DE102017201837A1 - Method for detecting and filtering precipitation on a radar sensor in a vehicle. - Google Patents

Abstract

Method for detecting precipitation on a radar sensor in a vehicle comprising the steps of performing a current radar measurement (R2) with a plurality of clusters (R2C1-R2C4); determining a frequency (NrRainClusCurrCycle) of cluster (R2C1-R2C4) in the current radar measurement ( R2) with the following conditions: a relative velocity (Vel) which is within a predetermined speed range; and a distance (Dis) which is within a predetermined distance range; wherein the current radar measurement (R2) is detected as being affected by precipitation if the determined frequency (NrRainClusCurrCycle) exceeds a frequency threshold.

Description

The invention relates to a method for detecting precipitation on a radar sensor in a vehicle. Furthermore, the invention comprises a radar sensor which is set up to carry out the method.

State of the art

Radar systems are increasingly used in vehicles. For example, it is off DE 10 2010 006 214 A1 a radar system is known that can perform emergency braking.

In particular, emergency braking places high demands on the reliability of the radar system. The radar system must quickly detect changes in the behavior of the preceding vehicle in order to be able to initiate emergency braking in good time.

Precipitation on the radar sensor affects the transmission and reception characteristics of the radar sensor, which can lead to errors.

From CA 2 276 975 a rain detection method for radar sensors is known in which the potential objects must be within a specified distance. Objects below this distance are classified as precipitation and discarded.

Task and solution

The object of the present invention is to recognize when a radar sensor is influenced by precipitation.

The object is achieved by a method according to the independent claim, and a radar sensor according to the independent claim. Further advantageous embodiments of the invention are the subject of the dependent claims.

The method according to the invention for detecting precipitation on a radar sensor in a vehicle comprises the following steps, performing a current radar measurement with a plurality of clusters, determining a frequency of clusters in the current radar measurement with the following conditions, a relative speed that lies within a predetermined speed range , and a distance that is within a predetermined range of distances, wherein the current radar measurement is detected to be affected by precipitation when the determined frequency exceeds a frequency threshold.

The distance may be an extrapolated distance. The extrapolated distance describes a distance where the object can actually be detected based on its properties.

Advantageously, only clusters are considered that could potentially be affected by precipitation. Precipitation on the radar sensor can only be assumed when a given frequency of the clusters potentially affected by precipitation occurs.

Preferably, an RCS average of RCS values of the clusters of the current radar detected as affected by precipitation may be formed, and the current radar measurement may be recognized as being affected by precipitation only when the RCS average is less than -25db.

Advantageously, radar measurements with a higher RCS average of the individual clusters are recognized as not being affected by precipitation. Objects with a high reflection are not mistakenly recognized as raindrops. The reliability of the radar sensor is thus improved.

More preferably, a plurality of radar measurements each having a plurality of clusters may be performed, wherein the plurality of radar measurements occur earlier in time than the current radar measurement, an RCS global average value from the RCS averages of the RCS values of the clusters detected to be affected by precipitation the plurality of radar measurements and the RCS average value of the current radar measurement is formed, and the current radar measurement only then as by Precipitation influence is detected when an RCS deviation of the RCS average value of the current radar measurement from the RCS global average value is below an RCS deviation threshold.

In an advantageous manner, it can be checked whether the RCS average values of the current radar wave are in a similar range as the values of the preceding radar measurements.

In a further embodiment of the invention, a plurality of radar measurements each having a plurality of clusters may be performed, wherein the plurality of radar measurements occurs earlier than the current radar measurement, an average frequency from the frequencies of the plurality of radar detected clusters affected by precipitation and the frequency of the current radar measurement is formed, the current radar measurement being recognized as being affected by precipitation only when the average frequency is above an average threshold.

Advantageously, a radar measurement is detected as being affected by precipitation when an average number of clusters within the current and earlier radar measurements have been detected as being affected by precipitation. In a particularly advantageous manner, a history of the radar measurements is thereby formed and only then concluded on precipitation, when a sufficient number of radar measurements has been recognized as influenced by precipitation.

Preferably, the plurality of radar measurements (R1) may include at least 100 radar measurements, each lasting at least 0.04 seconds. Advantageously, sufficient clusters can be determined during this time span.

In a further embodiment of the invention, the relative speed can be in the range of 0.5 to 2.5 m / s or -2.5 to -0.5 m / s. Advantageously, an influence can be expected in this area.

In one embodiment of the invention, the distance may be less than 35 m. Advantageously, clusters in this distance range can be expected to be affected by precipitation.

More preferably, the radar detected as affected by precipitation can be filtered out.

Advantageously, the accuracy of the radar sensor is thus further improved, since influencing the results of the radar sensor by precipitation is avoided.

According to the invention, a radar sensor is set up to carry out a method as described above.

figure description

1 shows two tables with information about two radar measurements. The upper table contains the individual clusters R1Cx ( 1 - 5 ) of a previous radar measurement R1 and the lower table the individual clusters R2Cx ( 1 - 4 ) of a current radar measurement R2. The current radar measurement R2 took place later than the earlier radar measurement R1.

Of the individual clusters R1C1-R1C5, R2C1-R2C4, the relative velocity Vel, the distance Dis and the value for the radar cross-section RCS are respectively stored in the table.

In the Rain column, there is a star among the clusters that have been recognized as being affected by rain. If a cluster has a relative velocity (Vel) in the range of [-2.5, -0.5] or [0.5, 2.5], and a distance Dis of less than 35m, then this cluster is considered to be influenced by precipitation.

2 shows the processing steps of the clusters of the radar measurements R1, R2. In the first step 11 the values of the clusters are recorded from the respective radar measurement. In step 13 The values of the clusters in a table are as in 1 shown registered. In step 15 a check is made of the values of the clusters and, if necessary, a return 14 to step 13 if the values of the clusters are incorrect. In step 15 it is further checked whether the relative velocity Vel and the distance Dis of the individual clusters are in the relevant range. If so, these clusters become the next processing step 17 further processed. If no cluster meets the conditions of the step 15 , so a jump takes place 16 to the step 21 ,

In step 17 is determined from the RCS values of the clusters R1C3,5; R2C1-3 per radar measurement an RCS average (AvgRCSRain) is formed as follows. The number of clusters detected by rainfall per radar measurement is determined. In the example off 1 The values of two clusters of R1 and three clusters of R2 are within the ranges of step 15 , These clusters are in 1 marked with a star in the Rain column. The RCS values of these clusters are added up per radar measurement, giving the sum for R1 -59 and R2 -102. This sum is divided by the number of recognized clusters, so AvgRCSRain is R1 -29.5 and R2 is -34.

In step 19 it is checked whether there are any further clusters of further radar measurements. If this is the case, a return occurs 18 to the step 15 and these clusters also go through the steps 15 and 17 , An ongoing determination of whether precipitation affects the clusters is thus possible.

In step 21 it is checked whether at least one cluster of each radar measurement the conditions of the step 15 Fulfills. The further processing of the clusters follows after receipt of the values of the radar measurement.

The radar measurement R1 took place before the radar measurement R2. R2 represents a current radar measurement in this embodiment.

In step 25 We take an RCS global average value (GlobalAvgRCSRain) by summing up the AvgRCSRain of each radar measurement from step 17 educated. Since it is to be determined whether the current radar measurement is influenced by precipitation, the radar measurements are subdivided into a current radar measurement R2 (current cycle) and chronologically earlier radar measurements R1 (till current cycle). In addition, a weighting factor alpha is introduced which represents a history for the previous radar measurements. This results in the following formula for the RCS global average value: $$\text{GlobalAvgRCSRain}=\text{alpha}*\text{AvgRCSRainCurrCycle}+\left(1-\text{alpha}\right)*{\text{AvgRCSRain}}_{\text{till current cycle}}$$

Since the detection is continuous, the step stops 25 an update of the RCS global average value of the previous time radar measurement R1 with the current radar measurement R2, taking into account only the values of the clusters that were used in step 15 as influenced by precipitation.

For the example 1 the result is the GlobalAvgRCSRain after the first radar measurement R1 as follows: $$\text{<figure-callout id="1" label="GlobalAvgRCSRain\_R" filenames="DE102017201837A1\_0011.png" state="{{state}}">GlobalAvgRCSRain\_R</figure-callout>}1=\mathrm{12:01}*\left(-29.5\right)+\left(1-\mathrm{12:01}\right)*0=-0295$$

After the second radar measurement R2: $$\text{GlobalAvgRCSRain\_R2}=\mathrm{12:01}*\left(-34\right)+\left(1-\mathrm{12:01}\right)*\left(-0295\right)=-0632$$

Here Alpha = 0.01 was selected, taking into account a history of 100 radar measurements.

If in step 21 no precipitation affected cluster is present, then the step 23 the GlobalAvgRCSRain is set to 0. In this case, a jump takes place 24 to the step 27 and step 25 is omitted.

In step 27 an RCS deviation (RCSdev) of the RCS average value (AvgRCSRainCurrCycle) of the current radar measurement R2 is determined by the RCS global average value (GlobalAvgRCSRain). $$\text{RCSdev}\mathrm{=}\text{GlobalAvgRCSRain}\mathrm{-}\text{AvgRCSRainCurrCycle}$$

$$\text{R1: RCSdev}\mathrm{ = -}\text{0}\text{.295}\mathrm{-}\left(-34\right)=33.\text{705}$$

For the example 1 : $$\text{R1: RCSdev}\mathrm{=}\text{0}\mathrm{-}\left(29.5\right)=29.5$$

$$\text{R1: RCSdev}\mathrm{=\; -}\text{0}\text{.295}\mathrm{-}\left(-34\right)=\mathrm{33rd}\text{705}$$

The GlobalAvgRCSRain changes with each radar measurement in step 25 will be processed.

In step 29 will check if the in step 25 determined RCS average value (AvgRCSRainCurrCycle) of the current radar measurement R2 is less than 25dB. If this is the case, then a reference is made to the step 33 ,

In step 33 An average frequency (AvgNoRainClus) is formed by low-pass filtering and summing the frequencies (NrRainClus) of clusters (R1C3, R1C5, R2C1-3) detected as affected by precipitation per radar measurement.

The weighting factor here too is alpha = 100, which takes into account a history of 100 radar measurements. $$\text{AvgNoRainClus}\mathrm{=}\text{alpha}*\text{NrRainClusCurrCycle}\mathrm{+}\left(1-\right)\text{alpha}*{\text{AvgNoRainClus}}_{\text{till current cycle}}$$

For the example 1 surrendered:

$$\text{R2: AvgNoRainClus}\mathrm{ = }\text{0}\text{.01}*\left(3\right)+\left(1-0.01\right)*002=0.049$$

NrRainClus_R1 = 2 and NrRainClus_R2 = 3, since in R1 two clusters and in R2 three clusters have been identified as being affected by precipitation. $$\text{R1: AvgNoRainClus}\mathrm{=}\text{0}\text{.01}*\left(2\right)+\left(1-\mathrm{12:01}\right)*0=\mathrm{12:02}$$

$$\text{R2: AvgNoRainClus}\mathrm{=}\text{0}\text{.01}*\left(3\right)+\left(1-\mathrm{12:01}\right)*002=0049$$

AvgNoRainClus changes with each radar wave in step 25 will be processed. After the current radar wave in step 29 into the average frequency (AvgNoRainClus), the average frequency is compared with a threshold value.

Should the exam in step 29 result in the RCS average being greater than 25 dB, the step refers 29 to step 31 , In step 31 the frequency (NrRainClusCurrCycle) of the current radar measurement R1 is set to 0. The then in step 33 the resulting calculation is shortened as follows: $$\text{AvgNoRainClus}\mathrm{=}\text{alpha}*0+\left(1-\text{alpha}\right)*{\text{AvgNoRainClus}}_{\text{till current cycle}}$$

In step 35 is then based on the in step 27 determined RCS deviation (RCSdev) and / or in step 33 average frequency determined (AvgNoRainClus) decided whether the radar measurement should be considered influenced by precipitation.

• R1: AvgNoRainClus = 0.02 => keine Beeinflussung
• R2: AvgNoRainClus = 0.049 => keine Beeinflussung

In the example off 1 Both radar measurements are detected as not affected by precipitation, since AvgNoRainClus is less than 4 and RCSdev is less than 10.

• R1: AvgNoRainClus = 0.02 => no influence
• R2: AvgNoRainClus = 0.049 => no influence

The thresholds for AvgNoRainClus and RCSdev were chosen for 100 radar measurements, with each radar measurement 12:04 s takes.

If a radar measurement has been found to be affected by precipitation, the radar measurement is filtered so that no erroneous conclusions can be drawn from this affected radar measurement.

QUOTES INCLUDE IN THE DESCRIPTION

This list of the documents listed by the applicant has been generated automatically and is included solely for the better information of the reader. The list is not part of the German patent or utility model application. The DPMA assumes no liability for any errors or omissions.

Cited patent literature

• DE 102010006214 A1 [0002]

Claims (9)

A method of detecting precipitation on a radar sensor in a vehicle comprising the steps of: performing a current radar measurement (R2) with a plurality of clusters (R2C1-R2C4); Determine a frequency (NrRainClusCurrCycle) of cluster (R2C1-R2C4) in the current radar measurement (R2) with the following conditions: - A relative speed (Vel), which is within a predetermined speed range; and a distance (Dis) lying within a predetermined distance range; wherein the current radar measurement (R2) is detected as being affected by precipitation if the determined frequency (NrRainClusCurrCycle) exceeds a frequency threshold. Method according to Claim 1 in which an RCS average value (AvgRCSRainCurrCycle) is formed from RCS values of the cluster (R2C1-R2C3) of the current radar measurement (R2), which is affected by precipitation, and the current radar measurement (R2) is only recognized as influenced by precipitation if the RCS average (AvgRCSRainCurrCycle) is less than 25db. Method according to one of the preceding claims, wherein a plurality of radar measurements (R1) are carried out, each with a plurality of clusters (R1C1-R1C5), wherein the plurality of radar measurements (R1) takes place earlier than the current radar measurement (R2), an RCS Global average value (GlobalAvgRCSRain) from the RCS (AvgRCSRain _{till current cycle} ) averages of the RCSs of the precipitation-affected clusters (R1C3, R1C5) from the plurality of radar measurements (R1) and the RCS average value (AvgRCSRainCurrCycle ) the current radar measurement (R2) is formed, and the current radar measurement (R2) is only recognized as being influenced by precipitation if an RCS deviation (RCSdev) of the RCS average value (AvgRCSRainCurrCycle) of the current radar measurement (R2) from the RCS Global average value (AvgRCSRain) is below an RCS deviation threshold. Method according to one of the preceding claims, wherein a plurality of radar measurements (R1) are carried out, each with a plurality of clusters (R1C1-R1C5), wherein the plurality of radar measurements (R1) takes place earlier than the current radar measurement (R2), an average frequency (AvgNoRainClus) is formed from the frequencies (NrRainClus) of the detected by precipitation influenced cluster from the plurality of radar measurement (R1) and the frequency (NrRainClusCurrCycle) of the current radar measurement (R2), the current radar measurement (R2) only then by Rainfall is detected when the average frequency (AvgNoRainClus) is above an average threshold. Method according to Claim 3 or 4 wherein the plurality of radar measurements (R1) comprise at least 100 radar measurements, each lasting at least 0.04 seconds. Method according to one of the preceding claims, wherein the relative speed is in the range of 0.5 to 2.5 m / s or -2.5 to -0.5 m / s. Method according to one of the preceding claims, wherein the distance (Dis) is less than 35 m. Method according to one of the preceding claims, wherein the detected as influenced by precipitation radar measurement is filtered out. Radar sensor in a vehicle, which is configured to carry out a method according to one of the preceding claims.

Images