment candidate positions are determined for each candidate position. These are all those candidate positions whose roadway distance to the candidate position under consideration is less than a predetermined threshold value, for example 10 m. The candidate position under consideration itself is also part of the environment candidate positions.

In the subsequent step S10, the minimum candidate position is determined for each candidate position, sometimes also called the median candidate position. To determine the minimum candidate position for the candidate position under consideration, the value of an optimization criterion is calculated for each environment candidate position that was determined for the candidate position under consideration. The optimization criterion is the sum of the weighted roadway distances from the respective environment candidate position currently under consideration to the other environment candidate positions. The weighting means that the roadway distance from the respective environment candidate position to the respective other environment candidate position is multiplied by the evaluation of the respective other environment candidate position. The environment candidate position that has the lowest value of the optimization criterion is the minimum candidate position for the candidate position under consideration. So, for a candidate position under consideration, an environment group of candidate positions is determined (all environment candidate positions for the candidate position under consideration, including the candidate position under consideration) and the minimum candidate position is determined for this environment group. This minimum candidate position is then assigned to the candidate position under consideration.

In step S11, the resulting assignment between each candidate position of the first group and the minimum candidate position determined for it is stored.

In the subsequent step S12, the cluster center candidate position is determined for each candidate position. This is done by searching for the associated minimum candidate position for a candidate position under consideration (or its previously determined minimum candidate position) using the assignment (lookup in the assignment). For the minimum candidate position found, the next minimum candidate position is then also searched for using the assignment. The minimum candidate position found is thus the starting candidate position for the next lookup in the assignment. This iterative lookup in the assignment ends when the minimum candidate position found corresponds to the candidate position for which the minimum candidate position was sought. In other words: when the input to the assignment corresponds to the output. The last starting candidate position is then the cluster center candidate position for the candidate position under consideration.

Then, in step S13, a rating is determined for each cluster center candidate position. The rating of a cluster center candidate position results from the sum of the ratings of those candidate positions in the first group for which this cluster center candidate position was determined. In other words, from the ratings of the candidate positions of a cluster. The rating of the candidate positions can be w _c,t or just w _{, t, dist} * w _{c, t, heading} .

In step S14, the cluster center candidate position having the highest score is selected as the estimate for the position of the vehicle.

In the last optional step S15, the method can also estimate a lane for the vehicle. To do this, first those candidate positions from the first group that are on the same section of the road or lane as the estimated vehicle position are eliminated. Then each lane is evaluated by considering the ratings of the candidate positions that are in the respective lane for each lane (for example summed and averaged). The lane with the highest rating is used as an estimate for the vehicle's lane.

4 shows schematically the clustering of candidate positions according to an embodiment. The illustration shows the branching according to 2 . The two lanes branching off at the junction are labelled 3 and 4. The candidate positions of the first group are shown by the individual lines, whereby the information of the lane is not shown, but only the “longitudinal position tion" on the lanes (which can be obtained from the distance information of the position of the candidate positions). The length of the lines symbolizes the evaluation of the respective candidate position. 4 also shows for which candidate positions the same media candidate position was determined. These candidate positions form a cluster. Membership in a cluster is shown by a similar symbol at the end of the respective line. Furthermore, 4 the cluster center candidate positions 5 to 8 for the respective cluster. The cluster center candidate positions 5 to 8 are marked by bold symbol repetitions at the level of the schematically shown lanes. In the case of 4 the cluster center candidate position 8 has the highest rating and would be used as an estimate for the candidate position.

Claims (9)

Method for estimating the position of a vehicle in a digital map, wherein the vehicle further comprises means for determining the absolute geographical position of the vehicle, in particular a satellite-based positioning system, wherein the map defines lanes, the method comprising: providing a group of candidate positions, wherein each candidate position is assigned a rating; wherein the rating assigned to each candidate position further takes into account a respective absolute distance which essentially corresponds to the distance between the respective candidate position and the determined absolute geographical position of the vehicle, in particular in a coordinate system; for each candidate position: determining a cluster center candidate position, wherein the cluster center candidate position is one of the candidate positions, wherein the lane distances between candidate positions are taken into account in the determination; wherein a lane distance is the distance from one candidate position to another candidate position along the lane or, in particular at junctions, along the lanes; Determining a score for each of the determined cluster center candidate positions, wherein the score for each cluster center candidate position is determined based on the scores of the candidate positions for which the respective cluster center candidate position was determined; Selecting a cluster center candidate position as an estimate for the position of the vehicle, taking into account the scores of the determined cluster center candidate positions. procedure according to claim 1 , wherein determining the cluster center candidate position for the respective candidate position comprises an iterative method in which candidate positions are selected one after the other that satisfy an optimization criterion, in particular a minimization criterion, based on the lane distance to other candidate positions, wherein the iterative method terminates and the cluster center candidate position is found when the same candidate position is selected directly one after the other, wherein this candidate position is determined as the cluster center candidate position. Method according to one of the preceding claims, wherein determining the cluster center candidate position for the respective candidate position comprises an iterative method in which the respective candidate position is the first starting candidate position, and the iterative method comprises: Determining the environment candidate positions, namely those candidate positions that are within a predetermined roadway distance from the starting candidate position, wherein the starting candidate position is also an environment candidate position; Determining a minimum candidate position, namely that environment candidate position that satisfies an optimization criterion, in particular the sum of the roadway distances to the other environment candidate positions or the sum of measures based on the roadway distances, in particular the roadway distances weighted with the respective evaluations of the other environment candidate positions, minimized; Repeating the two preceding steps, wherein the minimum candidate position is the starting candidate position; Stop the iteration if the starting candidate position and the designated minimum candidate position are identical, where the starting candidate position is then the cluster center candidate position. Method according to one of the Claims 1 or 2 , further comprising: For each candidate position: determining the surrounding candidate positions, namely those candidate positions that are within a predetermined roadway distance from the respective candidate position, wherein the respective Candidate position is also an environment candidate position; determining the minimum candidate position, namely that environment candidate position which satisfies an optimization criterion, in particular minimizes the sum of the roadway distances to the other environment candidate positions or the sum of measures based on the roadway distances, in particular the roadway distances weighted with the respective evaluations of the other environment candidate positions; assigning, and in particular storing the assignment, the minimum candidate position to the respective candidate position; determining the cluster center candidate position for each candidate position based on the assignment, this step comprising in particular an iterative method in which the respective candidate position or the minimum candidate position previously determined for it is the start candidate position, and the iterative method comprises: a) determining the minimum candidate position assigned to the start candidate position; b) repeating the previous step, whereby the determined minimum candidate position becomes the starting candidate position, until the starting candidate position corresponds to the associated minimum candidate position, whereby this starting candidate position is then the cluster center candidate position for the respective candidate position. Method according to one of the preceding claims, wherein the vehicle comprises: means for determining the orientation of the vehicle, in particular in relation to cardinal directions; wherein the rating assigned to each candidate position is based on the difference of the candidate orientation from the determined orientation of the vehicle, wherein the candidate orientation is the orientation of the roadway or a section of the roadway in the map on which the respective candidate position is located. Method according to one of the preceding claims, wherein those candidate positions for which the same cluster center candidate position was determined form a cluster group. Method according to one of the preceding claims, wherein the candidate positions are also assigned lanes; wherein the lane distance is determined in particular independently of the lanes; wherein the lanes are divided into sections; wherein the method further comprises: For each lane of the section on which the selected cluster center candidate position is located; summing or forming an average of the evaluations of the candidate positions that are located on the section in the respective lane; Determining the lane for estimating the vehicle position based on the sum or the average for the respective lane, in particular the highest sum or the highest average. Computer program which causes a computer to carry out a method according to one of the preceding claims when executing the computer program. Vehicle comprising: Electronic computing means for carrying out a method according to one of the Claims 1 until 7 are arranged, the vehicle comprising the means defined in the respective claim.

Images