KR20210044423A - Rainfall intensity estimation device and method using vehicle radar - Google Patents

Abstract

An apparatus and a method for estimating rainfall intensity using a vehicle radar are provided. In order to minimize a rainfall intensity estimation error after collecting and processing a radar signal and estimating the rainfall intensity, a preprocessing process of filtering an object is included to prevent overestimation of the rainfall intensity from the object. The present invention includes a pre-processing module.

Description

Rainfall intensity estimation device and method using vehicle radar

The present invention relates to an apparatus and method for estimating rainfall intensity using a vehicle radar, and more particularly, by observing rainfall using a w-band radar, which is a vehicle radar, and filtering an object rather than detecting an object from radar data, It relates to an apparatus and method for estimating rainfall intensity.

In general, a meteorological radar is a device that calculates the size of a radio signal reflected or scattered by a meteorological target by emitting an electromagnetic wave. It can monitor a large area (effective observation radius: 240 km) very quickly (10 minutes). Because of this, it is one of the most efficient remote sensing equipment that calculates precipitation over a large area.

Such a weather radar does not directly calculate precipitation, but calculates precipitation quantitatively by the size of a signal reflected or scattered from targets existing in the atmosphere by emitting electromagnetic waves.

Apart from this, recent advances in various sensing technologies and artificial intelligence technologies have made remarkable achievements for the commercialization of autonomous vehicles. Most of the automobiles produced these days are equipped with ADAS (Advanced Driver Assistance System), which has a high level of reliability, to improve the driver's convenience and safety.

However, studies on the stability of autonomous vehicles are still undergoing many trials and errors. In particular, in severe weather conditions such as snow, rain, and fog, there are many problems due to the limitations of observation of sensor devices such as radar, lidar, and camera.

In particular, the ultimate obstacle for fully driverless vehicle technology may be weather phenomena such as rain, snow, and fog rather than problems with algorithms and artificial intelligence.

In order to solve the above-described problems, the present invention aims to achieve the present invention by simultaneously observing objects and rainfall using a W-band vehicle radar, and classifying them to maintain the function of ADAS, while observing the current rainfall around the vehicle. The purpose of this is to improve the driving stability of the autonomous vehicle by estimating the intensity, and to secure the reliability of the data by enabling more accurate estimation of the rainfall intensity.

The problem of the present invention is not limited to those mentioned above, and other problems that are not mentioned will be clearly understood by those skilled in the art from the following description.

As a means for solving the above technical problem, according to an embodiment of the present invention, the rainfall intensity estimation apparatus using a vehicle radar performs data format conversion and quality control on the collected radar data, and determines the radar range and Doppler velocity profile. A radar data processing module that calculates; A preprocessing module for filtering objects such as people, cars, and street trees to minimize rainfall intensity estimation errors; And a rainfall intensity estimation module for estimating the rainfall intensity by using the range profile in which the object is filtered by the preprocessing module.

Here, the radar data processing module is characterized in that clutter filtering using Doppler speed is performed together.

In addition, the preprocessing module is characterized in that the object is filtered based on temporal statistics for each range bin data observed from the radar.

The preprocessing module is characterized in that it filters the object using the following equation (1).

<Equation 1>

는

에 대해 오브젝트 필터링이 수행된 후 치환된 range bin을 의미하며,

는 시간 축 상으로 관측된 N개의 i번째 range bin에 대한 집합

을 의미한다.

는

원소들에 대한 평균값이며,

는

원소들에 대한 표준편차를 의미하며,

는 오브젝트를 필터링하기 위한

의 임계치를 의미한다. only,

Is

It means the range bin replaced after object filtering is performed for,

Is the set of the n i- th range bins observed on the time axis

Means.

Is

Is the average value of the elements,

Is

Means the standard deviation of the elements,

For filtering objects

Means the threshold of.

The rainfall intensity estimation module is characterized in that the rainfall intensity R is calculated from the range profile from which the object has been removed using Equation 2 below.

<Equation 2>

은 range bin의 거리 인덱스이며,

는 각 거리인덱스에 따른 return power를 의미한다. 또한 Cov[I, b]는 I와 b에 대한 covariance를, Var[I]는 I에 대한 variance를 의미한다.only,

Is the distance index of the range bin,

Means return power according to each distance index. Also, Cov[I, b] means covariance for I and b , and Var[I] means variance for I.

According to another embodiment of the present invention, a method for estimating rainfall intensity using a vehicle radar is a radar data processing in which a radar data processing module performs data format conversion and quality control on the collected radar data, and calculates a radar range and a Doppler velocity profile. step; A pre-processing step in which the pre-processing module filters objects such as people, cars, and street trees to minimize rainfall intensity estimation errors; And a rainfall intensity estimation step in which the rainfall intensity estimation module estimates the rainfall intensity by using the range profile in which the object is filtered by the preprocessing module.

Here, the radar data processing step further comprises a step of clutter filtering using a Doppler speed.

In addition, the pre-processing step is characterized in that the object is filtered based on temporal statistics for each range bin data observed from the radar.

The pre-processing step is characterized in that the object is filtered using Equation 1 below.

<Equation 1>

는

에 대해 오브젝트 필터링이 수행된 후 치환된 range bin을 의미하며,

는 시간 축 상으로 관측된 N개의 i번째 range bin에 대한 집합

을 의미한다.

는

원소들에 대한 평균값이며,

는

원소들에 대한 표준편차를 의미하며,

는 오브젝트를 필터링하기 위한

의 임계치를 의미한다. only,

Is

It means the range bin replaced after object filtering is performed for,

Is the set of the n i- th range bins observed on the time axis

Means.

Is

Is the average value of the elements,

Is

Means the standard deviation of the elements,

For filtering objects

Means the threshold of.

The rainfall intensity estimation step is characterized in that the rainfall intensity R is obtained from the range profile from which the object has been removed using Equation 2 below.

<Equation 2>

은 range bin의 거리 인덱스이며,

는 각 거리인덱스에 따른 return power를 의미한다. 또한 Cov[I, b]는 I와 b에 대한 covariance를, Var[I]는 I에 대한 variance를 의미한다.only,

Is the distance index of the range bin,

Means return power according to each distance index. Also, Cov[I, b] means covariance for I and b , and Var[I] means variance for I.

According to the present invention,

Using W-band vehicle radar, objects and rainfall are simultaneously observed and classified to maintain the function of ADAS, while at the same time observing the current rainfall around the vehicle and estimating the rainfall intensity to improve the driving stability of autonomous vehicles. The reliability of data can be secured by enabling more accurate estimation of rainfall intensity.

The effects of the present invention are not limited to those mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the following description.

1 is a block diagram showing a rainfall intensity estimation apparatus according to an embodiment of the present invention;
2 is a flow chart for explaining a rainfall intensity estimation method according to an embodiment of the present invention;
3 is a flow chart of a system for estimating rainfall intensity in an ADAS system,
4 is a graph of an experiment of analyzing the influence of a range profile on rainfall of a W-band FMCW vehicle radar,
5 is an exemplary view in which a research FMCW radar is mounted on a vehicle,
6 is an exemplary diagram illustrating an object filtering result experiment when driving a vehicle in a rain-free environment;
7 is an experimental graph of a rainfall intensity error due to an object in a rain-free environment and a correction result by object filtering;
8 is an exemplary diagram illustrating a conceptual rainfall intensity influence analysis result by object filtering in a strong rainfall environment;
9 is an exemplary diagram illustrating rainfall intensity estimation at a point in time that changes from very weak rainfall to strong rainfall.

The above objects, other objects, features, and advantages of the present invention will be easily understood through the following preferred embodiments related to the accompanying drawings. However, the present invention is not limited to the embodiments described herein and may be embodied in other forms. Rather, the embodiments introduced herein are provided so that the disclosed contents may be thorough and complete, and the spirit of the present invention may be sufficiently conveyed to those skilled in the art.

In the present specification, when terms such as first and second are used to describe constituent elements, these constituent elements should not be limited by these terms. These terms are only used to distinguish one element from another element. The embodiments described and illustrated herein also include complementary embodiments thereof.

In addition, if an element, component, device, or system is stated to include a program or a component made of software, the element, component, device, or system is executed by the program or software, even if there is no explicit mention. Or it should be understood to include hardware (eg, memory, CPU, etc.) or other programs or software (eg, a driver required to drive an operating system or hardware) required to operate.

In addition, terms used in the present specification are for describing exemplary embodiments and are not intended to limit the present invention. In this specification, the singular form also includes the plural form unless specifically stated in the phrase. As used in the specification, "comprises" and/or "comprising" does not exclude the presence or addition of one or more other components.

In describing the specific embodiments below, a number of specific contents have been prepared to explain the invention in more detail and to aid understanding. However, readers who have knowledge in this field enough to understand the present invention can recognize that it can be used without these various specific contents.

In some cases, it should be mentioned in advance that parts that are commonly known in describing the invention and are not largely related to the invention are not described in order to avoid confusion without any reason in describing the invention.

Hereinafter, with reference to the accompanying drawings, specific technical content to be implemented in the present invention will be described in detail.

The present invention makes it possible to observe precipitation and estimate rainfall intensity through w-band radar by improving this in that ADAS malfunctions may occur due to observation limitations of sensor equipment such as radar, lidar, and camera in severe weather conditions. .

1 is a block diagram illustrating a rainfall intensity estimation apparatus according to an embodiment of the present invention, and FIG. 2 is a flowchart illustrating a rainfall intensity estimation method according to an embodiment of the present invention.

The present invention is a radar data processing step in which the radar data processing module 100 performs data format conversion and quality control on the collected radar data, and calculates a radar range and a Doppler velocity profile in estimating rainfall intensity using a vehicle radar ( S100), the pre-processing module 200 uses the range profile in which the object is filtered by the pre-processing module 200 through a pre-processing step (S200) in which objects such as people, automobiles, and street trees are filtered to minimize the rainfall intensity estimation error. Thus, the rainfall intensity estimation module 300 estimates the rainfall intensity through the rainfall intensity estimation step (S300) of estimating the rainfall intensity.

3 is a flowchart of a system for estimating rainfall intensity in an ADAS system.

In the radar data processing step (S100), clutter filtering using Doppler speed is performed together, specifically, data is acquired through radar (Radar Data Acquisition), and then the radar range and Doppler speed profile are calculated ( Range & Velocity FFTs).

In the preprocessing step S200, objects are filtered based on temporal statistics for each range bin data observed from the radar. In order to minimize the rainfall intensity estimation error, an important pre-processing process of filtering objects as opposed to object (object) detection is performed.

A range bin is a data cell that is a basic data cell when processing radar signals received from an arbitrary azimuth by unit distance, and is also called a range gate. There is one range bin for each bin space with the radar as the starting point.

In general, the intensity of the signal reflected from the detected object is much greater than that of the radar signal reflected from the rainfall in the ADAS system, so if object filtering is not performed, the detected object signal tends to overestimate the original rainfall intensity. This can happen. On the other hand, if the signal reflected from the object disappears in the process of object filtering, the rainfall intensity may be underestimated on the contrary, so the filtering of signals for surrounding objects obtained from a driving vehicle is a very important process in the proposed technique. .

Since the vehicle radar operates in a frequency modulated continuous wave (FMCW) method, it has the advantage of obtaining real-time location information of a fast moving object in a short distance with a high distance resolution. This means that the vehicle radar reacts sensitively to changes in the position or coordinates of objects occurring within a short time, and it may mean that the sensitivity to changes in rainfall with relatively little temporal variation is low.

In general, even weather phenomena such as sudden showers and squalls take at least a few seconds for the rainfall to intensify.

Accordingly, the present invention includes a method of filtering an object based on temporal statistics for each range bin data observed from a vehicle radar, and filtering the object using [Equation 1] below.

[Equation 1]

는

에 대해 오브젝트 필터링이 수행된 후 치환된 range bin을 의미하며,

는 시간 축 상으로 관측된 N개의 i번째 range bin에 대한 집합

을 의미한다.

는

원소들에 대한 평균값이며,

는

원소들에 대한 표준편차를 의미하며,

는 오브젝트를 필터링하기 위한

의 임계치를 의미한다. only,

Is

It means the range bin replaced after object filtering is performed for,

Is the set of the n i-th range bins observed on the time axis

Means.

Is

Is the average value of the elements,

Is

Means the standard deviation of the elements,

For filtering objects

Means the threshold of.

에 의해 결정되며, 이것은 N 번의 관측 동안 강우에 의한 레이더 반사강도의 변화보다 오브젝트에 의한 레이더 반사강도가 더 큰 물리적 현상에 기인한다. Therefore, object filtering is the change and threshold of the standard deviation for the last N range bins and bis.

This is due to a physical phenomenon in which the radar reflection intensity by the object is greater than the change in radar reflection intensity due to rainfall during N observations.

Meanwhile, the proposed algorithm is designed to be applied after N observations are performed after the radar is booted for stable filtering criteria.

5 is an exemplary view in which a research FMCW radar is mounted on a vehicle for a field test. For comparative verification, a webcam was additionally mounted on the top of the radar.

In addition, the performance of the FMCW radar used for the field test on the road is shown in [Table 1].

Configuratoion items value Carrier Frequency 77 GHz Azimuth Reslolution 15° Range Resolution 0.04m Unambiguous Range 31.0m Frame Duration 100 msec.

6 is an exemplary diagram illustrating an experiment result of object filtering when a vehicle is driven in a rain-free environment. 6-(a) shows the comparison of the result of the photographing with the camera mounted on the vehicle at the moment of observation and the actual coordinate values of the detected objects.

6-(b) shows a range profile and an object filtering result when an object is detected while driving a vehicle.

From Fig. 6-(b), strong peaks are generated on the range profile due to objects detected while driving the vehicle, and it can be confirmed that strong peaks are well suppressed from the object filtering result of the present invention.

In order to analyze the result of the rainfall intensity estimation of the present invention, first, a test for the rainfall intensity when the vehicle is driven in a rain-free environment was performed. The black dotted line shows the rainfall intensity R in no rainfall when there is no object, the gray line shows the rainfall intensity without object filtering, and the red line shows the rainfall intensity after performing the proposed object filtering on the time axis. Each is shown.

7 is an experimental graph of a rainfall intensity error due to an object in a no-rainfall environment and a correction result by object filtering, FIG. 8 is an exemplary view of a conceptual rainfall intensity influence analysis result by object filtering in a strong rainfall environment, and FIG. 9 is a very weak rainfall. This is an exemplary diagram of estimation of rainfall intensity at the point of change to strong rainfall in.

This may indicate that more accurate rainfall intensity estimation is possible when the pre-processing step according to an embodiment of the present invention is performed. Referring to FIG. 7, when the object is not filtered from, an error in estimation of the rainfall intensity occurs due to the object, and by applying the object filtering technique of the present invention, a rainfall intensity result close to the rainfall intensity in the case of no rainfall is derived to estimate the rainfall intensity. It can be seen that error correction is possible.

In addition, it was verified that the rainfall intensity estimation error can be corrected because it is shown even after the pre-processing step according to the embodiment of the present invention is performed.

In addition, FIG. 8 is a diagram illustrating a conceptual rainfall intensity influence analysis result by object filtering in a strong rainfall environment.

It can be seen that it does not have a significant effect on the rainfall intensity estimation (FIG. 8). From the results of FIG. 8, it was confirmed that the object filtering algorithm of the present invention does not significantly affect the rainfall intensity estimation as the object is removed. From this, it can be seen that the important advantage of enabling accurate rainfall estimation regardless of objects even when driving a vehicle has been verified.

In addition, finally, the observation result (FIG. 9) is compared at the point of sudden change from a very weak rainfall to a strong rainfall of 20 mm/h or more in an object-free environment, and is shown in FIG.

9 is an exemplary diagram illustrating rainfall intensity estimation at a point in time that changes from very weak rainfall to strong rainfall.

In the graph of FIG. 9, when the rainfall intensity estimation algorithm of the present invention was used, a phenomenon of changing from very weak rainfall to strong rainfall for about 20 sec could be clearly observed. From this result, it can be confirmed that the rainfall detection algorithm using the 77GHz vehicle radar can sensitively estimate the change in rainfall intensity.

In the rainfall intensity estimation step S300, the rainfall intensity is estimated from the range profile from which the object has been removed.

4 shows no rainfall (Fig. 4-(a)), weak rainfall (Fig. 4-(b)), strong rainfall (Fig. 4-(c)) to analyze the influence of the range profile on the rainfall of the W-band FMCW vehicle radar. )) is a graph showing the results of the experiment.

The data in FIG. 4 are results of setting the observation distance to 30 m. From FIG. 4, each black thin line represents an average of a range profile on a sunny day, and a blue colored oscillating solid line is a range profile measured in each environment. In order to facilitate the analysis, the measured range profile was shown as a bold line by means of mean average filtering (MAF).

As a result of the experiment, it can be seen that the return power of the range bins observed at close range varies depending on the rainfall intensity, and the sensitivity (change) of the return power due to rainfall decreases as the distance increases due to attenuation by rainfall that exists over the entire observation distance. Could know.

로부터 k 개의 연속한 range bin에 대해 [수학식 2] 와 같이 선형 추세선의 기울기로써 강우강도 R을 구할 수 있다.Therefore, in the present invention, the conceptual rainfall intensity is defined as the slope for the displacement of the return power over the observation distance using such characteristics. A-th range bin that is most affected by actual rainfall

For k consecutive range bins from, the rainfall intensity R can be obtained as the slope of the linear trend line as shown in [Equation 2].

[Equation 2]

은 range bin의 거리 인덱스이며,

는 각 거리인덱스에 따른 return power를 의미한다. 또한 Cov[I, b]는 I와 b에 대한 covariance를, Var[I]는 I에 대한 variance를 의미한다.only,

Is the distance index of the range bin,

Means return power according to each distance index. Also, Cov[I, b] means covariance for I and b , and Var[I] means variance for I.

In the rainfall intensity estimation step (S300), an error between the rainfall intensity derived using the relational expression and the actually recorded rainfall intensity is calculated to analyze the performance of the rainfall intensity relational expression, and accordingly, the rainfall intensity estimation error can be corrected. have.

In the above, even if all the constituent elements constituting the embodiments of the present invention have been described as being combined into one or operating in combination, the present invention is not necessarily limited to these embodiments. That is, within the scope of the object of the present invention, all the constituent elements may be selectively combined and operated in one or more. In addition, although all of the components may be implemented as one independent hardware, a program module that performs some or all functions combined in one or more hardware by selectively combining some or all of the components. It may be implemented as a computer program having Codes and code segments constituting the computer program may be easily inferred by those skilled in the art. Such a computer program is stored in a computer-readable storage medium, and is read and executed by a computer, thereby implementing an embodiment of the present invention.

100: radar data processing module
200: pre-processing module
300: rainfall intensity estimation module

Claims (10)

In the rainfall intensity estimation apparatus using a vehicle radar,
A radar data processing module that performs data format conversion and quality control on the collected radar data, and calculates a radar range and a Doppler velocity profile;
A preprocessing module for filtering objects such as people, cars, and street trees to minimize rainfall intensity estimation errors;
A rainfall intensity estimation module for estimating a rainfall intensity using a range profile in which an object is filtered by the preprocessing module;
Rain intensity estimation apparatus comprising a. The method of claim 1,
The radar data processing module,
A rainfall intensity estimation apparatus, characterized in that clutter filtering using a Doppler velocity is performed together. The method of claim 1,
The preprocessing module,
An apparatus for estimating rainfall intensity, comprising filtering objects based on temporal statistics for each range bin data observed from a radar. The method of claim 1,
The preprocessing module,
An apparatus for estimating rainfall intensity, characterized in that filtering an object using Equation 1 below.
<Equation 1>

only,

Is

It means the range bin replaced after object filtering is performed for,

Is the set of the n i- th range bins observed on the time axis

Means.

Is

Is the average value of the elements,

Is

Means the standard deviation of the elements,

For filtering objects

Means the threshold of. The method of claim 1,
The rainfall intensity estimation module,
The rainfall intensity estimation apparatus, characterized in that to obtain the rainfall intensity R using the following equation (2) from the range profile from which the object has been removed.
<Equation 2>

only,

Is the distance index of the range bin,

Means return power according to each distance index. Also, Cov[I, b] means covariance for I and b , and Var[I] means variance for I. In the rainfall intensity estimation method using a vehicle radar,
A radar data processing step in which the radar data processing module performs data format conversion and quality control on the collected radar data, and calculates a radar range and a Doppler velocity profile;
A pre-processing step in which the pre-processing module filters objects such as people, cars, and street trees to minimize rainfall intensity estimation errors;
A rainfall intensity estimation step in which the rainfall intensity estimation module estimates the rainfall intensity by using the range profile in which the object is filtered in the preprocessing module;
Precipitation intensity estimation method comprising a. The method of claim 6,
The radar data processing step,
Clutter filtering using a Doppler velocity;
The rainfall intensity estimation method further comprising a. The method of claim 6,
The pretreatment step,
A method for estimating rainfall intensity, comprising filtering objects based on temporal statistics for each range bin data observed from a radar. The method of claim 6,
The pretreatment step,
A method of estimating rainfall intensity, characterized in that the object is filtered using Equation 1 below.
<Equation 1>

only,

Is

It means the range bin replaced after object filtering is performed for,

Is the set of the n i- th range bins observed on the time axis

Means.

Is

Is the average value of the elements,

Is

It means the standard deviation of the elements,

For filtering objects

Means the threshold of. The method of claim 6,
The rainfall intensity estimation step,
The rainfall intensity estimation apparatus, characterized in that to obtain the rainfall intensity R using the following equation (2) from the range profile from which the object has been removed.
<Equation 2>

only,

Is the distance index of the range bin,

Means return power according to each distance index. Also, Cov[I, b] means covariance for I and b , and Var[I] means variance for I.

Images