DE102021212980B3 - Method for assigning received signals from a sensor system in a vehicle to an object - Google Patents

Abstract

The invention relates to a method for assigning received signals from a sensor system of a vehicle (F) to an object, the sensor system of the vehicle (F) having at least one distance-measuring sensor (S).

Description

The invention relates to a method and a system for assigning received signals from a sensor system in a vehicle to an object in the area surrounding the vehicle.

In principle, it is known to determine the position of objects in the area surrounding a vehicle using a plurality of sensors that only allow distance measurement but not determination of direction, and a tracking method in which received signals are determined in a number of consecutive transmission/reception cycles.

In DE 10 2009 027 936 A1 a method for determining position data of a vehicle driving ahead is described. In this case, the position of the vehicle traveling ahead is calculated using linear regression processing, starting from a primary data set provided by a vehicle sensor system, which consists of distance information and lateral position information of the vehicle traveling ahead.

A known method for tracking received signals in order to be able to assign them to an object uses a forward search window, i.e. directed towards future received signals. The size of the search window is variable and is increasingly being reduced in order to ensure the stability of the tracking process.

The problem with known methods is that they have a low level of robustness in the event of data loss, i.e. missing received signals, or complex object structures, and the tracking method frequently breaks off as a result. Known tracking methods also frequently tend to break off in the case of moving objects to be detected.

Proceeding from this, it is the object of the invention to specify a method for allocating received signals from a sensor system of a vehicle to an object, which is characterized by a high degree of robustness and insensitivity and thus reduces the aborts of the tracking method.

The object is solved by a method having the features of independent patent claim 1 . Preferred embodiments are the subject matter of the dependent claims. A system for allocating received signals from a sensor system in a vehicle to an object is the subject of independent claim 13.

• Zunächst werden mehrere zeitlich aufeinanderfolgende Sende-Empfangszyklen durch die Sensorik durchgeführt. In jedem Sende-Empfangszyklus erfolgt ein Aussenden eines Messsignals durch zumindest einen Sensor und das Empfangen eines reflektierten Anteils dieses Messsignals als Empfangssignal durch einen Sensor. Damit können in den Sende-Empfangszyklen mehrere Empfangssignale erhalten werden, die von einem oder mehreren Objekten stammen.

According to a first aspect, a method for assigning received signals from a sensor system of a vehicle to an object is disclosed. In particular, the method is used to detect a received signal sequence, referred to below as a cluster, and to assign further, newly received received signals to this received signal sequence, the received signal sequence describing an object in the area surrounding the vehicle. The sensor system of the vehicle has at least one distance-measuring sensor, for example at least one ultrasonic sensor. The procedure includes the following steps:

• First of all, several consecutive transmission/reception cycles are carried out by the sensor system. In each transmission-reception cycle, a measurement signal is transmitted by at least one sensor and a reflected portion of this measurement signal is received by a sensor as a reception signal. In this way, a number of received signals originating from one or more objects can be obtained in the send/receive cycles.

At least one cluster of received signals is then formed using a regression method. The regression method is preferably a backward-looking method, i.e. after receiving several received signals, an attempt is made to form a cluster from them. The received signals are assigned to a cluster based on their signal propagation time between the transmission of the measurement signal and the receipt of a reflected portion of this measurement signal, in such a way that those received signals are added to a cluster whose signal propagation times are linear or essentially linear over time, taking into account a change the given variance. In other words, the regression method searches for a group of received signals whose signal propagation time, plotted over time, has a profile that is sufficiently linear, taking into account a predetermined variance.

A tolerance range for the at least one cluster is then determined. The tolerance range defines a signal propagation time range in which a future received signal must lie in order to be able to be assigned to the cluster. The tolerance range can also specify a time interval from the cluster or cluster center within which the future received signal must occur in order to be able to be assigned to the cluster.

After a new received signal has been received, a check is made based on the tolerance range as to whether the new received signal can be assigned to the cluster.

If the check shows that the new received signal can be assigned to the tolerance range, updated cluster information of the cluster to which the received signal can be assigned is calculated based on the regression method, taking into account information on the received signals that are assigned to the cluster. In other words, the cluster is further developed based on the at least one newly assigned received signal. The cluster information by which the cluster is characterized is adjusted taking into account the received signals previously assigned to the cluster and the at least one newly assigned received signal.

The technical advantage of the method according to the invention is that it is extremely robust and insensitive, even with moving or contoured objects, and the risk of object tracking being aborted is therefore reduced.

According to one exemplary embodiment, the cluster, i.e. the initial cluster, is formed by determining an approximation line and assigning received signals to the cluster whose signal propagation times are in a predetermined area around the approximation line. In particular, an attempt is first made to find a group of received signals whose points lie approximately on a straight line, i.e. allowing for a predetermined variance. An approximation straight line is then preferably calculated on the basis of these points, for example by means of a method of the least error squares (least mean square method). The group of received signals that were used for calculating the approximation line is then expanded, namely by those points that lie in the specified surrounding area around the approximation line. The cluster is then formed on the basis of this extended group. This means that objects that have multiple reflection planes at different distances from the sensor (so that multiple received signals that are offset in time are generated from one transmitted signal) can be tracked better.

According to one exemplary embodiment, before the cluster is formed, the range in which received signals can lie is expanded by using further received signals to form the cluster, the signal propagation times of which lie in a predetermined surrounding range around the approximation line. This in turn makes the method more robust in relation to objects with a plurality of reflection planes.

wobei gilt:

• E(X): Mittelwert der Sendezeitpunkte der Empfangssignale des Clusters;
• E(Y): Mittelwert der Signallaufzeiten der Empfangssignale des Clusters;
• m: Steigung der Approximationsgeraden
• x_neu: Sendezeitpunkt des zuzuordnenden Empfangssignals;
• y_neu: Signallaufzeit des zuzuordnenden Empfangssignals;
• c: Abstandsgrenzwert.

According to one embodiment, those received signals are assigned to the cluster that meet the following condition: $$\left|E\left(Y\right)+m\cdot \left(x_{ne\mathrm{and}}-E\left(X\right)\right)-y_{ne\mathrm{and}}\right|<c$$

where:

• E(X): average of the transmission times of the received signals of the cluster;
• E(Y): average of the signal propagation times of the received signals of the cluster;
• m: Slope of the approximation line
• x _new : time of transmission of the received signal to be assigned;
• y _new : signal propagation time of the received signal to be assigned;
• c: distance limit.

According to one embodiment, the cluster information includes a cluster center and a cluster slope. The cluster center indicates an average of the transmission times and the signal propagation times of the received signals that are assigned to the cluster. The mean can be a linear mean or be a weighted average, with the more recent received signals in particular being weighted more highly than received signals from a longer time ago. The cluster gradient is defined, for example, by the ratio of the variance of the signal propagation times and the variance of the transmission times or the ratio of the covariance between the transmission times and the signal propagation times and the variance of the transmission times. When calculating the cluster gradient, the more recent received signals can also be weighted more highly than received signals that are longer in the past.

wobei gilt:

• p_i: Gewichtungsfaktor, der vorzugsweise 1/A beträgt, wobei A die Anzahl der Empfangssignale im Cluster ist;
• E(X): Mittelwert der Sendezeitpunkte der Empfangssignale des Clusters;
• E(Y): Mittelwert der Signallaufzeiten der Empfangssignale des Clusters;
• x_i: Sendezeitpunkt des i-ten Empfangssignals;
• y_i: Signallaufzeit des i-ten Empfangssignals;

According to one embodiment, the cluster slope is determined by the following formula: $$CS=\frac{COv\left(X,Y\right)}{Va\mathrm{right}\left(X\right)}=\frac{{\displaystyle {\sum }_i^A{p}_i\cdot \left(x_i-E\left(X\right)\right)\cdot \left(y_i-E\left(Y\right)\right)}}{{\displaystyle {\sum }_i^A{p}_i\cdot {\left(x_i-E\left(X\right)\right)}^2}}.$$

where:

• p _i : weighting factor, which is preferably 1/A, where A is the number of received signals in the cluster;
• E(X): average of the transmission times of the received signals of the cluster;
• E(Y): average of the signal propagation times of the received signals of the cluster;
• x _i : time of transmission of the i-th received signal;
• y _i : signal propagation time of the i-th received signal;

According to one embodiment, the tolerance range is determined based on the cluster center and the cluster slope. In particular, starting from the cluster center and based on the cluster gradient, a range of signal propagation times is defined in which the signal propagation time of the received signal may lie so that the received signal can be assigned to the cluster. The tolerance range also defines the time interval from the cluster center at which the transmission signal that leads to the reception signal must have been sent and/or the reception signal must be received so that it can be assigned to the cluster.

wobei gilt:

• CM_x: Mittelwert der Sendezeitpunkte der Empfangssignale des Clusters bzw. erste Koordinate des Clustermittelpunkts;
• CM_y: Mittelwert der Signallaufzeiten der Empfangssignale des Clusters bzw. zweite Koordinate des Clustermittelpunkts;
• CS: Clustersteigung;
• x_neu: Sendezeitpunkt des neu ermittelten Empfangssignals;
• y_neu: Signallaufzeit des neu ermittelten Empfangssignals;
• c: Abstandsgrenzwert.

According to one embodiment, newly determined received signals are associated with an existing cluster if they meet the following condition: $$\left|C{M}_y+CS\cdot \left(x_{ne\mathrm{and}}-C{M}_x\right)-y_{ne\mathrm{and}}\right|<c$$

where:

• CM _x : average of the transmission times of the received signals of the cluster or first coordinate of the cluster center;
• CM _y : mean value of the signal propagation times of the received signals of the cluster or second coordinate of the center of the cluster;
• CS: cluster slope;
• x _new : time of transmission of the newly determined received signal;
• y _new : signal propagation time of the newly determined received signal;
• c: distance limit.

According to one embodiment, after associating one or more newly determined received signals, the values of the cluster center and the cluster slope are updated. This ensures that after one or more received signals have been assigned to the cluster, the cluster is further developed, so that the cluster center point and the cluster gradient are adapted to the course of the received signals in the cluster.

According to one embodiment, the received signals are weighted when calculating the updated values of the cluster center point and the cluster slope, in such a way that received signals determined longer in the past are weighted less than received signals that are more recent in the past. This allows them to characterize the cluster Cluster information follows changing reception signal profiles in an improved manner, which result, for example, from a relative movement between the host vehicle and a surrounding object.

According to one embodiment, the sensor determines Doppler information about the received signal. The Doppler information is used to check whether the received signal can be assigned to a cluster. Improved tracking of objects is thus possible when there is a relative movement between the host vehicle and a surrounding object.

According to one exemplary embodiment, when checking whether the received signal can be assigned to a cluster, the Doppler information is compared with the cluster slope. Since the cluster slope is an indication of the change in the signal propagation time over time and thus of the relative speed between the host vehicle and a surrounding object, comparing the Doppler information with the cluster slope can reduce the probability of an incorrect assignment of a received signal to a cluster.

• - Durchführen mehrerer zeitlich aufeinanderfolgender Sende-Empfangszyklen durch die Sensorik, wobei die Sende-Empfangszyklen jeweils das Aussenden eines Messsignals durch zumindest einen Sensor und das Empfangen eines reflektierten Anteils dieses Messsignals als Empfangssignal durch einen Sensor umfassen;
• - Bilden zumindest eines Clusters von Empfangssignalen mittels eines Regressionsverfahrens, wobei die Empfangssignale basierend auf deren Signallaufzeit zwischen dem Aussenden des Messsignals und dem Empfangen eines reflektierten Anteils dieses Messsignals einem Cluster zugeordnet werden, und zwar derart, dass diejenigen Empfangssignale einem Cluster hinzugefügt werden, deren Signallaufzeiten sich über der Zeit linear oder im Wesentlichen linear unter Berücksichtigung einer vorgegebenen Varianz ändern;
• - Bestimmen eines Toleranzbereichs für das zumindest eine Cluster, wobei der Toleranzbereich einen Signallaufzeitbereich festlegt, in dem ein zukünftiges Empfangssignal liegen muss, um zu dem Cluster zugeordnet werden zu können;
• - Empfangen eines neuen Empfangssignals und prüfen basierend auf dem Toleranzbereich, ob das neue Empfangssignal dem Cluster zugeordnet werden kann; und
• - Berechnen von aktualisierten Clusterinformationen des Clusters, dem das Empfangssignal zugeordnet werden kann, basierend auf dem Regressionsverfahren unter Berücksichtigung von Informationen der Empfangssignale, die dem Cluster zugeordnet sind.

According to a further aspect, a system for assigning received signals from a sensor system of a vehicle to an object is disclosed. The system comprises at least one distance-measuring sensor and a computer unit for controlling the sensor system and providing received signals. The computing unit is configured to perform the following steps:

• - Carrying out a plurality of chronologically consecutive transmission/reception cycles by the sensor system, the transmission/reception cycles each comprising the transmission of a measurement signal by at least one sensor and the reception of a reflected portion of this measurement signal as a reception signal by a sensor;
• - Forming at least one cluster of received signals by means of a regression method, the received signals being assigned to a cluster based on their signal propagation time between the transmission of the measurement signal and the reception of a reflected portion of this measurement signal, in such a way that those received signals are added to a cluster whose signal propagation times change linearly or substantially linearly over time given a predetermined variance;
• - Determining a tolerance range for the at least one cluster, the tolerance range defining a signal propagation time range in which a future received signal must lie in order to be able to be assigned to the cluster;
• - Receiving a new received signal and checking based on the tolerance range whether the new received signal can be assigned to the cluster; and
• - calculating updated cluster information of the cluster to which the received signal can be assigned based on the regression method taking into account information of the received signals assigned to the cluster.

The terms “approximately”, “substantially” or “roughly” mean deviations from the exact value by +/-10%, preferably by +/-5% and/or deviations in the form of changes that are insignificant for the function .

Further developments, advantages and possible applications of the invention also result from the following description of exemplary embodiments and from the figures. All of the features described and/or illustrated are fundamentally the subject matter of the invention, either alone or in any combination, regardless of how they are summarized in the claims or how they relate back to them. The content of the claims is also made part of the description.

• 1 beispielhaft eine Draufsichtdarstellung eines Fahrzeugs mit einem Umgebungserfassungssystem, das mehrere Sensoren aufweist;
• 2 beispielhaft ein Diagramm, das die Signallaufzeit einer Vielzahl von Empfangssignalen aufgetragen über dem Sendezeitpunkt veranschaulicht;
• 3 beispielhaft und schematisch ein Ausschnitt des Diagramms gemäß 2, wobei mehrere Pfade zwischen benachbarten Empfangssignalen eingezeichnet sind, die das zeitlich rückwärtsgewandte Verfahren zur Suche einer Approximationsgeraden verdeutlicht;
• 4 beispielhaft und schematisch den Ausschnitt des Diagramms gemäß 3 mit einer Approximationsgerade, einem Umgebungsbereich um die Approximationsgerade und einer Ellipse, die Clusterinformationen eines initialen Clusters verdeutlicht;
• 5 beispielhaft und schematisch ein Ausschnitt des Diagramms gemäß 2, mit dem initialen Cluster, das durch die linke Ellipse veranschaulicht wird, weiteren Ellipsen, die jeweils Informationen von neu mit dem Cluster zu assoziierenden Empfangssignalen veranschaulichen und Toleranzbereichen zu den jeweiligen Ellipsen, die den Bereich veranschaulichen, in dem neue Empfangssignale liegen müssen, um mit dem Cluster assoziiert werden zu können;
• 6 beispielhaft und schematisch das Diagramm gemäß 2, mit einer Vielzahl von Ellipsen, die das fortlaufende Tracking der Empfangssignale zur Zuordnung der Empfangssignale zu einem Objekt verdeutlichen; und
• 7 beispielhaft ein Blockdiagramm, das die Verfahrensabläufe des Verfahrens zur Zuordnung von Empfangssignalen zu einem Objekt verdeutlicht.

The invention is explained in more detail below with reference to the figures of exemplary embodiments. Show it:

• 1 an exemplary top view illustration of a vehicle with an environment detection system that has a plurality of sensors;
• 2 by way of example a diagram which illustrates the signal propagation time of a large number of received signals plotted against the time of transmission;
• 3 an excerpt of the diagram according to FIG 2 , where several paths between adjacent received signals are drawn, which elucidates the backward-looking method for searching for an approximation line;
• 4 as an example and schematically the section of the diagram according to 3 with an approximation line, a surrounding area around the approximation line and an ellipse which illustrates cluster information of an initial cluster;
• 5 an excerpt of the diagram according to FIG 2 , with the initial cluster, which is illustrated by the ellipse on the left, further ellipses, each of which illustrate information on new received signals to be associated with the cluster, and tolerance ranges for the respective ellipses, which illustrate the range in which new received signals must lie in order to to be associated with the cluster;
• 6 as an example and schematically the diagram according to 2 , with a large number of ellipses, which illustrate the continuous tracking of the received signals for assigning the received signals to an object; and
• 7 a block diagram by way of example that illustrates the process sequences of the method for assigning received signals to an object.

1 shows a vehicle F, which has a large number of sensors S, by way of example and in a roughly schematic manner. These are in 1 indicated as circles. The vehicle F preferably has a plurality of sensors S which are distributed around the vehicle F. The sensors S are distance-measuring sensors, for example ultrasonic sensors. Alternatively, the sensors S can also be radar sensors. However, the sensors S preferably do not have the ability to determine the direction from which a received, reflected signal component of the transmission signal originates. Such sensors are often referred to as 1-D sensors. The distance of an object at which the reflection occurs can be determined based on the propagation time of the sensor signal between the time of transmission and the time of reception.

The sensors are coupled to a computer unit R, which has at least one processor and at least one memory unit. This computer unit R is designed to carry out the method sequences disclosed in this document, in particular to group and time-track the received signals that are caused by reflections on an object in order to be able to assign them to an object.

2 shows a diagram in which received signals (indicated by the respective points) of a sensor system of a vehicle F that is moving past a post-like object are entered. The horizontal x-axis indicates the transmission times at which measurement signals were emitted by the sensors. The signal propagation time between the transmission of the measurement signal and the reception of a reflected portion of the measurement signal is plotted on the vertical y-axis. The signal propagation time indicates the distance of the object from the sensor S.

The object is contoured, for example, ie it has a number of object planes with different geometric dimensions, for example different widths. As a result, as in 2 recognizable, several received signals with different signal propagation times result from one transmission cycle, i.e. several received signals are assigned to a transmission time that represents a transmission cycle (in 2 received signals lying vertically one above the other).

When approaching the object, the distance is first reduced, so that the signal propagation time decreases, then reaches a minimum value and increases again after driving past the object.

In order to be able to reliably detect and localize the object, after the detection of several received signals, an attempt is made to find a signal sequence in the received signals already detected which is approximately on a straight line, i.e. taking into account a variance of the received signals. This straight line is referred to below as the approximation line. The received signals that are on or near such an approximation line are then assigned to a cluster C. This cluster C indicates that the received signals assigned to it can be assigned to an object.

3 shows, by way of example and schematically, the determination of an approximation straight line, taking into account a number of received signals that have already been received. In other words, based on recent (for example in the last 100ms up to 500ms) detected received signals checked whether at least some of these received signals are on an approximation line taking into account a permissible variance or spread.

As indicated by the dashed lines, multiple paths can exist between the received signals. To form the approximation straight line, the path that has the greatest linearity, i.e. that comes closest to a straight line, is preferably used. Based on this path, an approximation line is then determined using a regression method, in particular a linear regression method.

4 shows the in 3 contained received signals with the closest path representing a straight line as a dashed line and an approximation straight line as a solid line, which was determined using a regression method, in particular a linear regression method, based on the received signals that lie on the path that has the greatest linearity.

Based on the approximation line, received signals from a surrounding area U around the approximation line are preferably also used to form a cluster C. The surrounding area U is in 4 indicated by the pair of dashed lines parallel to the approximation line. This means that received signals that originate from the same object but have different signal propagation times due to the shape of the object and therefore deviate from the approximation line are assigned to the cluster.

Dabei ist:

• E(X): Mittelwert der Sendezeitpunkte der Empfangssignale des Pfades, der zur Berechnung der Approximationsgerade verwendet wurde;
• E(Y): Mittelwert der Signallaufzeiten der Empfangssignale des Pfades, der zur Berechnung der Approximationsgerade verwendet wurde;
• m: Steigung der Approximationsgeraden
• x_neu: Sendezeitpunkt eines potentiell zuzuordnenden Empfangssignals;
• y_neu: Signallaufzeit eines potentiell zuzuordnenden Empfangssignals;
• c: Abstandsgrenzwert.

The surrounding area U around the approximation line can be defined, for example, by the following relationship: $$\left|E\left(Y\right)+m\cdot \left(x_{ne\mathrm{and}}-E\left(X\right)\right)-y_{ne\mathrm{and}}\right|<c$$

where:

• E(X): Mean value of the transmission times of the received signals of the path that was used to calculate the approximation line;
• E(Y): Mean value of the signal propagation times of the received signals of the path that was used to calculate the approximation line;
• m: Slope of the approximation line
• x _new : time of transmission of a received signal that can potentially be assigned;
• y _new : signal propagation time of a received signal that can potentially be assigned;
• c: distance limit.

Thus, for the initial formation of a cluster C, those received signals are used whose transmission times or signal propagation times are in a surrounding area defined by the distance limit value c to the received signals that were used for the calculation of the approximation line.

An initial cluster C is then formed based on the received signals located in the surrounding area U. The initial cluster C is in 4 indicated by the ellipse.

The initial cluster C has cluster information, in particular a cluster center CM and a cluster slope CS. These can be calculated, for example, using a least squares method (least mean square method). In the calculation of the cluster information, all received signals that are to be assigned to the cluster, i.e. that are in particular in the surrounding area U, are preferably weighted equally.

The cluster information can be calculated as follows:

wobei:

• A: Anzahl der Empfangssignale im Cluster; und
• x_i: Sendezeitpunkt des i-ten Empfangssignal

ist.The mean value of the transmission time E(X), ie the mean value over time of the cluster C, is determined by the time distribution (in the 3 and 4 Distribution of the received signals in the horizontal direction) of the received signals in cluster C determined. For example, this is calculated as follows: $$E\left(X\right)=\frac{1}{A}{\displaystyle \sum _{i=1}^A{x}_i}$$

whereby:

• A: number of received signals in the cluster; and
• x _i : time of transmission of the i-th received signal

is.

wobei:

• A: Anzahl der Empfangssignale im Cluster; und
• y_i: Signallaufzeit des i-ten Empfangssignal

ist.The mean value of the signal propagation time E(Y) is determined by the distribution of the signal propagation time (in the 3 and 4 Distribution of the received signals in the vertical direction) of the received signals in cluster C determined. For example, this is calculated as follows: $$E\left(Y\right)=\frac{1}{A}{\displaystyle \sum _{i=1}^A{y}_i}$$

whereby:

• A: number of received signals in the cluster; and
• y _i : signal propagation time of the i-th received signal

is.

In addition, for the received signals assigned to cluster C, the variance of the transmission times (in 3 and 4 the variance in the horizontal direction, ie in the x-direction) or the variance of the signal propagation times (in 3 and 4 the variance in the vertical direction, ie in the y-direction) can be calculated.

wobei

• E(X): Mittelwert der Sendezeitpunkte der Empfangssignale des Clusters;
• A: Anzahl der Empfangssignale im Cluster; und
• x_i: Sendezeitpunkt des i-ten Empfangssignal

$$Var\left(Y\right)=\frac{1}{A}{\displaystyle \sum _{i=1}^A{\left(y_i-E\left(Y\right)\right)}^2}$$

wobei

• E(Y): Mittelwert der Signallaufzeit der Empfangssignale des Clusters;
• A: Anzahl der Empfangssignale im Cluster; und
• y_i: Signallaufzeit des i-ten Empfangssignals.

The following applies: $$Va\mathrm{right}\left(X\right)=\frac{1}{A}{\displaystyle \sum _{i=1}^A{\left(x_i-E\left(X\right)\right)}^2}$$

whereby

• E(X): average of the transmission times of the received signals of the cluster;
• A: number of received signals in the cluster; and
• x _i : time of transmission of the i-th received signal

$$Va\mathrm{right}\left(Y\right)=\frac{1}{A}{\displaystyle \sum _{i=1}^A{\left(y_i-E\left(Y\right)\right)}^2}$$

whereby

• E(Y): Mean value of the signal propagation time of the received signals of the cluster;
• A: number of received signals in the cluster; and
• y _i : signal propagation time of the i-th received signal.

The covariance between the time of transmission and the signal propagation time can be determined as follows: $$COv\left(X,Y\right)=\frac{1}{A}{\displaystyle \sum _{i=1}^A\left(x_i-E\left(X\right)\right)\left(y_i-E\left(Y\right)\right)}$$

The cluster gradient CS can be determined from the variance Var(X) and the covariance Cov (X,Y).

The following applies: $$CS=\frac{COv\left(X,Y\right)}{Va\mathrm{right}\left(X\right)}.$$

The cluster center CM is determined by the mean E(X) of the transmission times of the received signals of cluster C and the mean E(Y) of the signal propagation times of the received signals of cluster C, with E(X) and E(Y) specifying the coordinates of the cluster center CM. The cluster gradient CS indicates the change in the signal propagation time over time and thus indicates the signal propagation time range in which a future received signal will lie if it originates from the same object.

If new received signals are received after the initialization of a cluster C, it is checked whether the new received signal can be assigned to cluster C, i.e. it is checked in particular whether the received signal is within a tolerance range T, in which a received signal to be associated with the cluster expect is. The tolerance range T is in 5 sketched by the sector opening to the right from the cluster center CM. For example, it is checked whether the received signal has a signal propagation time that corresponds to the course of the signal propagation time that is indicated by the cluster gradient CS.

wobei gilt:

• CM_x: Mittelwert der Sendezeitpunkte der Empfangssignale des Clusters bzw. erste Koordinate des Clustermittelpunkts;
• CM_y: Mittelwert der Signallaufzeiten der Empfangssignale des Clusters bzw. zweite Koordinate des Clustermittelpunkts;
• CS: Clustersteigung;
• x_neu: Sendezeitpunkt des neu ermittelten Empfangssignals;
• y_neu: Signallaufzeit des neu ermittelten Empfangssignals;
• c: Abstandsgrenzwert.

To check whether the new received signal can be associated with cluster C, the following formula can be used, for example: $$\left|C{M}_y+CS\cdot \left(x_{ne\mathrm{and}}-C{M}_x\right)-y_{ne\mathrm{and}}\right|<c$$

where:

• CM _x : average of the transmission times of the received signals of the cluster or first coordinate of the cluster center;
• CM _y : mean value of the signal propagation times of the received signals of the cluster or second coordinate of the center of the cluster;
• CS: cluster slope;
• x _new : time of transmission of the newly determined received signal;
• y _new : signal propagation time of the newly determined received signal;
• c: distance limit.

If the new received signal is within the tolerance range, in particular if the above condition is met, the newly determined received signal is associated with the cluster and the cluster information, in particular the cluster center point CM and the cluster slope CS, is updated using the newly determined received signal.

The cluster information is preferably updated by a time-dependent weighting of the received signals. In this case, the received signals that are longer in the past are preferably weighted less than the currently received received signal. In particular, the weighting factor can be reduced linearly or non-linearly with the time interval. The result of this is that received signals that are longer in the past are weighted increasingly less, so that a non-linear change in the cluster information becomes possible and the cluster C can thus adapt to non-linear received signal profiles.

The cluster information is preferably updated in that the previously applicable cluster information, in particular the previously applicable mean value of the transmission times of the received signals, the previously applicable mean value of the signal propagation time of the received signals, the previously applicable variance of the transmission times, the previously applicable variance of the signal propagation times and the previously applicable covariance be provided with a first weighting factor between the time of transmission and the signal propagation time in order to obtain weighted, previously valid cluster information. This weighted, previously valid cluster information is modified based on one or more received signals to be newly associated in order to thereby obtain updated cluster information. As a result, the cluster information can be updated in a way that saves storage resources, since the information about all of the received signals does not have to be stored individually, but rather only the cluster information that was previously valid.

• Zunächst werden basierend auf dem zumindest einen neu zu assoziierenden Empfangssignal die Mittelwerte E(X) und E(Y) des Clusters C aktualisiert. Dabei werden die bisher geltenden Mittelwerte E_alt(X) und E_alt(Y), die mit einem Gewichtungsfaktor gewichtet werden, durch Addition von Mittelwerten E_neu(X) und E_neu(Y) des zumindest einen neu zu assoziierenden Empfangssignals modifiziert. Dabei werden die Mittelwerte E_neu(X) und E_neu(Y) ebenso gewichtet. Die Gewichtungsfaktoren können entweder fest gewählt oder adaptiv sein. Durch die adaptive Wahl der Gewichtungsfaktoren kann erreicht werden, dass der Einfluss des zumindest einen neu zu assoziierenden Empfangssignals fahrsituationsabhängig oder in Abhängigkeit der zu erfassenden Umgebung veränderbar ist. So kann beispielsweise in Situationen, in denen konturierte Objekte erfasst werden und/oder in denen eine Relativbewegung zwischen dem Fahrzeug und dem Objekt stattfindet, der Einfluss des zumindest einen neu zu assoziierenden Empfangssignals erhöht werden, damit das Objekt sicher getrackt und damit erfasst werden kann. In umgekehrten Fall kann in Situationen, in denen wenig konturierte Objekte erfasst werden und/oder in denen sich das Fahrzeug nicht bewegt, der Einfluss des zumindest einen neu zu assoziierenden Empfangssignals reduziert werden, so dass der Einfluss von Rauschen reduziert wird.

The cluster information is preferably updated as follows:

• First, the mean values E(X) and E(Y) of the cluster C are updated based on the at least one received signal to be newly associated. The previously applicable mean values E _old (X) and E _old (Y), which are weighted with a weighting factor, are modified by adding mean values E _new (X) and E _new (Y) of the at least one new received signal to be associated. The mean values E _new (X) and E _new (Y) are also weighted. The weighting factors can either be fixed or adaptive. Through the adaptive selection of the weighting factors, it can be achieved that the influence of the at least one received signal to be newly associated can be changed as a function of the driving situation or as a function of the environment to be detected. For example, in situations in which contoured objects are detected and/or in which there is a relative movement between the vehicle and the object, the influence of the at least one received signal to be newly associated can be increased so that the object can be reliably tracked and thus detected. Conversely, in situations in which objects with few contours are detected and/or in which the vehicle is not moving, the influence of the at least one received signal to be newly associated can be reduced, so that the influence of noise is reduced.

$$E\left(Y\right)=E_{alt}\left(Y\right)\cdot \left(1-w\right)+w\cdot \frac{1}{B}{\displaystyle \sum _{i=1}^B{y}_i}$$

wobei:

• B: Anzahl der neu zu assoziierenden Empfangssignale;
• E_alt(X): bisher geltender Mittelwert der Sendezeitpunkte;
• E_alt(Y): bisher geltender Mittelwert der Signallaufzeiten;
• w: Gewichtungsfaktor
• x_i: Sendezeitpunkt des i-ten neu zu assoziierenden Empfangssignals;
• y_i: Signallaufzeit des i-ten neu zu assoziierenden Empfangssignals;

The updating of the mean values E(X), E(Y) can be achieved using the following formulas: $$E\left(X\right)=E_{alt}\left(X\right)\cdot \left(1-w\right)+w\cdot \frac{1}{B}{\displaystyle \sum _{i=1}^B{x}_i}$$

$$E\left(Y\right)=E_{alt}\left(Y\right)\cdot \left(1-w\right)+w\cdot \frac{1}{B}{\displaystyle \sum _{i=1}^B{y}_i}$$

whereby:

• B: number of received signals to be newly associated;
• E _old (X): previously applicable average of the transmission times;
• E _old (Y): previously applicable mean value of the signal propagation times;
• w: weighting factor
• x _i : time of transmission of the i-th received signal to be newly associated;
• y _i : signal propagation time of the i-th received signal to be newly associated;

$$Var\left(Y\right)=Va{r}_{alt}\left(Y\right)\cdot \left(1-w\right)+w\cdot \frac{1}{B}{\displaystyle \sum _{i=1}^B{\left(y_i-E\left(Y\right)\right)}^2}$$

$$Cov\left(X,Y\right)=Co{v}_{alt}\left(X,Y\right)\cdot \left(1-w\right)+w\cdot \frac{1}{B}{\displaystyle \sum _{i=1}^B\left(x_i-E\left(X\right)\right)\left(y_i-E\left(Y\right)\right)}$$

wobei:

• B: Anzahl der neu zu assoziierenden Empfangssignale;
• Cov_alt(X,Y): bisher geltende Kovarianz des Clusters;
• E(X): aktualisierter Mittelwert der Sendezeitpunkte;
• E_alt(Y): aktualisierter Mittelwert der Signallaufzeiten;
• Var_alt(X): bisher geltende Varianz der Sendezeitpunkte des Clusters;
• Var_alt(Y): bisher geltende Varianz der Signallaufzeiten des Clusters;
• w: Gewichtungsfaktor
• x_i: Sendezeitpunkt des i-ten neu zu assoziierenden Empfangssignals;
• y_i: Signallaufzeit des i-ten neu zu assoziierenden Empfangssignals;

The variances Var(X) and Var(Y) and the covariance Cov(X,Y) of the cluster can be updated as follows: $$Va\mathrm{right}\left(X\right)=Va{\mathrm{right}}_{alt}\left(X\right)\cdot \left(1-w\right)+w\cdot \frac{1}{B}{\displaystyle \sum _{i=1}^B{\left(x_i-E\left(X\right)\right)}^2}$$

$$Va\mathrm{right}\left(Y\right)=Va{\mathrm{right}}_{alt}\left(Y\right)\cdot \left(1-w\right)+w\cdot \frac{1}{B}{\displaystyle \sum _{i=1}^B{\left(y_i-E\left(Y\right)\right)}^2}$$

$$COv\left(X,Y\right)=CO{v}_{alt}\left(X,Y\right)\cdot \left(1-w\right)+w\cdot \frac{1}{B}{\displaystyle \sum _{i=1}^B\left(x_i-E\left(X\right)\right)\left(y_i-E\left(Y\right)\right)}$$

whereby:

• B: number of received signals to be newly associated;
• Cov _old (X,Y): previously valid covariance of the cluster;
• E(X): updated mean of transmission times;
• E _old (Y): updated average of the signal propagation times;
• Var _old (X): previously applicable variance of the sending times of the cluster;
• Var _old (Y): previously applicable variance of the signal propagation times of the cluster;
• w: weighting factor
• x _i : time of transmission of the i-th received signal to be newly associated;
• y _i : signal propagation time of the i-th received signal to be newly associated;

The weighting factor w is preferably selected in the range 0<w<0.5, in particular in the range 0.1<w<0.3.

6 shows the diagram of 2 with ellipses inscribed therein. The individual ellipses illustrate the mean values in the x and y directions or the variances Var(X) and Var(Y) or the covariances Cov(X,Y) of the received signals to be newly associated with the cluster.

In one embodiment, the sensor system can capture and output Doppler information, i.e. information on the Doppler shift that results from a relative movement of the vehicle and the detected object. This Doppler information can be used to decide whether to associate one or more new received signals with the cluster. In particular, the cluster gradient can be compared with the Doppler information in order to decide whether the newly acquired received signal should be assigned to the cluster. This can reduce false associations.

In one embodiment, the formation of random clusters can be reduced by using received signals that can be assigned to a cluster to remove other received signals that have arisen due to the same transmission of a transmitted signal in the same and/or other sensors. The determination that the received signals originate from the same transmission of the transmitted signal can be made, for example, on the basis of a time coding of the transmitted signals or a frequency coding of the transmitted signals.

In one embodiment, multiple reflections between the vehicle and the object to be detected can be used to confirm the existence of an object or a cluster. If the vehicle approaches an object, multiple reflections of level n can occur. These have a speed n times higher than in the case of single reflection. The traceability of a speed component is limited. However, the information from a detected cluster can be used explicitly to specifically identify a pseudo-cluster with a speed component that is approximately n times as large and at a distance of approximately n times. This reduces noise in other sensors and confirms the existence of the actual cluster.

7 shows a diagram that explains the method steps for assigning received signals to an object.

First, the sensor system performs a number of consecutive transmission/reception cycles (S10), each transmission/reception cycle including the transmission of a measurement signal by at least one sensor and the receipt of a reflected portion of this measurement signal as a reception signal by a sensor. At least one cluster of received signals is then formed using a regression method, the received signals being assigned to a cluster based on their signal propagation time between the transmission of the measurement signal and the receipt of a reflected portion of this measurement signal, in such a way that those received signals are added to a cluster whose Signal propagation times change linearly or essentially linearly over time, taking into account a predetermined variance (S11).

A tolerance range is then determined for the at least one cluster, with the tolerance range specifying a signal propagation time range in which a future received signal must lie in order to be able to be assigned to the cluster (S12).

After receiving a new received signal, it is checked based on the tolerance range whether the new received signal can be assigned to the cluster (S13).

Lastly, updated cluster information of the cluster to which the reception signal can be assigned is calculated based on the regression method considering information of the reception signals assigned to the cluster (S14).

The invention has been described above using exemplary embodiments. It goes without saying that numerous changes and modifications are possible without leaving the scope of protection defined by the patent claims.

Reference List

C

clusters

CM

cluster center

CS

cluster slope

f

vehicle

R

computing unit

S

sensor

T

tolerance range

u

surrounding area

Claims (13)

Method for assigning received signals from a sensor system of a vehicle (F) to an object, the sensor system of the vehicle (F) having at least one distance-measuring sensor (S), the method comprising the following steps: - The sensor system (S10) carries out several consecutive transmission/reception cycles, the transmission/reception cycles each comprising the transmission of a measurement signal by at least one sensor (S) and the reception of a reflected portion of this measurement signal as a reception signal by a sensor (S); - Forming at least one cluster (C) of received signals by means of a regression method, the received signals being assigned to a cluster (C) based on their signal propagation time between the transmission of the measurement signal and the receipt of a reflected portion of this measurement signal, in such a way that those received signals Clusters (C) are added whose signal propagation times change linearly or essentially linearly over time, taking into account a predetermined variance (S11); - Determining a tolerance range (T) for the at least one cluster (C), the tolerance range (T) defining a signal propagation time range in which a future received signal must lie in order to be able to be assigned to the cluster (C) (S12); - Receiving a new received signal and checking based on the tolerance range (T) whether the new received signal can be assigned to the cluster (C) (S13); and - calculating updated cluster information of the cluster (C) to which the received signal can be assigned based on the regression method taking into account information of the received signals assigned to the cluster (C) (S14). procedure after claim 1 , characterized in that the cluster (C) is formed by determining an approximation line and assigning received signals to the cluster whose signal propagation times are in a predetermined surrounding area (U) around the approximation line. procedure after claim 2 , characterized in that before the cluster (C) is formed, the range in which the received signals can lie is expanded by further received signals being used to form the cluster (C), the signal propagation times of which are in a predetermined surrounding area (U) around the approximation line lay. Method according to one of the preceding claims, characterized in that those received signals are assigned to the cluster (C) which meet the following condition: $$\left|E\left(Y\right)+m\cdot \left(x_{ne\mathrm{and}}-E\left(X\right)\right)-y_{ne\mathrm{and}}\right|<c$$

where: E(X): mean value of the transmission times of the received signals of the cluster; E(Y): average of the signal propagation times of the received signals of the cluster; m: slope of the approximation line x _new : transmission time of the received signal to be assigned; y _new : signal propagation time of the received signal to be assigned; c: distance limit. Method according to one of the preceding claims, characterized in that the cluster information comprises a cluster center (CM) and a cluster slope (CS). Method according to one of the preceding claims, characterized in that a cluster center (CM) and a cluster slope (CS) are determined for the cluster (C), the cluster center (CM) being determined by averaging the transmission times and averaging the signal propagation times of the cluster (C ) associated received signals is calculated and the cluster slope (CS) is determined by the following formula: $$CS=\frac{COv\left(X,Y\right)}{Va\mathrm{right}\left(X\right)}=\frac{{\displaystyle {\sum }_i^A{p}_i\cdot \left(x_i-E\left(X\right)\right)\cdot \left(y_i-E\left(Y\right)\right)}}{{\displaystyle {\sum }_i^A{p}_i\cdot {\left(x_i-E\left(X\right)\right)}^2}}.$$

where: p _i : weighting factor, which is preferably 1/A, where A is the number of received signals in the cluster; E(X): average of the transmission times of the received signals of the cluster; E(Y): average of the signal propagation times of the received signals of the cluster; x _i : time of transmission of the i-th received signal; y _i : signal propagation time of the i-th received signal; procedure after claim 6 , characterized in that the tolerance range is determined based on the cluster center (CM) and the cluster slope (CS). procedure after claim 7 , characterized in that newly determined received signals are associated with an existing cluster (C) if they meet the following condition: $$\left|C{M}_y+CS\cdot \left(x_{ne\mathrm{and}}-C{M}_x\right)-y_{ne\mathrm{and}}\right|<c$$

where: CM _x : mean value of the transmission times of the received signals of the cluster or first coordinate of the cluster center; CM _y : mean value of the signal propagation times of the received signals of the cluster or second coordinate of the center of the cluster; CS: cluster slope; x _new : time of transmission of the newly determined received signal; y _new : signal propagation time of the newly determined received signal; c: distance limit. Procedure according to one of Claims 6 until 8th , characterized in that after the association of one or more newly determined received signals, the values of the cluster center (CM) and the cluster slope (CS) are updated. procedure after claim 9 , characterized in that when calculating the updated values of the cluster center point (CM) and the cluster slope (CS), the received signals are weighted in such a way that received signals determined longer in the past are weighted less than received signals that are shorter in the past. Method according to one of the preceding claims, characterized in that the sensor (S) determines Doppler information on the received signal and that the Doppler information is used to check whether the received signal can be assigned to a cluster (C). procedure after claim 11 , characterized in that the Doppler information is compared with the cluster gradient (CS) during the test. System for assigning received signals from a sensor system of a vehicle (F) to an object, the system comprising at least one distance-measuring sensor (S) and a computer unit (R) for controlling the sensor system and providing received signals, the computer unit (R) is configured to carry out the following steps: - the sensor system carries out a plurality of chronologically consecutive transmission/reception cycles, the transmission/reception cycles each involving the transmission of a measurement signal by at least one sensor (S) and the reception of a reflected portion of this measurement signal as a reception signal by a include sensor (S); - Forming at least one cluster (C) of received signals by means of a regression method, the received signals being assigned to a cluster (C) based on their signal propagation time between the transmission of the measurement signal and the receipt of a reflected portion of this measurement signal, in such a way that those received signals Clusters (C) are added whose signal propagation times change linearly or essentially linearly over time, taking into account a predetermined variance; - Determining a tolerance range (T) for the at least one cluster (C), the tolerance range (T) defining a signal propagation time range in which a future received signal must lie in order to be able to be assigned to the cluster (C); - Receiving a new received signal and checking based on the tolerance range (T) whether the new received signal can be assigned to the cluster (C); and - calculating updated cluster information of the cluster (C) to which the received signal can be assigned based on the regression method taking into account information of the received signals which are assigned to the cluster (C).

Images