DE102020204736A1 - Method and device for determining a probability of a vehicle's presence in one of several lanes of a roadway - Google Patents

Abstract

The invention relates to a method (200) for determining a probability (145) of a location of a vehicle (100) in one of several lanes (110) of a roadway (115), the method (200) comprising the one step of reading in (210) having lane data (127) representing a number of lanes (110) of a lane (115) traveled by the vehicle (100) and / or a possibility to switch between the lanes (110), and lane change data (137) representing one of represent a lane change of the vehicle (100) detected by a lane change sensor (135), and lane delimitation type data (141) are read in, which a type of a lane delimitation (143, 117) detected by a lane delimitation type sensor (137) of the lane currently used by the vehicle (100) ( 110) represents. The method (200) further comprises a step of linking (220) the lane data (127), the lane change data (137) and the lane delimitation type data (141) in order to determine a probability (145) of the vehicle (100) for one of several lanes ( 110) of the roadway (115) to be determined. Finally, the method (200) comprises a step of selecting (230) that lane (110) as the lane (110) in which the vehicle (100) is currently located and which has the highest determined probability of being (145).

Description

State of the art

The invention is based on a device or a method according to the preamble of the independent claims. The present invention also relates to a computer program.

Up to now, lane-accurate navigation can be carried out primarily with a very computationally expensive evaluation of signals from very simple sensors (for example video-based) in order to enable vehicles to be located on a map with precise lane-accuracy. Alternatively, a vehicle can be located down to the centimeter using expensive technology (high-precision GNSS sensors, radar sensors, lidar sensors). If highly accurate maps are also available on board the vehicle, a lane can be determined. However, this also requires a high level of available computing power in the vehicle, which often should not be kept available on this scale or should be available in real time for other driver assistance tasks, especially in the field of highly autonomous driving. For this reason, there is an increased need to create a simpler solution for determining the position of a vehicle on a roadway with lane accuracy.

Disclosure of the invention

Against this background, with the approach presented here, a method, furthermore a device which is set up to execute or control this method, and finally a corresponding computer program according to the main claims are presented. The measures listed in the dependent claims make advantageous developments and improvements of the device specified in the independent claim possible.

• - Einlesen von
  • ◯ Spurdaten, die eine Anzahl von Fahrspuren einer von dem Fahrzeug befahrenen Fahrbahn und/oder eine Wechselmöglichkeit zwischen den Fahrspuren repräsentieren, und von
  • ◯ Spurwechseldaten, die einen von einem Spurwechselsensor erkannten Fahrspurwechsel des Fahrzeugs repräsentieren, und von
  • ◯ Fahrbahnbegrenzungstypdaten, die einen Typ einer von einem Fahrbahnbegrenzungstypsensor erkannten Fahrbahnbegrenzung der aktuell von dem Fahrzeug benutzten Fahrspur repräsentieren;
• - Verknüpfen der Spurdaten, der Spurwechseldaten und der Fahrbahnbegrenzungstypdaten, um eine Aufenthaltswahrscheinlichkeit des Fahrzeugs für je eine von mehreren Fahrspuren der Fahrbahn zu bestimmen; und
• - Auswählen derjenigen Fahrspur als Fahrspur, auf der sich das Fahrzeug aktuell befindet, welche die höchste bestimmte Aufenthaltswahrscheinlichkeit aufweist.

The approach presented here creates a method for determining a probability of a vehicle's presence in one of several lanes of a roadway, the method having the following steps:

• - Reading from
  • ◯ Lane data that represent a number of lanes of a roadway on which the vehicle is traveling and / or a possibility to switch between the lanes, and from
  • ◯ Lane change data, which represent a lane change of the vehicle detected by a lane change sensor, and from
  • ◯ lane boundary type data representing a type of a lane boundary of the lane currently used by the vehicle detected by a lane boundary type sensor;
• Linking the lane data, the lane change data and the lane delimitation type data in order to determine a probability of the vehicle being present for one of several lanes in the lane; and
• - Selecting that lane as the lane in which the vehicle is currently located, which has the highest certain probability of being present.

In the present case, lane data can be understood to mean information about how many lanes and / or which options to change from the available lanes are available on the current lane on which the vehicle is traveling. In the present case, lane change data can be understood to mean information from a lane change sensor as to whether this lane change sensor has detected a lane change. This information can, for example, also indicate a direction in which the vehicle has changed or be present or supplemented in the form of a probability that gives an indication of how probable an actual lane change is that has been recognized as such by the lane change sensor. In the present case, lane delimitation type data can be understood to mean information about a type of lane delimitation. For example, such information about a kind or type of the lane boundary in the form of a solid line, a dashed line, a lawn edge, a sand bank or the like can be mapped by the lane boundary type data. Linking can be understood to mean, for example, a logical, algebraic or algorithmic mathematical link. In the linking step, the probabilities of the location of the vehicle for several available lanes on the lane (currently being used by the vehicle) can be determined.

The approach presented here is based on the knowledge that by linking information that can usually be supplied by already existing and / or inexpensive sensors, a very precise detection of the position of a vehicle in one of several possible lanes of a roadway is technically very simple can be. Here, a certain uncertainty can be allowed for the determination of this position, but this is due to the processing of different types of information, preferably from different sensors or one different evaluation of information from a sensor can be determined, can be kept very low. In this way, a numerical or circuit-technical effort for the highly precise determination of the position of the vehicle on a roadway can be kept very low and thus an advantage can be achieved over previously known solutions.

An embodiment of the approach proposed here is particularly favorable in which, in the reading-in step, the lane data are read in from a digital map and / or an optical sensor and / or the lane change data and / or the lane delimitation type data are read in using an optical sensor. Such an embodiment of the approach proposed here offers the advantage that, by reading out the track data from a digital map, it is usually possible to obtain quite precise information from sources that are often already available. Such a digital map can, for example, be carried on board a vehicle for use in a vehicle navigation system. As an alternative or in addition, a digital map stored in an online database can also be accessed via a communication link, for example via an Internet connection, or it can be loaded into the vehicle, in particular excerpts. However, it is also conceivable to determine the lane data using signals from an optical sensor such as a camera, although in this case it should be noted that even modern camera systems already enable a fairly precise classification of the roadway into different lanes with technically relatively simple algorithms. Advance position data can be understood to mean, for example, information about the position of the vehicle in a specific lane of the roadway at a previous point in time, that is to say a point in time that is earlier than the point in time at which it was viewed, here the point in time before or an advance point in time. Analogously, the advance lane data can correspond to the aforementioned information on lane data that was current at this previous point in time, that is, when the vehicle was driving on the roadway at this previous point in time. Likewise, the advance lane delimitation type data can correspond to the aforementioned information on lane delimitation type data which was also current at this previous point in time, that is to say corresponds to a type of lane delimitation when the vehicle was driving on the roadway at this previous point in time. Such an embodiment of the approach proposed here offers the advantage of achieving a further improvement in precision when determining this probability of the vehicle's presence in one of several lanes of a road by taking into account previous information, i.e. an iterative calculation of the probability of presence.

According to a further embodiment of the approach proposed here, advance position data, advance lane data, advance lane change data and / or advance lane delimitation type data can also be read in in the step of reading in, and in the step of linking the advance position data, advance lane data, advance -Lane change data and / or advance lane delimitation type data are linked to the lane data, the lane change data and / or the lane delimitation type data in order to determine the probability of stay, the advance position data representing a position of the vehicle at a previous point in time on the lane, the advance lane data which represent a number of advance lanes of a lane traveled by the vehicle at a previous point in time and an option to change between the advance lanes, the advance lane change data indicating a lane change of the vehicle detected by the lane change sensor at the previous point in time and the advance lane boundary type data represents a type of a lane boundary of the lane used by the vehicle at the previous point in time that is recognized by the lane boundary type sensor.

An embodiment of the approach proposed here is also advantageous in which, in the reading-in step, advance position data are read in that correspond to a lane that had the highest probability of staying in a previous method step of selection. Such an embodiment of the approach presented here offers the advantage of a very precise, consecutive, iterative calculation of the position of the vehicle through the use of the advance position data corresponding to a position which had the highest probability of being in a previous method step of selection to enable. In this way, a further increase in the forecast quality for determining the position of the vehicle in one of the available lanes can be achieved with technically simple means.

An embodiment of the approach proposed here is particularly advantageous in which at least one lane change data probability and / or at least one lane delimitation type data probability is imported in the step of reading and in the step of linking the at least one lane change data probability and / or at least one lane delimitation type data probability to determine the probability of the Vehicle is used. In this case, lane change data probability can represent a probability that a lane change detected by the lane change sensor has actually taken place. The lane delimitation type data probability can also represent a probability that a lane delimitation recognized by the lane delimitation type sensor has corresponded to a type of an actually present lane delimitation. A lane change data probability can be understood, for example, as a value of how likely a change in lanes that has actually taken place will also be recognized as such and / or how likely that a change in lane that has actually taken place will not be recognized or that a change in lane that has not taken place will be recognized as a lane change Change of lanes is recognized. Similarly, a lane delimitation type data probability can also be understood as a value of how likely an actually present type of a lane delimitation, for example on a left and / or a right side of the lane currently being traveled by the vehicle, has been correctly recognized or how likely an actually present type of one has been recognized Lane delimitation is classified as a different type of lane delimitation or how likely another type of lane delimitation is classified as the actually present type of lane delimitation of the lane currently being driven in by the vehicle. Such an embodiment of the approach proposed here offers the advantage of being able to keep a numerical and / or circuit-related complexity of the approach proposed here low by taking into account the corresponding probabilities and still having sufficiently precise results for the probability of the vehicle being in one of several possible lanes of a roadway to obtain.

An embodiment of the approach proposed here is also conceivable in which the at least one lane change data probability is read from a predetermined lane change data probability table and / or the at least one lane delimitation type data probability from a predetermined lane delimitation type data probability table. Such a lane change data probability table can be understood as a tabular overview or data arrangement in which the probabilities of a correct assignment of the actually present lane to a recognized lane or the probability of a lane other than the actually present lane is entered. Analogously, a table overview or data arrangement can also be understood in a lane delimitation type data-probabilities table, in which the probabilities of a correct assignment of the actually present type of a lane delimitation to a recognized type of a lane delimitation or the probability of another recognized type of the lane delimitation to the actually present type the lane boundary is entered. Such an embodiment of the approach proposed here offers the advantage that, for example, the corresponding sensors for determining the lane change data or the lane delimitation type data are classified or measured in advance with regard to their recognition quality, so that the use of corresponding, for example predetermined, tables significantly simplifies the determination of the The likelihood of the vehicle being on one of several lanes of the roadway is made possible.

According to a particular embodiment of the approach proposed here, in the reading-in step, the predetermined lane change data probability table can also be selected from a plurality of different lane change data probability tables as a function of at least one environmental parameter. Alternatively or additionally, the predetermined lane delimitation type data probability table can be selected from a plurality of different lane delimitation type data probability tables as a function of at least one environmental parameter. In particular, the environmental parameter can represent a lighting situation in an area around the vehicle and / or a road surface condition in the area around the vehicle. Such an embodiment of the approach proposed here offers the advantage of being able to take into account the detection quality of the corresponding sensors with regard to different environmental parameters, for example whether a lane is wet or caved due to rain or snow or whether at night it is less possible to differentiate between the individual lanes or types of lane boundaries occurs. In this way, depending on the application scenario, a different table can be used to determine the lane change data or the lane delimitation type data, so that a further increase in the precision of the probability of the vehicle being in one of several lanes of a lane can be achieved with technically simple means.

An embodiment of the approach proposed here is particularly easy to implement, in which, in the linking step, the at least one lane change data probability is multiplicatively linked to the at least one lane delimitation type data probability. Such an embodiment of the approach proposed here also has the advantage circuitry and / or numerically very simple means to be able to determine a very precise probability of the vehicle being in one of several lanes of a roadway. For example, the probabilities of stay over all currently available lanes can also be normalized in this context, so that irritation can be avoided if, for example, a sum over this on the probabilities of stay of all available lanes does not lead to a value of 1.

An embodiment of the approach proposed here, in which the steps of the method are carried out repeatedly, in particular with a frequency of not more than 1 Hertz, is particularly reliable. Such an embodiment of the approach proposed here offers the advantage of being able to monitor the position of the vehicle at small time intervals as possible through the recurring determination of the probability of the vehicle being located in one of several possible lanes of a roadway, so that there may be an impending traffic safety hazard, for example through an unintentional departure of the vehicle from the current lane can be detected as early as possible and appropriate countermeasures such as issuing a warning or active intervention in the vehicle control system can be initiated. It is also conceivable that the approach presented here for checking the plausibility of another method for determining the position of the vehicle on a roadway can be used.

This method can be implemented, for example, in software or hardware or in a mixed form of software and hardware, for example in a control device.

The approach presented here also creates a device which is designed to carry out, control or implement the steps of a variant of a method presented here in corresponding devices. The object on which the invention is based can also be achieved quickly and efficiently by means of this embodiment variant of the invention in the form of a device.

For this purpose, the device can have at least one processing unit for processing signals or data, at least one storage unit for storing signals or data, at least one interface to a sensor or an actuator for reading in sensor signals from the sensor or for outputting data or control signals to the Have actuator and / or at least one communication interface for reading in or outputting data, which are embedded in a communication protocol. The computing unit can be, for example, a signal processor, a microcontroller or the like, wherein the storage unit can be a flash memory, an EEPROM or a magnetic storage unit. The communication interface can be designed to read in or output data wirelessly and / or wired, a communication interface that can read in or output wired data, for example, can read this data electrically or optically from a corresponding data transmission line or output it into a corresponding data transmission line.

In the present case, a device can be understood to mean an electrical device that processes sensor signals and outputs control and / or data signals as a function thereof. The device can have an interface which can be designed in terms of hardware and / or software. In the case of a hardware design, the interfaces can, for example, be part of a so-called system ASIC which contains a wide variety of functions of the device. However, it is also possible that the interfaces are separate, integrated circuits or at least partially consist of discrete components. In the case of a software-based design, the interfaces can be software modules that are present, for example, on a microcontroller alongside other software modules.

A computer program product or computer program with program code, which can be stored on a machine-readable carrier or storage medium such as a semiconductor memory, a hard disk memory or an optical memory, and for carrying out, implementing and / or controlling the steps of the method according to one of the embodiments described above is also advantageous is used, especially when the program product or program is executed on a computer or device.

• 1 zeigt ein Blockschaltbilddarstellung eines Fahrzeugs mit einem Ausführungsbeispiel einer Vorrichtung zum Ermitteln einer Aufenthaltswahrscheinlichkeit eines Aufenthalts des Fahrzeugs auf einer von mehreren Fahrspuren einer Fahrbahn;
• 2 ein Ablaufdiagramm eines Ausführungsbeispiels eines Verfahrens zum Ermitteln einer Aufenthaltswahrscheinlichkeit eines Aufenthalts eines Fahrzeugs auf einer von mehreren Fahrspuren einer Fahrbahn;
• 3 eine beispielhafte Verwechslungsmatrix zur Abbildung der korrekten Erfassung eines Spurwechsels durch den Spurwechselsensor;
• 4 beispielhaft eine Straßensituation mit einer beispielhaften Konnektivität der Spuren zwischen zwei Straßen;
• 5 eine grafische Repräsentation für das in der 4 wiedergegebene Beispiel der oben beschriebenen Berechnung von Qⁱ;
• 6 eine beispielhafte Verwechslungsmatrix für einen Spurmarkierungstyperkennungsensor zur Abbildung der korrekten Erfassung eine Qualität der Erkennung eines Typs einer Fahrbahnbegrenzung durch den Fahrbahnbegrenzungstypsensor; und
• 7 ein Diagramm zur Darstellung eines Informationsflusses eines Ausführungsbeispiels des hier vorgestellten Ansatzes.

Embodiments of the approach presented here are shown in the drawings and explained in more detail in the description below. It shows:

• 1 shows a block diagram representation of a vehicle with an embodiment of a device for determining a probability of a stay of the vehicle in one of several lanes of a roadway;
• 2 a flowchart of an exemplary embodiment of a method for determining a probability of a location of a vehicle in one of several lanes of a roadway;
• 3 an exemplary confusion matrix for mapping the correct detection of a lane change by the lane change sensor;
• 4th exemplary a street situation with exemplary connectivity of the lanes between two streets;
• 5 a graphical representation for that in the 4th reproduced example of the above-described calculation of Q ⁱ ;
• 6th an exemplary confusion matrix for a lane marking type recognition sensor for mapping the correct detection of a quality of the recognition of a type of a lane boundary by the lane boundary type sensor; and
• 7th a diagram to illustrate an information flow of an embodiment of the approach presented here.

In the following description of advantageous exemplary embodiments of the present invention, identical or similar reference numerals are used for the elements shown in the various figures and having a similar effect, a repeated description of these elements being dispensed with.

1 Figure 3 shows a block diagram representation of a vehicle 100 with an embodiment of a device 105 to determine a probability of a location of the vehicle 100 on one of several lanes 110 a roadway 115 . The roadway 115 includes according to the 1 example shown two lanes 110a and 110b , with the vehicle 100 in the right lane 110a is located. The right lane 110a is on the right-hand side by a solid lane delimitation 117a to mark an edge of the roadway and limited on its left side by means of a dashed line 117b limited to the right lane 110a from a left lane 110b to separate.

The vehicle also includes an optical sensor 120 , which is designed for example in the form of a mono camera or a stereo camera, and which is an optical image of the roadway 115 or the lanes arranged on it 110a and 110b as an image signal 122 supplies. The image signal 122 becomes a track data unit according to this embodiment 125 fed in which track data 127 can be determined which have a number of lanes 110 that of the vehicle 100 traffic lane 115 and / or an option to switch between the lanes 110 represents. Alternatively or additionally, the track data unit 125 Data from a digital map 129 read out the track data 127 which also represent a number of lanes 110 that of the vehicle 100 traffic lane 115 and / or an option to switch between the lanes 110 represents. The track data 127 are via an interface 131 the device 105 read in and a linking unit 133 fed.

The vehicle also includes 100 a lane change sensor 135 that the image signal 122 from the optical sensor 120 reads in and from this lane change data 137 generated and via the interface 131 the linking unit 133 provides, with the lane change data 137 one of the lane change data for 135 recognized change of lane 110 (ie a lane change) of the vehicle 100 represent. For example, this lane change sensor 135 can detect whether one of the lane markings 117a or 117b from the vehicle 100 was transmitted and the lane change data responding to this 137 to generate.

Also includes the vehicle 100 a lane boundary type sensor 139 , which is designed to store the image data 123 read in and from this lane delimitation type data 141 to generate the one type of the lane boundary type sensor 139 recognized lane boundary 143 , such as the solid road marking here 117a , the current of the vehicle 100 traffic lane 110 represents. The lane boundary type data 141 are also used via the interface 131 the linking unit 133 the device 105 fed.

In the linking unit 133 will be track data below 127 , the lane change data 137 and the lane boundary type data 141 linked to a probability of presence 145 of the vehicle 100 fixes one of several lanes 110 the roadway 115 to determine. For example, a first probability of being 145a determines which represents the likelihood that the vehicle will move 100 in the right lane 110a is located. A second probability of presence becomes analogous 145b determines which represents the likelihood that the vehicle will move 100 in the left lane 110b is located. These probabilities of residence 145 become a selection unit 150 in which that lane 110 is determined to be the lane on which the vehicle is located 100 is currently located, which is the highest of the specific probabilities of residence 145 exhibit. This lane is called the current position of the vehicle 100 in a corresponding lane signal 152 coded and, for example, on a driver assistance system 155 of the vehicle 100 forwarded for further processing. For example, the driver assistance system 155 be a lane guidance system or a navigation system, which depends on the recognized lane 110 that the vehicle is currently located, steering interventions or route recommendations are determined and transferred to an automatic steering support system of the vehicle 100 or a driver of the vehicle 100 issues.

The device also includes 105 a first storage unit 157 , in which a lane change data-probability table 159 is stored and from which at least one lane change data probability 161 through the interface 131 in linkage unit 133 is loaded. The lane change data probability table 159 contains several lane change data probabilities 161 that a quality of the detection of a lane change by the lane change sensor 135 map, for example by the fact that these probabilities 161 map whether a lane change sensor 135 recognized the lane change data actually took place or the probability with which a lane change that has not taken place is assessed as a lane change that has actually taken place.

Analogously, the device comprises 105 also a second storage unit 163 , in which a lane boundary type data probability table 165 is stored and from which at least one lane delimitation type data probability 167 through the interface 131 into the linking unit 133 is loaded. The lane boundary type data probabilities table 165 also includes multiple lane boundary type data probabilities 167 that is a quality of detection of a type of lane boundary by the lane boundary type sensor 139 map, for example by the fact that these probabilities map whether a lane delimitation type sensor 139 recognized type of lane boundary 143 an actually existing type of lane boundary 143 corresponds to or as a (wrong) different type of lane boundary 143 is interpreted.

In the linking unit 133 are then the probabilities of residence 145 also using the lane change data probabilities 161 and the lane boundary type data probabilities 167 determined.

It is also conceivable that a third memory 169 is provided in which the linking unit 133 supplied or from the linking unit 133 detected signals, i.e. the track data 127 , the lane change data 137 , the lane boundary type data 141 who have favourited lane change data probabilities table 159 , the lane change data probability 161 , the lane boundary type data probabilities table 165 and / or the lane delimitation type data probability 167 or the probabilities of stay 145 for the respective lane 110 as input signals 171 can be read in and stored over a certain period of time. Alternatively or additionally, the current position of the vehicle can also be used 100 that are in the corresponding lane signal 152 is encoded, as an advance position signal of the position of the vehicle 100 as an advance lane signal 152 ' in the third memory 169 be stored and for a subsequent step of linking in the linking unit 133 be used. After this period of time, for example when a new integration cycle for determining the probabilities of stay 145 can be used to determine the current probabilities of residence 145 from memory 169 the signals stored in it as advance values, i.e. as advance track data 127 ' , as advance lane change data 137 ' , as advance lane boundary type data 141 ' , as advance lane change data probabilities table 159 ' , as advance lane change data probability 161 ' , as an advance lane boundary type data probabilities table 165 ' , the pre-stay probabilities 145 ' for the respective lane 110 and / or as advance lane delimitation type data probability 167 ' into the linking unit 133 are loaded and thereby to determine the probabilities of stay 145 must be taken into account. In this way, the successive or continuous determination of the probabilities of residence 145 of the vehicle in a lane 110 the roadway 115 these probabilities of residence 145 very precisely determined.

It is also conceivable that in the first memory 157 several lane change data probability tables 159 are stored, the probabilities for the correct detection of the lane change by the lane change sensor 135 are assigned in different environmental conditions in which the vehicle 100 is currently located (for example, when the road surface is dry or wet, or in daylight or at night). For example, in the interface 131 and / or the linking unit 133 for the current environmental conditions around the vehicle 100 the appropriate table 159 or the appropriate probabilities 161 selected and for the determination of the current probability of stay 145 used. Analog can also be done in the second memory 165 multiple lane boundary type data probability tables 165 be the probabilities of correctly detecting the type of lane boundary by the lane boundary type sensor 139 are assigned in different environmental conditions in which the vehicle is 100 is currently located (for example, when the road surface is dry or wet, or in daylight or at night). For example, in the interface 131 and / or the linking unit 133 for the current environmental conditions around the vehicle 100 the appropriate table 165 or the appropriate probabilities 167 selected and for the determination of the current probability of stay 145 used.

2 shows a flowchart of an exemplary embodiment of a method 200 for determining a probability of a vehicle's presence in one of several lanes of a roadway. The procedure 200 includes one step 210 the reading in of lane data, which represents a number of lanes of a lane traveled by the vehicle and / or a possibility to change between the lanes, and of lane change data, which represent a lane change of the vehicle detected by a lane change sensor, and of lane delimitation type data, which is a type one of represents a lane delimitation type sensor recognized lane delimitation of the lane currently used by the vehicle. The method also includes 200 one step 220 the linking of the lane data, the lane change data and the lane delimitation type data in order to determine a probability of the vehicle being present for one of several lanes in the lane. Finally, the procedure includes 200 one step 230 the selection of that lane as the lane on which the vehicle is currently located and which has the highest specific probability of being present.

• Der Spurwechselsensor 135 erkennt beispielsweise, ob das Fahrzeug 100 einen Spurwechsel, also einen Wechsel der Fahrspuren 100, vollzogen hat und kann beispielsweise auch dessen Richtung angeben (nach Links gewechselt / nach Rechts gewechselt). Der Fahrbahnbegrenzungstyperkennungssensor, der vorliegend auch als Fahrbahnbegrenzungstypsensor 139 bezeichnet wird, erkennt die linke und rechte Fahrbahnbegrenzung 143 bzw. Fahrbahnbegrenzungsmarkierung 117 und kann dessen Typ (durchgezogene/gestrichelte Linie, Bordstein, Rasenkante, etc.) als entsprechenden Typ der Fahrbahnbegrenzung ermitteln.

Further details of the determination of the probability of a vehicle's presence are given below 100 on one of several lanes 110 a roadway 115 are explained in more detail. It should be noted here that the approach presented here is a computation model for the fusion of sensor data with the aim of locating a vehicle with lane accuracy 100 on a road or lane 115 contains. A system that enables the vehicle to be precisely located on the street is particularly helpful in this context. The method described here then determines, for example, on which lane of the road (which, for example, is also synonymous as a lane 115 can be designated) the vehicle 100 stops. The data from two sensors (here specifically the lane change sensor 135 and the lane boundary type sensor 139 ) merged with each other:

• The lane change sensor 135 detects, for example, whether the vehicle 100 a lane change, i.e. a change of lanes 100 , and can, for example, also indicate its direction (changed to the left / changed to the right). The lane delimitation type recognition sensor, which in the present case also acts as a lane delimitation type sensor 139 is designated, recognizes the left and right lane boundary 143 or lane marking 117 and can determine its type (solid / dashed line, curb, lawn edge, etc.) as the corresponding type of lane boundary.

Neither of the two sensors 135 respectively. 139 always works reliably, ie individual measurements from these sensors 135 respectively. 139 can be faulty. The approach presented here combines the data from the two sensors 135 respectively. 139 and thus achieves a higher quality of location.

Together with a lane-accurate map, for example as a digital map 129 is stored in a memory, in observations, recorded over a coherent time interval, can be used to determine the exact lane position of the vehicle at any point in time 100 in the map 129 to investigate.

A preprocessing step can be carried out for this purpose. This determines how good the sensors are 135 respectively. 139 work. Corresponding confusion matrices are determined for both sensors, such as those in the first memory, for example 157 as a lane change data probability table 159 with corresponding lane change data probability stored therein 161 and in the second memory 163 as a lane boundary type data probability table 165 with corresponding lane delimitation type data probabilities stored therein 167 are filed. These confusion matrices indicate how often one of the sensors is 135 respectively. 139 confused a given situation with another. For example, the vehicle is leading 100 a lane change to the left, but a lane change to the right or no lane change at all is recognized. This preprocessing step only needs to be performed once, before executing the procedure. If necessary, such a preprocessing step can be used for the sensors 135 respectively. 139 can also be carried out several times under different environmental conditions, for example for a wet or dry road 115 or for the detection of the lane change or the type of lane boundary in daylight or at night, since a different probability of correct or incorrect detection of a lane change or the type of lane boundary can also occur in these different environmental conditions.

Furthermore, if the vehicle 100 drives, for example, periodically, z. B. 1 time per Second, the sensor data, here for example the lane change data 137 and the lane boundary type data 141 queried and with the aid of this data and preferably with the aid of map information as the lane data 127 the position of the vehicle 100 at lane level or the current lane 110 certainly. This is done to each lane 110 the probability 145 calculated with which the vehicle 100 on this lane 110 is located. The process runs iteratively: The probabilities 145 ' or the other advance data which were determined in the last or chronologically preceding iteration, that is to say the advance data which are in the third memory 169 are used as the starting point for the next, i.e. the chronologically following, iteration. An important aspect of the approach presented here is the calculation of these (residence) probabilities 145 .

The approach presented here enables a very cost-effective determination of the lane 110 of a vehicle 100 can be made, which can be built on mostly already existing sensor signals.

1. 1. Spurdaten aus Karte für aktuelle Position entnehmen
2. 2. Sensordaten empfangen
3. 3. Berechnung der Aufenthalteswahrscheinlichkeiten für die Spuren an der aktuellen Position aus Sensordaten der Spurwechselerkennung
4. 4. Berechnung der Aufenthalteswahrscheinlichkeiten für die Spuren an der aktuellen Position aus Sensordaten der Spurmarkierungstyperkennung
5. 5. Bestimmen der aktuellen Fahrspur des Fahrzeuges

An important aspect of the approach presented here is the calculation of the probabilities of stay over all lanes of the road currently being driven on by a vehicle on the basis of sensor information from lane change detection and lane marking type detection. For example, the iterative procedure is as follows:

1. 1. Obtain lane data from the map for the current position
2. 2. Receive sensor data
3. 3. Calculation of the probabilities of stay for the lanes at the current position from sensor data of the lane change detection
4. 4. Calculation of the probabilities of stay for the lanes at the current position from sensor data of the lane marking type recognition
5. 5. Determine the current lane of the vehicle

• Spurdaten aus Karte für aktuelle Position entnehmen Wie die straßengenaue Position eines Fahrzeuges 100 auf einer digitalen Karte 129 ermittelt wird, kann auf unterschiedliche Weise erfolgen. Der hier vorgestellten Ansatz baut beispielsweise darauf auf, dass in der digitalen Karte 129 zu jeder verzeichneten Straße Informationen über die Anzahl der auf der Straße verfügbaren Spuren und deren Spurbegrenzungsmarkierungstypen, z. B.: durchgezogen, gestrichelt, Bordstein, etc., vorhanden sind. Darüber hinaus ist in der Karte 129 verzeichnet, wie Spuren von unterschiedlichen Straßen miteinander verbunden sind, das heißt die Konnektivität oder die Wechselmöglichkeit zwischen den einzelnen Fahrspuren 110. Ist die straßengenaue Position des Fahrzeuges 100 ermittelt, kann somit aus der Karte 129 die Typen der Spurbegrenzungen sowie die Konnektivität ausgelesen werden.

An exemplary embodiment for each step is described in more detail below:

• Take lane data from the map for the current position Like the street-precise position of a vehicle 100 on a digital map 129 determined can be done in different ways. The approach presented here is based, for example, on the fact that in the digital map 129 for each recorded road information on the number of lanes available on the road and their types of lane delimitation markings, e.g. E.g .: solid, dashed, curb, etc., are present. It is also in the map 129 records how lanes from different roads are connected to one another, i.e. the connectivity or the ability to switch between the individual lanes 110 . Is the exact street position of the vehicle 100 can thus be determined from the map 129 the types of lane boundaries and the connectivity can be read out.

Receive sensor data

The vehicle is capable of the lane marking types of each to the vehicle 100 next lane markings 117 to investigate. These are further described with a left 117b or right 117a Lane markings, each marked in the direction of travel. In addition, the vehicle can 100 determine when it has changed lanes. The direction, i.e. in particular to the left / right in the direction of travel, can also be recognized for each lane change.

Calculation of the probabilities of stay for the lanes at the current position from sensor data of the lane change detection

3 shows an exemplary mix-up matrix W. for mapping the correct detection of a lane change by the lane change sensor 135 , this confusion matrix W. for example the one in the first memory 157 stored lane change data probabilities table 159 with the individual lane change data probabilities 161 is equivalent to. the end W. the uncertainty of the sensor can be taken. Here, the designation R is a change to the right with the designation L, a change to the left with the designation K no recognized lane change is shown. This matrix shows (in the columns how the actual change is shown) the probability with which an event has occurred, provided that the sensor has made a certain observation - the detected change is shown in the rows. In the following we make a note W. [a; b] the probability that event b has occurred but the sensor has detected event a. W. [To the right); K (one)], here = 5%, thus indicates that there is a 5 percent probability that no lane change has been carried out, although the sensor has detected a lane change to the right.

wobei $$\text{Lage}\left(u,v\right)=\{\begin{array}{ll}Links\hfill & \text{wenn }v\text{ links von }u\text{ liegt}\hfill \\ Rechts\hfill & \text{wenn }v\text{ rechts von }u\text{ liegt}\hfill \\ Keiner\hfill & \text{wenn }v\text{ gerade aus von }u\text{ liegt}\hfill \end{array}$$

4 zeigt beispielhaft eine Straßensituation. Hierbei ist eine beispielhafte Konnektivität der Spuren zwischen zwei Straßen. In diesem Fall lauten die Konnektivitäten von Vorab-Fahrspuren 110' zu - im nächsten Iterationsschritt befahrbaren Fahrspuren 110 - wie folgt: 1 →{1, 2, 3}, 2→ {2, 3, 4}, 3→{3, 4, 5}. Die Spuren des Aufenthaltsortes der letzten (Vorab-) Iteration (i-1) im unteren Bereich der abgebildeten Fahrspuren 110' die Spuren des aktuellen (i) Aufenthaltsortes im oberen Bereich. Der Pfeil gibt an, dass der Spurwechselsensor 135 einen Spurwechsel nach rechts detektiert hat.The input of the calculation in this step of the method is the result P ^{i-1 of} the calculation of the probabilities of stay 145 from the last iteration (i-1) as well as the connectivity K ^{i of} the tracks 110 between the whereabouts (the street 115 ) the last iteration (i-1) and this iteration (i). May be S ⁱ = {s ⁱ ₁ ; ...; s ⁱ _n } is the set of traces in the digital map in iteration i. K ⁱ [s ^i-1 _u , s ⁱ _v] indicates whether there is connectivity between track s ^i-1 _u ε S ^i-1 (from the last iteration) and track s ⁱ _v ε S ⁱ (from the current Iteration). Whereby a connectivity between two lanes 110 exists if at most one lane change is necessary (and legal) to switch between lanes. If there is connectivity between lanes s ^i-1 _u and s ⁱ _v , then K ⁱ [s ^i-1 _u ; s ⁱ _v] = 1, otherwise K ⁱ [s ^i-1 _u ; s ⁱ _v] = 0. The goal of the calculation in this step, Q ', specifies the probabilities of staying in iteration i if only the lane change detection sensor is used. The probabilities of stay Q ⁱ = {q ⁱ ₁ , ..., q ⁱ _n } the current iteration i, if a lane change to X ε {left; To the right; None} was recognized, can now be calculated for each track u as follows: $$q_u^i={\displaystyle \sum _{s_v^{i-1}\in s^{i-1}}{\mathrm{P.}}_v^{i-1}}.\frac{K^i\left[s_v^{i-1},s_u^i\right]\cdot \mathrm{W.}\left[X,\text{location}\left(s_v^{i-1},s_u^i\right)\right]}{{\sum }_{s_l^i\in s^i}{K}^i\left[s_v^{i-1},s_l^i\right]\cdot \mathrm{W.}\left[X,\text{location}\left(s_v^{i-1},s_l^i\right)\right]}$$

whereby $$\text{location}\left(u,v\right)=\{\begin{array}{ll}\mathrm{L.}inks\hfill & \text{if}v\text{to the left of}u\text{lies}\hfill \\ \mathrm{R.}ecHts\hfill & \text{if}v\text{right of}u\text{lies}\hfill \\ Keiner\hfill & \text{if}v\text{straight out of}u\text{lies}\hfill \end{array}$$

4th shows an example of a street situation. Here is an exemplary connectivity of the lanes between two streets. In this case, the connectivities are from advance lanes 110 ' to - in the next iteration step drivable lanes 110 - as follows: 1 → {1, 2, 3}, 2 → {2, 3, 4}, 3 → {3, 4, 5}. The traces of the whereabouts of the last (preliminary) iteration (i-1) in the lower area of the lanes shown 110 ' the traces of the current (i) whereabouts in the upper area. The arrow indicates that the lane change sensor 135 has detected a lane change to the right.

5 shows a graphical representation for that in the 4th reproduced example of the above-described calculation of Q ⁱ .

Calculation of the probabilities of stay for the lanes at the current position from sensor data of the lane marking type recognition

From the digital map 129 is for the current location of the vehicle 100 known which lane 110 what types of lane markers 117 in the street 115 are upset. In addition, there is the mix-up matrix M. known for the lane marking type recognition sensor.

6th shows an exemplary mix-up matrix M. for a lane marking type recognition sensor 139 for mapping the correct detection, a quality of the detection of a type of a lane boundary by the lane boundary type sensor 139 , this confusion matrix M. for example the one in the second memory 163 stored lane delimitation type data probability table 165 with the associated lane delimitation type data probabilities 167 is equivalent to. In this case, the designation N means that no lane markings are recognized 117 , with the designation D a dashed lane marking 117 and with the designation S a solid road marking is shown. This matrix (the actual type of lane marking is shown in the columns) shows the probability with which an event has occurred, provided that the sensor has made a certain observation - the recognized type of lane marking is shown in the rows. In the following we make a note M. [a; b] the probability that event b has occurred but the sensor has detected event a. M. [D; N], here = 10% - thus indicates that there is a 10 percent probability that the sensor has detected a dashed lane marking, although there is no lane marking 117 was present.

Wobei Pr(leftMark(s_x) I d_I) und Pr(rightMark(s_x) I d_r) direkt aus der Verwechlungsmatrix M abgelesen werden können.Let T = {t ₁ ; ...; t _m } now the set of different types of lane markings, e.g. For example: “solid line” ε T. Let the recognized left and right lane marker types be d _I ε T or d _r ε T for the current iteration. Let S = {s ₁ ; ...; s _n } is the set of tracks in the digital map and leftMark: S → T is the function that specifies the left lane marking type, and rightMark: S → T is the function that specifies the right lane marking type for a lane. Then the probability with which the vehicle is in a lane s _x ε S if only the lane marking type recognition sensor is used can be calculated as the probability that the marking type leftMark (s _x ) is to the left of the vehicle, provided that the vehicle has recognized the marking type d _I , multiplied by the probability that the marking type rightMark (s _x ) is to the right of the vehicle, provided that the vehicle has recognized _{the marking type d r.} Or, expressed as a formula: $$\text{Pr}\left(s_x\right)=\text{Pr}\left(\text{leftMark}\left(s_x\right)|d_l\right)\cdot \text{Pr}\left(\text{rightMark}\left(s_x\right)|d_r\right)$$

Where Pr (leftMark (s _x ) I d _I ) and Pr (rightMark (s _x ) I d _r ) directly from the confusion matrix M. can be read.

die Wahrscheinlichkeit mit der sich das Fahrzeug auf einer Spur s_x ε S aufhält, wenn nur der Spurmarkierungstyperkennungssensor 139 genutzt wird berechnet werden.If it is assumed that the digital map is correct with regard to the number of lanes and the types of lane markings, then with $$\text{Pr}\text{'}\left(s_x\right)=\frac{\text{Pr}\left(s_x\right)}{{\displaystyle {\sum }_{s_y\in s}\text{Pr}\left(s_y\right)}}$$

the probability with which the vehicle is in a lane s _x ε S if only the lane marking type recognition sensor 139 used will be calculated.

wobei n die Anzahl der Spuren auf der aktuell befahrenen Straße notiert. In Iteration i wird die aktuelle Fahrspur s_j entsprechend geschätzt als die Fahrspur s_j, für die pⁱ _j maximal ist.Determining the current lane of the vehicle In this step, Q ^{i is} checked for plausibility with the aid of the lane marking type recognition, in that the probabilities of the location of the lane marking type recognition are weighted with the probabilities of the lane marking type recognition. For this we define P ⁱ = {p ⁱ ₁ ; ...; p ⁱ _n } with $$p_j^i=\frac{q_j^i\cdot \text{Pr}\text{'}\left(s_j\right)}{{\displaystyle {\sum }_{l=1}^n{q}_l^i\cdot \text{Pr}\text{'}\left(s_l\right)}}$$

where n is the number of lanes on the currently driven road. In iteration i, the current lane s _j is estimated accordingly as the lane s _j for which p ⁱ _{j is} a maximum.

As an alternative to using a digital map 129 , from which the lane information (mark types and connectivities as the lane data 127 ) are read, it would also be possible to read this z. B. with a front camera 120 capture. This should be done from the front camera 120 Image data 122 and the corresponding sensors 135 respectively. 139 provide so that this the number of tracks 110 determine the types of lane markings and the connectivity of the lanes to one another.

In one embodiment variant of this invention, the lane marking type recognition sensor is a (video) camera or processes data from such a camera 120 . Possibly a front camera, rear camera, side camera. The lane marker type may be taken from the camera image 122 determined by a neural network. In one embodiment variant of this invention, the lane change detection sensor is a (video) camera or processes data from such a camera 120 . Possibly a front camera, rear camera, side camera. Lane changes may always be recognized when the vehicle drives over a lane marking boundary. In the variant described here, the confusion matrices are fixed. However, it is conceivable that these change depending on the situation. B. the detection will possibly get worse. Depending on the situation, this should then be entered in the confusion matrices M. respectively. W. be reflected. This requires a sensor that detects the external circumstances, e.g. B. light, dark, rain, ice, snow, etc., and confusion matrices M. , W. appropriate to each of these situations. Thus the corresponding matrix can then M. , W. to be selected.

7th shows a diagram for the representation of an information flow of an embodiment of the approach presented here. Thereby lane data 127 for example from a digital map 129 , Lane change data 137 from a lane change sensor 135 and lane boundary type data 141 from a lane boundary type sensor 139 read in and one of the linking unit 133 using a mathematical model linked to probabilities of residence 145 for one of several lanes each.

The approach presented here for an application as a plausibility check method for a feature-based localizer is particularly favorable. Use as an independent locating method is also possible, but this may result in a very high effort compared to a plausibility check - the method works particularly well with a road map that is accurate to the lane.

If an exemplary embodiment comprises an “and / or” link between a first feature and a second feature, this is to be read in such a way that the exemplary embodiment according to one embodiment includes both the first feature and the second feature and, according to a further embodiment, either only the has the first feature or only the second feature.

Claims (12)

A method (200) for determining a probability (145) of a location of a vehicle (100) in one of several lanes (110) of a roadway (115), the method (200) having the following steps: - Reading (210) ◯ Lane data (127), which represent a number of lanes (110) of a lane (115) traveled by the vehicle (100) and / or a possibility to change between the lanes (110), and of lane change data (137), which contain one of a The lane change sensor (135) represent a lane change of the vehicle (100) detected, and from ◯ lane delimitation type data (141), which a type of a lane delimitation (143, 117) detected by a lane delimitation type sensor (137) of the lane (110) currently used by the vehicle (100) ) represent; - Linking (220) the lane data (127), the lane change data (137) and the lane delimitation type data (141) to determine a probability (145) of the vehicle (100) for one of each to determine a plurality of lanes (110) of the roadway (115); and - selecting (230) that lane (110) as the lane (110) in which the vehicle (100) is currently located and which has the highest determined probability of being present (145). Method (200) according to Claim 1 , characterized in that in the reading-in step (210) the lane data (127) are read in from a digital map (129) and / or from an optical sensor (120) and / or the lane change data (137) and / or the lane delimitation type data ( 141) can be read in using an optical sensor (120). The method (200) according to any one of the preceding claims, characterized in that in the step (210) of reading in further advance position data (152 '), advance lane data (127'), advance lane change data (137 ') and / or advance Lane delimitation type data (141 ') are read in, and in the step (220) of linking the advance position data (152'), the advance lane data (127 '), advance lane change data (137') and / or advance lane delimitation type data (141 ') ) are linked to the lane data (127), the lane change data (137) and / or the lane delimitation type data (141) in order to determine the probability of stay (145), the advance position data (152 ') relating to a position of the vehicle (100) represent a previous point in time (i-1) on the roadway (115), the advance lane data (127 ') representing a number of advance lanes (110') of a roadway (115) traveled by the vehicle (100) at a previous point in time ) and an option to switch between the advance driving lanes (110), the advance lane change data (137) representing a lane change of the vehicle (100) detected by the lane change sensor (135) at the previous point in time, and the advance lane delimitation type data (141) representing a type of one of the lane delimitation type sensor (137) the detected lane boundary (143, 117) of the lane (110) used by the vehicle (100) at the previous point in time. Method (200) according to Claim 3 , characterized in that in the step (210) of reading in advance position data are read in, which correspond to a lane (110) which had the highest probability of stay (145) in a previous method step of selection. Method (200) according to one of the preceding claims, characterized in that in step (210) of reading in at least one lane change data probability (161) and / or at least one lane delimitation type data probability (167) and in step (220) of linking the at least one lane change data probability (161) and / or at least one lane delimitation type data probability (167) is used to determine the probability of being present (145) of the vehicle (100), the lane change data probability (161) representing a probability that one of the Lane change sensor (135), the lane change detected has actually taken place and the lane delimitation type data probability (167) represents a probability that a lane delimitation (143, 117) recognized by the lane delimitation sensor (137) corresponded to a type of an actually existing lane delimitation (143, 117) . Method (200) according to Claim 5 , characterized in that in the step (210) of reading in the at least one lane change data probability (161) from a predetermined lane change data probability table (159) and / or the at least one lane delimitation type data probability (167) from a predetermined lane delimitation type data probabilities -Table (165) can be read. Method (200) according to Claim 6 characterized in that in the reading-in step (210) the predetermined lane change data probability table (159) is selected from a plurality of different lane change data probability tables (159) as a function of at least one environmental parameter and / or the predetermined lane delimitation type data Probability table (165) is selected as a function of at least one environmental parameter from a plurality of different lane delimitation type data probability tables (165), in particular wherein the environmental parameter is a lighting situation in an area around the vehicle (100) and / or a road surface condition in the Represents the environment around the vehicle (100). Method (200) according to one of the Claims 5 until 7th , characterized in that in the linking step (220) the at least one lane change data probability (161) is multiplicatively linked with the at least one lane delimitation type data probability (167). Method (200) according to one of the preceding claims, characterized in that the steps (210, 220, 230) of the method (200) are repeated cyclically, in particular are carried out at a frequency of no more than 1 Hertz. Device (105) which is set up to execute and / or control the steps (210, 220, 230) of the method (200) according to one of the preceding claims in corresponding units (131, 133, 150). Computer program which is set up to execute and / or control the steps (210, 220, 230) of the method (200) according to one of the preceding claims. Machine-readable storage medium on which the computer program is based Claim 11 is stored.

Images